diff --git a/.agents/notes/implemented/architecture/2026-06-18-shared-persistence-write-coordinator.i18n.yaml b/.agents/notes/archived/architecture/2026-06-18-shared-persistence-write-coordinator.i18n.yaml similarity index 66% rename from .agents/notes/implemented/architecture/2026-06-18-shared-persistence-write-coordinator.i18n.yaml rename to .agents/notes/archived/architecture/2026-06-18-shared-persistence-write-coordinator.i18n.yaml index a6f873d0f8..1d5dbfcc3e 100644 --- a/.agents/notes/implemented/architecture/2026-06-18-shared-persistence-write-coordinator.i18n.yaml +++ b/.agents/notes/archived/architecture/2026-06-18-shared-persistence-write-coordinator.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-06-18-shared-persistence-write-coordinator.md -2026-06-18-shared-persistence-write-coordinator.md: a61ceb9b2197a6dd8ed86c1c971373a2706607aa -2026-06-18-shared-persistence-write-coordinator.zh.md: 777d5f5972ac1096c2e3434f9e0ac5aec27e8c26 +2026-06-18-shared-persistence-write-coordinator.md: 5c324f2c0c2b951f664bcf92725a36c4975fd3a1 +2026-06-18-shared-persistence-write-coordinator.zh.md: 51ef76189cb9276214bf05bbd1dbd8e3755daa35 diff --git a/.agents/notes/implemented/architecture/2026-06-18-shared-persistence-write-coordinator.md b/.agents/notes/archived/architecture/2026-06-18-shared-persistence-write-coordinator.md similarity index 99% rename from .agents/notes/implemented/architecture/2026-06-18-shared-persistence-write-coordinator.md rename to .agents/notes/archived/architecture/2026-06-18-shared-persistence-write-coordinator.md index a61ceb9b21..5c324f2c0c 100644 --- a/.agents/notes/implemented/architecture/2026-06-18-shared-persistence-write-coordinator.md +++ b/.agents/notes/archived/architecture/2026-06-18-shared-persistence-write-coordinator.md @@ -1,6 +1,7 @@ # Agent Note: Shared persistence write coordinator Status: implemented +Archived: 2026-08-31 English | [中文](2026-06-18-shared-persistence-write-coordinator.zh.md) diff --git a/.agents/notes/implemented/architecture/2026-06-18-shared-persistence-write-coordinator.zh.md b/.agents/notes/archived/architecture/2026-06-18-shared-persistence-write-coordinator.zh.md similarity index 99% rename from .agents/notes/implemented/architecture/2026-06-18-shared-persistence-write-coordinator.zh.md rename to .agents/notes/archived/architecture/2026-06-18-shared-persistence-write-coordinator.zh.md index 777d5f5972..51ef76189c 100644 --- a/.agents/notes/implemented/architecture/2026-06-18-shared-persistence-write-coordinator.zh.md +++ b/.agents/notes/archived/architecture/2026-06-18-shared-persistence-write-coordinator.zh.md @@ -1,6 +1,7 @@ # Agent Note: 共享持久化写入协调器 Status: implemented +Archived: 2026-08-31 [English](2026-06-18-shared-persistence-write-coordinator.md) | 中文 diff --git a/.agents/notes/archived/manifest.json b/.agents/notes/archived/manifest.json index 18173d7d3d..fa3abbea4f 100644 --- a/.agents/notes/archived/manifest.json +++ b/.agents/notes/archived/manifest.json @@ -10,6 +10,9 @@ "architecture/2026-06-15-turn-enclosure-invariant.i18n.yaml": "sha256:7eb471a53b7bef104c57e9343b80d672763f062ecb086b01b318b65b488d3c02", "architecture/2026-06-15-turn-enclosure-invariant.md": "sha256:afefa3a268c84f26cf5461e08933245352a9e63cff688d3c398c8064a4ac6e85", "architecture/2026-06-15-turn-enclosure-invariant.zh.md": "sha256:c54fdac980abc922cdc252a8fef59e4bdd7567316c7fbb6f7dbc035e470d95fa", + "architecture/2026-06-18-shared-persistence-write-coordinator.i18n.yaml": "sha256:3c5c22e9e6a63598ba648cad46d783af322cf3afd6021426a2d738f4b026bf65", + "architecture/2026-06-18-shared-persistence-write-coordinator.md": "sha256:d5242c770101086b6f0a0c40eab500d405ef4a98cae07e28d9ec21e89d94f90e", + "architecture/2026-06-18-shared-persistence-write-coordinator.zh.md": "sha256:3dce52e302600a0eea29b4821a2718b6bbc1ebe4c6c2ae372cd0cbe66cb05519", "architecture/2026-06-20-extract-example-app-packages.i18n.yaml": "sha256:d99b612cc1051c86d883d74737c72e921735e7a28e0b5e6351d3870c664bdcc4", "architecture/2026-06-20-extract-example-app-packages.md": "sha256:9c7aca3a1e9a1ccc3729961663bc649b90076e671cae23e3db8203305983ccce", "architecture/2026-06-20-extract-example-app-packages.zh.md": "sha256:19bd50232d9f25d35aa3f9dc72d9af0df457dd0eaca8b982d5aa625e5b95bcff", diff --git a/.agents/notes/implemented/architecture/2026-06-11-dev-invariants-over-deep-readonly.i18n.yaml b/.agents/notes/implemented/architecture/2026-06-11-dev-invariants-over-deep-readonly.i18n.yaml index 44272f7ca1..6c6ebd4788 100644 --- a/.agents/notes/implemented/architecture/2026-06-11-dev-invariants-over-deep-readonly.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-06-11-dev-invariants-over-deep-readonly.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-06-11-dev-invariants-over-deep-readonly.md -2026-06-11-dev-invariants-over-deep-readonly.md: 66980f1ee09c6112f72786d6c3a147aadbc57f6c -2026-06-11-dev-invariants-over-deep-readonly.zh.md: 576e53e0b27e65f6fa071ff649509223a7bc30ff +2026-06-11-dev-invariants-over-deep-readonly.md: 7e5f55e8910797bd46674050ea5eb8abbca4aef7 +2026-06-11-dev-invariants-over-deep-readonly.zh.md: c1413e9c89284312af41b6c115b7e50a99d1b5e3 diff --git a/.agents/notes/implemented/architecture/2026-06-11-dev-invariants-over-deep-readonly.md b/.agents/notes/implemented/architecture/2026-06-11-dev-invariants-over-deep-readonly.md index 66980f1ee0..7e5f55e891 100644 --- a/.agents/notes/implemented/architecture/2026-06-11-dev-invariants-over-deep-readonly.md +++ b/.agents/notes/implemented/architecture/2026-06-11-dev-invariants-over-deep-readonly.md @@ -22,7 +22,7 @@ Responsibility is split between an always-on storage boundary and optional devel `Session` accepts an event only after one recursive pass has materialized a lossless JSON snapshot. That pass rejects unsupported values and produces the exact detached record that enters the log, so validation and storage cannot observe different values from a stateful getter or retain caller-owned nested references. -The accepted event and all of its descendants are deep-frozen before publication. `append()` returns that owned frozen event, `session/event` observers receive the same record, and `session.events` returns a frozen array snapshot. A previously returned array does not grow after a later append. Seed records pass through the same validation, snapshot, and freeze boundary before construction succeeds. +The accepted event and all of its descendants are deep-frozen before publication. `append()` returns that owned frozen event, and `session/event` observers and `eventAt(seq)` receive the same record. `snapshotEvents(fromSeq?, toSeqExclusive?)` returns a frozen array snapshot; a previously returned array does not grow after a later append. `seq` and `eventAt()` avoid array materialization when a caller needs only the current length or one event. Seed records pass through the same validation, snapshot, and freeze boundary before construction succeeds. This guarantee belongs in `Session`, not in an optional listener, because every composition relies on trustworthy history. A production deployment, a focused test, or a custom embedding receives the same storage semantics whether or not development support plugins are registered. @@ -32,7 +32,7 @@ This guarantee belongs in `Session`, not in an optional listener, because every ### Package-owned invariant companions check relationships -`dsh-invariants` registers the configurable `ctx.invariants` service and contains no product checks. Every package publishes a `./invariant` ownership companion; `dsh-session`, `dsh-agent`, `dsh-scope`, and `dsh-agent-loop` currently add the rules that require trace state or observation of another seam: monotonic sequence numbers, turn and step nesting, tool-call/result pairing, legal agent-status transitions, subject-correct scoped dispatch, and equality between a loop-built request and the request reconstructed from its session-log prefix. Global enablement and package-name regex filters belong to the service ([package-owned invariant service](2026-07-19-package-owned-invariant-service.md)). +`dsh-invariants` registers the configurable `ctx.invariants` service and contains no product checks. A package publishes a `./invariant` ownership companion only for an independently observable runtime relationship; packages without one omit the companion and record the reason in their README. `dsh-session`, `dsh-agent`, `dsh-scope`, and `dsh-agent-loop` provide the initial rules that require trace state or observation of another seam: monotonic sequence numbers, turn and step nesting, tool-call/result pairing, legal agent-status transitions, subject-correct scoped dispatch, and equality between a loop-built request and the request reconstructed from its session-log prefix. Global enablement and package-name regex filters belong to the service ([package-owned invariant service](2026-07-19-package-owned-invariant-service.md); [omission decision](../simplification/2026-08-28-omit-unneeded-invariant-companions.md)). When the session companion attaches to an existing or seeded session, it replays the immutable log to rebuild trace state. The service gives each contribution a disposable child fiber, so hot reload is safe in the middle of a turn without giving diagnostics ownership of session storage. @@ -48,12 +48,12 @@ Freezing history only when an invariants plugin is installed would make the core ### Clone only when deriving messages -Detaching `deriveMessages()` would protect the most common request path but leave other readers of `session.events`, append return values, and session-event observers able to mutate durable history. The log must protect its own boundary; derived projections are an additional isolation boundary, not a substitute. +Detaching `deriveMessages()` would protect the most common request path but leave other readers of `snapshotEvents()`, `eventAt()`, append return values, and session-event observers able to mutate durable history. The log must protect its own boundary; derived projections are an additional isolation boundary, not a substitute. ## Consequences - Every accepted live or seeded session event is detached from caller-owned inputs and deeply immutable before any observer can receive it. -- `session.events` exposes stable immutable snapshots instead of the private growing array. +- `snapshotEvents()` exposes stable immutable snapshots instead of the private growing array; `seq` and `eventAt()` serve scalar reads without copying that array. - Request-side mutation cannot reach stored history through derived messages. - Development builds can enable relational assertions without changing storage behavior, and disposing or filtering a companion does not weaken log immutability. - `dsh-invariants` configures global enablement plus package allow/block regex lists; each check remains owned and tested by its product package. diff --git a/.agents/notes/implemented/architecture/2026-06-11-dev-invariants-over-deep-readonly.zh.md b/.agents/notes/implemented/architecture/2026-06-11-dev-invariants-over-deep-readonly.zh.md index 576e53e0b2..c1413e9c89 100644 --- a/.agents/notes/implemented/architecture/2026-06-11-dev-invariants-over-deep-readonly.zh.md +++ b/.agents/notes/implemented/architecture/2026-06-11-dev-invariants-over-deep-readonly.zh.md @@ -22,7 +22,7 @@ TypeScript readonly 类型不是充分的运行时边界。它们在程序运行 `Session` 仅在一次递归遍历完成无损 JSON 快照的物化之后才接受事件。该遍历拒绝不支持的值，并产出进入日志的已分离的确切记录，因此验证与存储不会从有状态的 getter 观察到不同的值，也不会保留调用方拥有的嵌套引用。 -被接受的事件及其所有后代在发布前被深度冻结。`append()` 返回由 Session 拥有的冻结事件，`session/event` 观察者接收同一记录，`session.events` 返回冻结的数组快照。先前返回的数组不会因后续 append 而增长。种子记录在构造成功前经过相同的验证、快照与冻结边界。 +被接受的事件及其所有后代在发布前被深度冻结。`append()` 返回由 Session 拥有的冻结事件，`session/event` 观察者和 `eventAt(seq)` 接收同一记录。`snapshotEvents(fromSeq?, toSeqExclusive?)` 返回冻结的数组快照；先前返回的数组不会因后续 append 而增长。调用方只需要当前长度或单个事件时，`seq` 和 `eventAt()` 不会物化数组。种子记录在构造成功前经过相同的验证、快照与冻结边界。 此保证属于 `Session` 而非可选监听器，因为每种组合都依赖可信的历史。无论是否注册了开发支持插件，生产部署、聚焦测试或自定义嵌入都获得相同的存储语义。 @@ -32,7 +32,7 @@ TypeScript readonly 类型不是充分的运行时边界。它们在程序运行 ### 包拥有的不变式配套插件检查关系 -`dsh-invariants` 注册可配置的 `ctx.invariants` 服务，本身不包含产品检查。每个包发布一个 `./invariant` 所有权配套插件；`dsh-session`、`dsh-agent`、`dsh-scope` 和 `dsh-agent-loop` 目前添加需要跟踪状态或观察另一个 seam 的规则：单调递增的序列号、轮次与步骤嵌套、工具调用/结果配对、合法的 agent（智能体）状态转换、主体正确的作用域分发，以及循环构建的请求与从其会话日志前缀重建的请求之间的相等性。全局启用和包名 regex 过滤器归该服务所有（见[包拥有的不变式服务](2026-07-19-package-owned-invariant-service.zh.md)）。 +`dsh-invariants` 注册可配置的 `ctx.invariants` 服务，本身不包含产品检查。只有拥有可独立观察的运行时关系时，包才发布 `./invariant` 所有权配套插件；没有该关系的包会省略 companion 并在 README 中记录原因。`dsh-session`、`dsh-agent`、`dsh-scope` 和 `dsh-agent-loop` 提供首批需要跟踪状态或观察另一个 seam 的规则：单调递增的序列号、轮次与步骤嵌套、工具调用/结果配对、合法的 agent（智能体）状态转换、主体正确的作用域分发，以及循环构建的请求与从其会话日志前缀重建的请求之间的相等性。全局启用和包名 regex 过滤器归该服务所有（见[包拥有的不变式服务](2026-07-19-package-owned-invariant-service.zh.md)与[省略决策](../simplification/2026-08-28-omit-unneeded-invariant-companions.zh.md)）。 当会话配套插件附加到已有会话或以种子记录初始化的会话时，它回放不可变日志以重建跟踪状态。服务为每项贡献提供一个可 dispose（资源释放）的子 fiber，因此轮次中途热重载是安全的，同时不赋予诊断逻辑对会话存储的所有权。 @@ -48,12 +48,12 @@ TypeScript readonly 类型不是充分的运行时边界。它们在程序运行 ### 仅在派生消息时克隆 -分离 `deriveMessages()` 能保护最常见的请求路径，但 `session.events` 的其他读取者、append 返回值和会话事件观察者仍能修改持久历史。日志必须保护自身的边界；派生投影是额外的隔离边界，而非替代品。 +分离 `deriveMessages()` 能保护最常见的请求路径，但 `snapshotEvents()`、`eventAt()` 的其他读取者、append 返回值和会话事件观察者仍能修改持久历史。日志必须保护自身的边界；派生投影是额外的隔离边界，而非替代品。 ## 后果 - 每个被接受的实时或种子会话事件在任何观察者接收之前，都已从调用方拥有的输入中分离并深度不可变。 -- `session.events` 暴露稳定的不可变快照，而非持续增长的私有数组。 +- `snapshotEvents()` 暴露稳定的不可变快照，而非持续增长的私有数组；`seq` 和 `eventAt()` 为标量读取提供无需复制数组的路径。 - 请求侧的修改无法通过派生消息触及已存储的历史。 - 开发构建可以启用关系断言而不改变存储行为；dispose 或过滤一个配套插件不会削弱日志不可变性。 - `dsh-invariants` 配置全局启用状态以及包名允许/阻止 regex 列表；每项检查仍由其产品包拥有并测试。 diff --git a/.agents/notes/implemented/architecture/2026-06-14-session-persistence.i18n.yaml b/.agents/notes/implemented/architecture/2026-06-14-session-persistence.i18n.yaml index 3bdde9ece6..34927aeae7 100644 --- a/.agents/notes/implemented/architecture/2026-06-14-session-persistence.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-06-14-session-persistence.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-06-14-session-persistence.md -2026-06-14-session-persistence.md: 50ec79de83f0cef4a3ec94b689cc25937e334016 -2026-06-14-session-persistence.zh.md: 7b66aed6f077ac484802cfa1e23e1ba7ac3ae985 +2026-06-14-session-persistence.md: a7e06af78c4a372be7a68f3e0f6dc18e38cbead1 +2026-06-14-session-persistence.zh.md: 6d458d4f4c31793212d674bb406204c3882a25ed diff --git a/.agents/notes/implemented/architecture/2026-06-14-session-persistence.md b/.agents/notes/implemented/architecture/2026-06-14-session-persistence.md index 50ec79de83..a7e06af78c 100644 --- a/.agents/notes/implemented/architecture/2026-06-14-session-persistence.md +++ b/.agents/notes/implemented/architecture/2026-06-14-session-persistence.md @@ -14,22 +14,22 @@ The [event-sourced model](2026-06-11-event-sourced-sessions.md) makes the append Persistence is a **capability seam** with an abstract Service Definition ([capability seams](2026-06-13-capability-seams.md), the `dsh-shell` template), not loop or core logic: -1. **Interface** (`dsh-session-persistence`, `ctx.sessionPersistence`) — an abstract `SessionPersistence` service: `locate`/`create`/`append`/`prepare`/`load`/`inspect`/`readFrom`/`list`/`listSnapshots`. Its persisted unit IS the existing `SessionEvent` (`{ type, seq, time, data }`), reused verbatim — no conversion type. +1. **Interface** (`dsh-session-persistence`, `ctx.sessionPersistence`) — an abstract `SessionPersistence` service: `create`/`open`/`stat`/`list`/`export`, with `create`/`open` returning per-session `SessionHandle`s that carry `read`/`append`/`flush`/`close` ([handle-based seam](2026-08-27-handle-based-session-persistence.md)). Its persisted unit IS the existing `SessionEvent` (`{ type, seq, time, data }`), reused verbatim — no conversion type. 2. **Implementation** (`dsh-session-persistence-jsonl`) — an append-only logical JSONL log per session: a `SessionHeader` line followed by storage records that losslessly represent the contiguous `SessionEvent` stream. Eligible `assistant/chunk` delta runs use packed rows by default; [checksummed Zstandard frames](2026-07-19-zstandard-jsonl-session-logs.md) are the default physical encoding, with raw lines configurable. Key durable, contested choices: - **The canonical durable log persists every `SessionEvent` losslessly, including `assistant/chunk`.** JSONL storage may encode a consecutive delta run as one packed row, but logical readers reconstruct the exact event boundaries, sequence numbers, and timestamps. `deriveMessages()` skips chunks, and a chunk-filtered rollout (Codex's `policy.rs`) is tempting — but `seq = log.length` and validation of `events[i].seq === i` require a *contiguous* logical log; filtering chunks out would leave holes and break both the contract and resume. A chunk-filtered projection is possible later as a derived view with its own renumbering, but it is NOT the canonical log. -- **Append-only; a crashed turn is closed, never truncated.** Flushed events are never rewritten. The [semantic checkpoint policy](../bug-fix/2026-07-21-semantic-session-checkpoints.md) drains the request before model dispatch, a recorded top-level call before tool dispatch, and the complete response/result batch after a step; the loop drains the final turn boundary. Because one interrupted turn may contain substantial valid work, cold inspection preserves its contiguous, parseable events and adds risk-classified error results for unanswered assistant calls, a missing `step/end`, and `turn/end` with `{ kind: 'interrupted' }` to the in-memory logical view. `prepare` or `load` commits those closers before returning a recoverable view; the synthetic results keep resumed provider transcripts valid. Only an incomplete final record is discarded during committed repair; a parse error or sequence gap at or before the last real `turn/end` is corruption and makes the session unloadable. -- **The file backend is canonical while the service remains extensible.** `dsh-session-persistence-jsonl` is the sole first-party provider and passes `runPersistenceContract`; the abstract service and coordinator remain available to out-of-tree providers. The [JSONL-only persistence decision](../simplification/2026-08-30-jsonl-only-session-persistence.md) owns removal of the first-party database provider and its deliberate compatibility cut. +- **Append-only; a crashed turn is closed, never truncated.** Flushed events are never rewritten. The [semantic checkpoint policy](../bug-fix/2026-07-21-semantic-session-checkpoints.md) drains the request before model dispatch, a recorded top-level call before tool dispatch, and the complete response/result batch after a step; the loop drains the final turn boundary. Because one interrupted turn may contain substantial valid work, persistence returns its contiguous, parseable events unmodified; the reader owns balancing — resume computes risk-classified error results for unanswered assistant calls, a missing `step/end`, and `turn/end` with `{ kind: 'interrupted' }` (`interruptedTurnClosers`) and appends them through its write handle, while read-only observers add the same closers in memory. The synthetic results keep resumed provider transcripts valid. Only the incomplete fragment of a torn final append is discarded — complete records recovered from it are durably rewritten by the write path before its first new append; a parse error or sequence gap in the committed prefix is corruption and makes the session unloadable. +- **The file backend is canonical while the service remains extensible.** `dsh-session-persistence-jsonl` is the sole first-party provider and passes `runPersistenceContract`; the abstract service remains available to out-of-tree providers. The [JSONL-only persistence decision](../simplification/2026-08-30-jsonl-only-session-persistence.md) owns removal of the first-party database provider and its deliberate compatibility cut. - **Metadata is out-of-log.** Format version, cwd, and lineage are storage concerns, not replayable conversation state, so they live in a `SessionHeader` owned by `dsh-session` and attached to a `Session` via a new readonly `session.header` — never in `SessionEventMap`, never reaching `deriveMessages()`. `createdAt` is non-negative safe-integer Unix epoch milliseconds: live creation and persistence registration reject fractional values, and JSONL validates the decoded header. The alternative (a merge-extensible `session/meta` event as log line 0) was rejected: an in-log event would ride along with a seeded/forked session for free, but metadata is not replayable state, so the explicit out-of-log header boundary is the cleaner cost. (The header was originally split into an immutable `SessionHeader` plus a mutable `SessionSummary` whose union was `SessionMeta`; the mutable summary was later removed as dead state — see [Drop the mutable session summary](../simplification/2026-06-19-drop-mutable-session-summary.md).) -- **`ctx.agents.create()` and `ctx.agents.resume()` are async factories; resume additionally crosses the persistence boundary.** `ctx.agents.resume({ resumeSessionId })` obtains the exact unpublished Session through `ctx.sessionPersistence.prepare()`, publishes it under the persisted id, and continues its projections. The [Session preparation decision](2026-08-05-session-preparation.md) owns reuse between history inspection and resume. The agent-loop does NOT hard-inject `sessionPersistence` (that would pend non-persistent demos forever); `resume` rejects with a clear error when it is absent. +- **`ctx.agents.create()` and `ctx.agents.resume()` are async factories; resume additionally crosses the persistence boundary.** `ctx.agents.resume({ resumeSessionId })` opens the session's write handle, reads the stored log, and publishes the prepared Session under the persisted id, continuing its projections. The [Session preparation decision](2026-08-05-session-preparation.md) owns the unpublished-Session ownership window. The agent-loop does NOT hard-inject `sessionPersistence` (that would pend non-persistent demos forever); `resume` rejects with a clear error when it is absent. ## Alternatives considered Each key choice above records its rejected alternative where the choice is stated: a **chunk-filtered canonical log** (Codex's `policy.rs` shape) — breaks the contiguous-seq contract; **truncating a crashed turn** — silently destroys a long autonomous run's real work; an **in-log `session/meta` event as log line 0** — metadata is not replayable state; **finite fractional `createdAt` values** — have no producer and diverge from integer Unix-millisecond storage; **hard-injecting `sessionPersistence` into the loop** — would pend non-persistent demos forever. -Format versioning: the header carries a `version`; cold reads reject any non-current version. The pre-release session format stays pinned at `SESSION_FORMAT_VERSION = 0` and carries no broad compatibility promise, while the coordinator may own an explicit narrow import upgrade when persisted user data requires it ([pre-identity message recovery](../bug-fix/2026-07-28-load-pre-identity-session-messages.md)). Append-only + flush is robust to partial trailing writes tolerated during cold preparation; a future provider or write-ahead log needs its own power-loss and recovery contract. +Format versioning: the header carries a `version`; cold reads reject any non-current version. The pre-release session format stays pinned at `SESSION_FORMAT_VERSION = 0` and carries no compatibility promise: reads validate current v0 records only, and retired same-version shapes refuse fail-closed ([export and pre-release trims](../simplification/2026-08-27-persistence-export-and-pre-release-trims.md)). Append-only + flush is robust to partial trailing writes (tolerated during cold preparation) but not to fsync-less power loss mid-line; a DB/WAL backend is the stronger option there. ## Consequences diff --git a/.agents/notes/implemented/architecture/2026-06-14-session-persistence.zh.md b/.agents/notes/implemented/architecture/2026-06-14-session-persistence.zh.md index 7b66aed6f0..6d458d4f4c 100644 --- a/.agents/notes/implemented/architecture/2026-06-14-session-persistence.zh.md +++ b/.agents/notes/implemented/architecture/2026-06-14-session-persistence.zh.md @@ -14,22 +14,22 @@ Status: implemented 持久化是一个具有抽象 Service Definition 的**能力 seam**（[能力 seam](2026-06-13-capability-seams.zh.md)，`dsh-shell` 模板），而非循环或核心逻辑： -1. **接口**（`dsh-session-persistence`，`ctx.sessionPersistence`）：一个抽象的 `SessionPersistence` 服务，提供 `locate`/`create`/`append`/`prepare`/`load`/`inspect`/`readFrom`/`list`/`listSnapshots`。其持久化单元就是现有的 `SessionEvent`（`{ type, seq, time, data }`），原样复用，无转换类型。 +1. **接口**（`dsh-session-persistence`，`ctx.sessionPersistence`）：一个抽象的 `SessionPersistence` 服务，提供 `create`/`open`/`stat`/`list`/`export`，其中 `create`/`open` 返回逐会话的 `SessionHandle`，句柄承载 `read`/`append`/`flush`/`close`（[基于句柄的 seam](2026-08-27-handle-based-session-persistence.zh.md)）。其持久化单元就是现有的 `SessionEvent`（`{ type, seq, time, data }`），原样复用，无转换类型。 2. **实现**（`dsh-session-persistence-jsonl`）：每个会话一个仅追加的逻辑 JSONL 日志：先是一行 `SessionHeader`，随后是无损表示连续 `SessionEvent` 流的存储记录。符合条件的 `assistant/chunk` 增量连续段默认使用打包行；[带校验和的 Zstandard 帧](2026-07-19-zstandard-jsonl-session-logs.zh.md)是默认物理编码，也可通过配置使用原始行。 长期有效、存在争议的关键选择： - **规范的持久日志无损保留每个 `SessionEvent`，包括 `assistant/chunk`。** JSONL 存储可以将一段连续的增量事件编码为一条打包行，但逻辑读取方会重建精确的事件边界、序号与时间戳。`deriveMessages()` 跳过分片，而过滤分片的方案（Codex 的 `policy.rs`）很有吸引力，但 `seq = log.length` 以及 `events[i].seq === i` 验证要求*连续*的逻辑日志；过滤掉分片会留下空洞，同时破坏约定和恢复功能。基于分片过滤的投影可以作为派生视图在后续实现（带有自己的重新编号），但它不是规范日志。 -- **仅追加；崩溃的轮次被关闭，而非截断。** 已刷写的事件永不被重写。[语义检查点策略](../bug-fix/2026-07-21-semantic-session-checkpoints.zh.md)会在调用模型前排空请求、在调用工具前排空已记录的顶层调用，并在步骤结束后排空完整的响应/结果批次；循环则排空最终轮次边界。由于一个被中断的轮次可能包含大量有效工作，冷检查会保留其连续、可解析的事件，并在内存逻辑视图中为未应答的 assistant 调用添加按风险分类的错误结果、补一个缺失的 `step/end`，以及带 `{ kind: 'interrupted' }` 的 `turn/end`。`prepare` 或 `load` 在返回可恢复视图前提交这些收尾事件；合成结果保证恢复后的提供方 transcript（文本记录）仍然有效。只有不完整的最后一条记录会在提交修复时被丢弃；在最后一个真实 `turn/end` 处或之前出现解析错误或序号间隙，属于数据损坏，会使该会话不可加载。 -- **文件后端为规范实现，服务保持可扩展。** `dsh-session-persistence-jsonl` 是唯一 first-party provider，并通过 `runPersistenceContract`；抽象服务与 coordinator 继续供仓库外 provider 使用。[JSONL-only 持久化决策](../simplification/2026-08-30-jsonl-only-session-persistence.zh.md)负责 first-party 数据库 provider 的删除及其明确 compatibility cut。 +- **仅追加；崩溃的轮次被关闭，而非截断。** 已刷写的事件永不被重写。[语义检查点策略](../bug-fix/2026-07-21-semantic-session-checkpoints.zh.md)会在调用模型前排空请求、在调用工具前排空已记录的顶层调用，并在步骤结束后排空完整的响应/结果批次；循环则排空最终轮次边界。由于一个被中断的轮次可能包含大量有效工作，持久化会原样返回其连续、可解析的事件；配平是读方的职责——resume 会为未应答的 assistant 调用计算按风险分类的错误结果、补一个缺失的 `step/end`，以及带 `{ kind: 'interrupted' }` 的 `turn/end`（`interruptedTurnClosers`），并通过其写句柄追加它们，而只读观察方仅在内存中添加同样的收尾事件。合成结果保证恢复后的提供方 transcript（文本记录）仍然有效。只有撕裂的最终 append 中不完整的碎片会被丢弃——从中恢复的完整记录由写路径在第一次新 append 之前持久重写；已提交前缀中的解析错误或序号间隙，属于数据损坏，会使该会话不可加载。 +- **文件后端为规范实现，服务保持可扩展。** `dsh-session-persistence-jsonl` 是唯一 first-party provider，并通过 `runPersistenceContract`；抽象服务继续供仓库外 provider 使用。[JSONL-only 持久化决策](../simplification/2026-08-30-jsonl-only-session-persistence.zh.md)负责 first-party 数据库 provider 的删除及其明确 compatibility cut。 - **元数据在日志之外。** 格式版本、cwd 和谱系是存储关注点，不是可回放的对话状态，因此它们存放在 `dsh-session` 拥有的 `SessionHeader` 中，并通过新的只读属性 `session.header` 附加到 `Session` 上——永远不进入 `SessionEventMap`，永远不到达 `deriveMessages()`。`createdAt` 是以 Unix epoch 毫秒表示的非负安全整数：运行时创建和持久化注册会拒绝小数值，JSONL 会验证解码后的 header。替代方案（一个可合并扩展的 `session/meta` 事件作为日志第 0 行）被否决：日志内事件会自然随 seed/fork 的会话携带，但元数据不是可回放状态，因此显式的日志外 header 边界是更清晰的取舍。（header 最初被拆分为不可变的 `SessionHeader` 加可变的 `SessionSummary`，二者的联合类型为 `SessionMeta`；可变 summary 后来因属于死状态而被移除——见 [移除可变会话摘要](../simplification/2026-06-19-drop-mutable-session-summary.zh.md)。） -- **`ctx.agents.create()` 和 `ctx.agents.resume()` 是异步工厂；恢复还跨越持久化边界。** `ctx.agents.resume({ resumeSessionId })` 通过 `ctx.sessionPersistence.prepare()` 取得精确的未发布 Session，以持久化 id 发布它，并继续其投影。[Session 准备阶段决策](2026-08-05-session-preparation.zh.md)定义历史检查与恢复之间的复用。agent loop（智能体循环）不会硬注入 `sessionPersistence`（那样会让非持久化的演示永远挂起）；当它不存在时，`resume` 会以明确的错误拒绝。 +- **`ctx.agents.create()` 和 `ctx.agents.resume()` 是异步工厂；恢复还跨越持久化边界。** `ctx.agents.resume({ resumeSessionId })` 打开该会话的写句柄，读取已存储的日志，并以持久化 id 发布准备好的 Session，继续其投影。[Session 准备阶段决策](2026-08-05-session-preparation.zh.md)定义未发布 Session 的所有权窗口。agent loop（智能体循环）不会硬注入 `sessionPersistence`（那样会让非持久化的演示永远挂起）；当它不存在时，`resume` 会以明确的错误拒绝。 ## 曾考虑的替代方案 上述每个关键选择都在陈述处记录了被否决的替代方案：**过滤分片的规范日志**（Codex 的 `policy.rs` 形式）破坏连续 seq 约定；**截断崩溃的轮次**会静默销毁长时间自主运行中的真实工作；**日志内 `session/meta` 事件作为第 0 行**——元数据不是可回放状态；**有限的非整数 `createdAt` 值**没有生产方，且与整数 Unix 毫秒存储不一致；**将 `sessionPersistence` 硬注入循环**会让非持久化的演示永远挂起。 -格式版本控制：header 携带一个 `version`；冷读取拒绝任何非当前版本。预发布阶段的会话格式仍固定为 `SESSION_FORMAT_VERSION = 0`，不承诺广泛兼容；当持久化用户数据确有需要时，协调器可以负责显式且范围受限的导入升级（[消息标识机制引入前的消息恢复](../bug-fix/2026-07-28-load-pre-identity-session-messages.zh.md)）。仅追加 + 刷写能承受冷准备时可容忍的尾部不完整写入；未来 provider 或 write-ahead log 需要自有的断电与恢复约定。 +格式版本控制：header 携带一个 `version`；冷读取拒绝任何非当前版本。预发布阶段的会话格式仍固定为 `SESSION_FORMAT_VERSION = 0`，不作兼容承诺：读取只校验当前 v0 记录，已废弃的同版本形态会以 fail-closed 方式拒绝（[导出与预发布精简](../simplification/2026-08-27-persistence-export-and-pre-release-trims.zh.md)）。仅追加 + 刷写对尾部的不完整写入具有健壮性（冷准备时可容忍），但无法抵御未使用 fsync 时在行写入中途断电；数据库/WAL 后端是该场景下更强的选项。 ## 后果 diff --git a/.agents/notes/implemented/architecture/2026-06-20-branded-ids.i18n.yaml b/.agents/notes/implemented/architecture/2026-06-20-branded-ids.i18n.yaml index 2c4a5e3544..e847027821 100644 --- a/.agents/notes/implemented/architecture/2026-06-20-branded-ids.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-06-20-branded-ids.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-06-20-branded-ids.md -2026-06-20-branded-ids.md: 954fd89aa229ba587cd1293973b4038cfeb20473 -2026-06-20-branded-ids.zh.md: 0dd761da2e5b5fc3e864fe03c250b9781be9ee59 +2026-06-20-branded-ids.md: 1f579a7afb7ac5f7facd6c5e8040d5df719c4bed +2026-06-20-branded-ids.zh.md: eaf032027f2f61bec9e4ab624622a5012c97e9b9 diff --git a/.agents/notes/implemented/architecture/2026-06-20-branded-ids.md b/.agents/notes/implemented/architecture/2026-06-20-branded-ids.md index 954fd89aa2..1f579a7afb 100644 --- a/.agents/notes/implemented/architecture/2026-06-20-branded-ids.md +++ b/.agents/notes/implemented/architecture/2026-06-20-branded-ids.md @@ -51,7 +51,7 @@ Kept deliberately narrow per the "not every string needs a brand" policy. Each o - **`ModelId`** (`GenerateOptions.model`, the `LlmRuntime` adapter-registry key) — a real cross-package lookup key (config → agent → llm → adapter); a reasonable next brand, left out only to keep this decision's blast radius focused. - **`ToolName`** (the `ToolRuntime` key) — author-defined, human-readable, and rarely confused with another id; the weakest candidate, likely not worth a brand. - **`ErrorCode`** (`HarnessError.code`) — a closed vocabulary (`ABORTED`, `NO_ADAPTER`, …), not a per-instance id; better served by a string-literal union than a brand, if anything. -- **Numeric ordinals** — turn number, step number, and the event `seq` are `number`, not `string`, so `Branded` does not apply; a parallel `number & { readonly [BRAND]: B }` variant could brand them, but they are positional ordinals rarely passed across boundaries, so the payoff is low. +- **Other numeric ordinals** — the [Session sequence and log-offset decision](2026-08-31-session-sequence-and-log-offset-brands.md) brands event identities and log gaps because they cross persistence and reference seams. Turn and step numbers remain plain numbers: they are payload-local ordinals and are not interchangeable with Session event positions. - **Validated construction** — `brandString()` performs no runtime check, and every boundary (ACP `sessionId`, provider-issued `call.id`, the empty-string fallback in `dsh-llm-deepseek`) trusts the raw string. A `SessionId.parse()` / `isValid()` companion that throws on malformed input at boundaries is a genuine gap, but it is a runtime-behavior change with its own design (what is "malformed"? what happens on failure?) and belongs in its own decision. ## Verification diff --git a/.agents/notes/implemented/architecture/2026-06-20-branded-ids.zh.md b/.agents/notes/implemented/architecture/2026-06-20-branded-ids.zh.md index 0dd761da2e..eaf032027f 100644 --- a/.agents/notes/implemented/architecture/2026-06-20-branded-ids.zh.md +++ b/.agents/notes/implemented/architecture/2026-06-20-branded-ids.zh.md @@ -51,7 +51,7 @@ const owner = brandString('session-1') - **`ModelId`**（`GenerateOptions.model`，`LlmRuntime` 适配器注册表的键）：一个真正的跨包查找键（config → agent → llm → 适配器）；合理的下一个 brand，仅为控制本决策的影响范围而暂不纳入。 - **`ToolName`**（`ToolRuntime` 的键）：由作者定义、人类可读，且很少与其他 id 混淆；最弱的候选，可能不值得加 brand。 - **`ErrorCode`**（`HarnessError.code`）：一个封闭词汇（`ABORTED`、`NO_ADAPTER`……），不是逐实例的 id；如果要做，string 字面量联合类型比 brand 更合适。 -- **数值序号**：轮次号、步骤号和事件 `seq` 是 `number` 而非 `string`，`Branded` 不适用；可以用并行的 `number & { readonly [BRAND]: B }` 变体来 brand 它们，但它们是位置序号、很少跨边界传递，收益较低。 +- **其他数值序号**：[Session 序列号与日志偏移决策](2026-08-31-session-sequence-and-log-offset-brands.zh.md)会为事件身份与日志间隙加 brand，因为它们跨越 persistence 与引用 seam。turn 与 step number 保持普通 number：它们是 payload-local ordinal，不会与 Session 事件位置互换。 - **带校验的构造**：`brandString()` 不执行运行时检查，且每个边界（ACP `sessionId`、提供方签发的 `call.id`、`dsh-llm-deepseek` 中的空字符串回退）都信任裸 string。一个在边界处对格式错误的输入抛异常的 `SessionId.parse()` / `isValid()` 配套工具确实是缺口，但它属于运行时行为变更，有自己的设计问题（什么算「格式错误」？失败时会怎样？），应在独立决策中处理。 ## 验证 diff --git a/.agents/notes/implemented/architecture/2026-07-19-package-invariant-runtime-contracts.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-19-package-invariant-runtime-contracts.i18n.yaml index 6ec0a2a7e5..ec285a946f 100644 --- a/.agents/notes/implemented/architecture/2026-07-19-package-invariant-runtime-contracts.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-19-package-invariant-runtime-contracts.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-19-package-invariant-runtime-contracts.md -2026-07-19-package-invariant-runtime-contracts.md: b5799a37a61244193b46db6ea4ae15f306d144b2 -2026-07-19-package-invariant-runtime-contracts.zh.md: e6035bbabba7188017746c57c5b6a48761710658 +2026-07-19-package-invariant-runtime-contracts.md: a1b635cc40844f1846c04e203dbb842d1c7328ed +2026-07-19-package-invariant-runtime-contracts.zh.md: d315c440f1c100911386f57d6f82d5f16be631ca diff --git a/.agents/notes/implemented/architecture/2026-07-19-package-invariant-runtime-contracts.md b/.agents/notes/implemented/architecture/2026-07-19-package-invariant-runtime-contracts.md index b5799a37a6..a1b635cc40 100644 --- a/.agents/notes/implemented/architecture/2026-07-19-package-invariant-runtime-contracts.md +++ b/.agents/notes/implemented/architecture/2026-07-19-package-invariant-runtime-contracts.md @@ -14,20 +14,20 @@ Some packages genuinely own no continuously observable relation. Pure utilities, ## Decision -### Registration is exhaustive; assertions must be meaningful +### Published assertions must be meaningful -Every workspace package publishes a separately built `./invariant` companion and registers its exact npm package name. A companion does one of two things: +A workspace package publishes a separately built `./invariant` companion only when it owns an independently observable runtime relationship. A published companion: -- installs a package-owned check over an event stream or relevant mutable data structure and reports violations through its bound `fail(message)` reporter; or -- uses an empty installer whose declaration has an owner-specific `No runtime invariant:` comment explaining why the package has no plausible runtime relation to observe. +- installs a package-owned check over an event stream or relevant mutable data structure and reports violations through its bound `fail(message)` reporter; and +- registers the package's exact npm name while keeping diagnostics outside the root entrypoint. -The empty form is an explicit architectural conclusion, not a generated placeholder. A future package change that introduces mutable state or an event protocol must replace the explanation with the corresponding check. +When no plausible relationship exists, the package omits the companion and publication wiring and records its package-specific reason in the README. A future change that introduces an independently observable relationship must replace the explanation with the corresponding check. The omission mechanics and current audit are owned by the [omit-unneeded-companions decision](../simplification/2026-08-28-omit-unneeded-invariant-companions.md). The central `dsh-invariants` service owns only configuration, registration uniqueness, child-fiber lifecycle, rollback, disposal, and package-attributed failure. It exposes no generic plugin-shape, service-shape, or startup-assertion helpers and imports no product package. -### Implemented checks +### Representative implemented checks -The current 103-package workspace has 21 executable companions and 82 justified empty companions. +Published companions are enumerated mechanically by `verify-package-invariants`; the current audit count is recorded in the [omit-unneeded-companions decision](../simplification/2026-08-28-omit-unneeded-invariant-companions.md). The table below samples representative runtime relationships rather than listing every companion. | Owner | Runtime relationship | |---|---| @@ -57,13 +57,13 @@ Session-backed companions validate existing durable events when they load, using ### Repository gate and tests -`verify-package-invariants` discovers every workspace package and enforces companion source, exact-name registration, named-only Loader shape, `./invariant` exports, publication files, dependencies, TypeScript references, and bundle entries. Its AST rule rejects generated markers, default exports, and unexplained empty installers. A non-empty installer must accept and use the failure reporter, and registration must pass that checked local `install` function. The gate deliberately does not infer semantic quality from method names or helper calls. +`verify-package-invariants` discovers every workspace package. It accepts clean omission, rejects stale or partial companion wiring, and enforces exact-name registration, named-only Loader shape, `./invariant` exports, publication files, dependencies, TypeScript references, and bundle entries for published companions. Its AST rule rejects generated markers, default exports, and empty installers. Every installer must accept and use the failure reporter, and registration must pass that checked local `install` function. The gate deliberately does not infer semantic quality from method names or helper calls. -Vitest mounts `InvariantRegistry` with `{ enabled: true }` for every package test topology and loads the owning companion. The invariant subpath path mapping resolves source companions instead of stale built output. Focused suites cover every executable companion's valid and invalid observations, and the exhaustive topology runs every source companion through the real Loader namespace normalization. After the structural gate validates each publication map, an artifact gate stages its manifest-declared `lib/` files, imports the compiled `./invariant` self-reference under plain Node, and repeats that Loader-shape check, so a companion that imports an undeclared runtime chunk fails before release. Tests that synthesize event streams must produce a valid surrounding lifecycle unless the test is intentionally asserting a violation. +Vitest mounts `InvariantRegistry` with `{ enabled: true }` for every package test topology and loads the owning companion when one is published. The invariant subpath path mapping resolves source companions instead of stale built output. Focused suites cover every published companion's valid and invalid observations, and the exhaustive topology runs every source companion through real Loader namespace normalization. After the structural gate validates each publication map, an artifact gate stages its manifest-declared `lib/` files, imports the compiled `./invariant` self-reference under plain Node, and repeats that Loader-shape check, so a companion that imports an undeclared runtime chunk fails before release. Tests that synthesize event streams must produce a valid surrounding lifecycle unless the test is intentionally asserting a violation. ## Alternatives considered -- **Keep generated empty companions.** Rejected because an unexplained placeholder can survive after a package gains a meaningful runtime relation. +- **Keep explained empty companions.** Rejected because source, publication, dependency, and test wiring are disproportionate machinery for a negative conclusion that belongs in the package README. - **Require an assertion from every package.** Rejected because method-presence, plugin-shape, and fixed-example assertions duplicate stronger type, load, and unit-test contracts without checking runtime consistency. - **Keep generic shape helpers in the service.** Rejected because they blur compile-time API validation with runtime invariants and encourage centrally defined product assumptions. - **Move the product checks into the service.** Rejected because product vocabulary, dependencies, tests, and change ownership belong with the package that emits the data. @@ -71,8 +71,8 @@ Vitest mounts `InvariantRegistry` with `{ enabled: true }` for every package tes ## Consequences -- Every package has visible ownership and publication wiring, but only packages with a plausible runtime relation add listeners or trace state. -- Empty companions remain reviewable decisions with package-specific explanations and fail the gate if the explanation is removed. +- Packages with a plausible runtime relation have visible ownership and publication wiring; packages without one record the omission reason in their README. +- Empty companions fail the gate, and partial omission wiring fails before build or release. - Type declarations, Cordis loadability, plugin metadata, service method APIs, and pure algebra remain covered by their owning compile, load, unit, or integration gates. - Runtime failures identify the owning npm package and point to an inconsistent observation rather than restating a required API shape. - The original selection, blocklist precedence, duplicate ownership, rollback, disposal, and HMR service contracts remain unchanged. diff --git a/.agents/notes/implemented/architecture/2026-07-19-package-invariant-runtime-contracts.zh.md b/.agents/notes/implemented/architecture/2026-07-19-package-invariant-runtime-contracts.zh.md index e6035bbabb..d315c440f1 100644 --- a/.agents/notes/implemented/architecture/2026-07-19-package-invariant-runtime-contracts.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-19-package-invariant-runtime-contracts.zh.md @@ -14,20 +14,20 @@ Status: implemented ## 决策 -### 注册必须全覆盖；断言必须有意义 +### 已发布的断言必须有意义 -每个 workspace 包都发布单独构建的 `./invariant` companion，并用完整 npm 包名注册。companion 只能采用以下两种形式之一： +只有拥有可独立观察的运行时关系时，workspace 包才发布单独构建的 `./invariant` companion。已发布 companion 必须： -- 安装包自有的事件流或相关可变数据结构检查，并通过绑定的 `fail(message)` 报告器报告违规；或 -- 使用空安装器，并在其声明前写一条该包专属的 `No runtime invariant:` 注释，说明为什么该包没有合理的运行时关系可供观测。 +- 安装包自有的事件流或相关可变数据结构检查，并通过绑定的 `fail(message)` 报告器报告违规；并且 +- 用该包的准确 npm 包名注册，同时保持诊断逻辑不进入根入口。 -空形式是明确的架构结论，不是生成占位符。如果后续包变更引入可变状态或事件协议，就必须用相应检查替换该说明。 +没有合理关系时，包会省略 companion 与发布接线，并在 README 中记录该包的具体原因。如果后续变更引入可独立观察的关系，就必须用相应检查替换该说明。省略机制与当前审计由[省略不必要 companion 的决策](../simplification/2026-08-28-omit-unneeded-invariant-companions.zh.md)负责。 中央 `dsh-invariants` 服务只负责配置、注册唯一性、子 fiber 生命周期、回滚、dispose（资源释放）和归属到包的失败。它不暴露通用插件形状、服务形状或启动断言 helper，也不导入产品包。 -### 已实施的检查 +### 已实施检查示例 -当前 103 个包的 workspace 包含 21 个可执行 companion 和 82 个有理由的空 companion。 +已发布 companion 由 `verify-package-invariants` 机械枚举；当前审计数量记录在[省略不必要 companion 的决策](../simplification/2026-08-28-omit-unneeded-invariant-companions.zh.md)中。下表仅展示有代表性的运行时关系，不会逐项列出所有 companion。 | 所有者 | 运行时关系 | |---|---| @@ -57,13 +57,13 @@ Status: implemented ### 仓库门禁与测试 -`verify-package-invariants` 发现每个 workspace 包，并强制 companion 源文件、完整名称注册、仅含具名 export 的 Loader 形状、`./invariant` export、发布文件、依赖、TypeScript reference 和 bundle entry 完整。其 AST 规则拒绝生成标记、默认导出和没有解释的空安装器。非空安装器必须接收并使用失败报告器，注册时还必须传入该经检查的本地 `install` 函数。门禁不会通过方法名或 helper 调用推断语义质量。 +`verify-package-invariants` 发现每个 workspace 包。它接受完整省略，拒绝陈旧或不完整的 companion 接线，并对已发布 companion 强制完整名称注册、仅含具名 export 的 Loader 形状、`./invariant` export、发布文件、依赖、TypeScript reference 和 bundle entry 完整。其 AST 规则拒绝生成标记、默认导出和空 installer。每个 installer 都必须接收并使用失败报告器，注册时还必须传入该经检查的本地 `install` 函数。门禁不会通过方法名或 helper 调用推断语义质量。 -Vitest 为每个包测试拓扑使用 `{ enabled: true }` 挂载 `InvariantRegistry`，并加载所有者 companion。不变量 subpath 的 path mapping 会解析源 companion，而不是陈旧的构建输出。聚焦 suite 覆盖每个可执行 companion 的有效和无效观测；穷举拓扑通过真实 Loader 命名空间归一化运行每个源 companion。结构门禁验证每个包的发布映射后，产物门禁会暂存其 manifest（元数据清单）声明的 `lib/` 文件，在 plain Node 下导入已编译的 `./invariant` 自引用，并重复执行该 Loader 形状检查；这样，若 companion 导入未声明的运行时分片，门禁就会在发布前失败。合成事件流的测试必须构造有效的外围生命周期，除非测试本身就是在断言违规。 +Vitest 为每个包测试拓扑使用 `{ enabled: true }` 挂载 `InvariantRegistry`，并在所有者发布 companion 时加载它。不变量 subpath 的 path mapping 会解析源 companion，而不是陈旧的构建输出。聚焦 suite 覆盖每个已发布 companion 的有效和无效观测；穷举拓扑通过真实 Loader 命名空间归一化运行每个源 companion。结构门禁验证每个包的发布映射后，产物门禁会暂存其 manifest（元数据清单）声明的 `lib/` 文件，在 plain Node 下导入已编译的 `./invariant` 自引用，并重复执行该 Loader 形状检查；这样，若 companion 导入未声明的运行时分片，门禁就会在发布前失败。合成事件流的测试必须构造有效的外围生命周期，除非测试本身就是在断言违规。 ## 考虑过的替代方案 -- **保留生成的空 companion。** 拒绝，因为包获得有意义的运行时关系后，没有解释的占位符仍可能继续存在。 +- **保留带说明的空 companion。** 拒绝，因为只为表达 README 可以直接记录的否定结论而保留源码、发布、依赖与测试接线，成本过高。 - **要求每个包都执行断言。** 拒绝，因为方法存在性、插件形状和固定示例断言会重复更强的类型、加载和单元测试约定，却没有检查运行时一致性。 - **在服务中保留通用形状 helper。** 拒绝，因为这会混淆编译期 API 验证和运行时不变量，并鼓励在中央定义产品假设。 - **把产品检查移入服务。** 拒绝，因为产品词汇、依赖、测试和变更所有权应归属于产生这些数据的包。 @@ -71,8 +71,8 @@ Vitest 为每个包测试拓扑使用 `{ enabled: true }` 挂载 `InvariantRegis ## 后果 -- 每个包都有可见的所有权与发布 wiring，但只有具备合理运行时关系的包才会增加 listener 或 trace 状态。 -- 空 companion 是带包专属说明、可评审的决策；删除说明后门禁会失败。 +- 拥有合理运行时关系的包具有可见的所有权与发布 wiring；没有该关系的包会在 README 中记录省略原因。 +- 空 companion 会让门禁失败，不完整的省略接线也会在构建或发布前失败。 - 类型声明、Cordis 可加载性、插件 metadata、服务方法 API 和纯代数继续由所属的编译、加载、单元或集成门禁覆盖。 - 运行时失败会标明所属 npm 包，并指出不一致的观测，而不是复述必要的 API 形状。 - 原有 selection、blocklist 优先级、重复所有权、回滚、dispose 和 HMR（热模块替换）服务约定保持不变。 diff --git a/.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.i18n.yaml index 8e3a2bc01c..8a58bc4d9b 100644 --- a/.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.md -2026-07-19-package-owned-invariant-service.md: 88fb870eac6f72307fa2bfeaa714fe59a6a9130d -2026-07-19-package-owned-invariant-service.zh.md: 46b7789034c4a359fed95744e403f0a3a4dee6de +2026-07-19-package-owned-invariant-service.md: b955c99a2576b6b2f8208a16ad2792af181c3468 +2026-07-19-package-owned-invariant-service.zh.md: 4fc0fb5d615753c0c057e927f59359847fc1328f diff --git a/.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.md b/.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.md index 88fb870eac..b955c99a25 100644 --- a/.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.md +++ b/.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.md @@ -10,7 +10,7 @@ Runtime invariant checks span session traces, agent state, scoped dispatch, and Deployments that opt into diagnostics need more than presence or absence of one plugin. Such a composition carries the known invariant contributions while permitting a global off switch and package-selective diagnostics. Selection must remain stable when a package loads later or reloads under HMR, and disabled contributions must not allow two plugins to claim the same package name silently. -Package ownership must also be exhaustive. Without a mechanical repository rule, a new package can omit the companion, dependency, or publication wiring and remain invisible to diagnostics until a maintainer notices the gap. +Published ownership must be mechanically complete. Without a repository rule, a package can expose a partial companion, dependency, or publication map and remain broken until a maintainer notices the gap; packages that publish none must keep their reason reviewable in the README. ## Decision @@ -18,7 +18,7 @@ Package ownership must also be exhaustive. Without a mechanical repository rule, `@deepseek-ai/dsh-invariants` is a product-independent Cordis service plugin that registers `ctx.invariants`. It owns configuration, registration uniqueness, child-fiber lifecycle, and package-attributed failures. It imports no session, agent, scope, or agent-loop package and contains none of their checks. -Every workspace package publishes a `./invariant` companion plugin that registers its exact full npm name. A companion checks a meaningful event or mutable-data relationship when its owner has one; otherwise it carries an owner-specific explanation for its empty installer. Generated ownership placeholders and synthetic API-shape assertions are forbidden by the follow-up [runtime-contract Agent Note](2026-07-19-package-invariant-runtime-contracts.md). Package root entrypoints do not import or register diagnostics implicitly, so loading a root package does not change runtime checking or require the invariant service. +A workspace package publishes a `./invariant` companion plugin only when it owns an independently observable event or mutable-data relationship. The companion registers its exact full npm name. Packages without such a relationship omit the companion and publication wiring and record the reason in their README; generated placeholders, empty installers, and synthetic API-shape assertions are forbidden by the [runtime-contract Agent Note](2026-07-19-package-invariant-runtime-contracts.md) and [omission decision](../simplification/2026-08-28-omit-unneeded-invariant-companions.md). Package root entrypoints do not import or register diagnostics implicitly, so loading a root package does not change runtime checking or require the invariant service. ### Configuration and selection @@ -64,9 +64,9 @@ The former functional-plugin entry point and one-argument `InvariantError` const | `@deepseek-ai/dsh-scope/invariant` | `@deepseek-ai/dsh-scope` | scoped-event carrier presence and subject consistency | | `@deepseek-ai/dsh-agent-loop/invariant` | `@deepseek-ai/dsh-agent-loop` | model-request reconstruction | -These four owners supplied the initial stateful checks. The follow-up runtime-contract decision adds checks for seventeen more owners with real event or mutable-data relationships and records justified empty companions for the rest. Every companion is a separately bundled `./invariant` export with its own declarations and Loader-safe namespace plugin shape; the service package's own companion imports its local service type to avoid a self-dependency. +These four owners supplied the initial stateful checks. Later owners add companions for real event or mutable-data relationships, while packages without one omit the companion and document why. Every published companion is a separately bundled `./invariant` export with its own declarations and Loader-safe namespace plugin shape. -`verify-package-invariants` discovers every workspace package and rejects missing companion source, generated markers, unexplained empty installers, non-empty installers that omit or ignore the reporter, foreign or unresolved registration names, missing `./invariant` exports or published files, missing invariant peer/development dependencies and project references, and bundle overrides that omit the companion entry. +`verify-package-invariants` discovers every workspace package, accepts clean omission, and rejects partial companion wiring, generated markers, empty installers, installers that omit or ignore the reporter, foreign or unresolved registration names, missing `./invariant` exports or published files, missing invariant peer/development dependencies and project references, and bundle overrides that omit a published companion entry. ### Scoped-event semantic map @@ -84,7 +84,7 @@ Service tests cover defaults, global disablement, allow/block selection, blockli Composition tests cover standard-spine forwarding and generated SDK entries. Loader tests preserve each companion namespace, while built plain-Node smokes exercise the compiled subpath exports. The scoped-event freshness gate reruns its semantic Program analysis. -Every Vitest configuration loads a test host that mounts an explicitly enabled service before an ordinary Cordis root's first plugin and adds the current test package's companion. One exhaustive topology mounts all package companions once; focused service and owner tests construct their own invariant topology so they can exercise disablement, filtering, rollback, and reload without duplicate ownership. Gate tests also execute every companion's `apply` function and verify that it calls `register` with its manifest name, rather than accepting source text alone. +Every Vitest configuration loads a test host that mounts an explicitly enabled service before an ordinary Cordis root's first plugin and adds the current test package's companion when one exists. One exhaustive topology mounts all published companions once; focused service and owner tests construct their own invariant topology so they can exercise disablement, filtering, rollback, and reload without duplicate ownership. Gate tests also execute every published companion's `apply` function and verify that it calls `register` with its manifest name, rather than accepting source text alone. ## Alternatives considered @@ -96,10 +96,10 @@ Every Vitest configuration loads a test host that mounts an explicitly enabled s ## Consequences - Product packages own and test their relational assertions while the service stays product-independent. -- Every package pays the publication and dependency cost of a companion; only owners with a meaningful runtime relationship add listener or trace-state cost. +- Only owners with a meaningful runtime relationship pay the publication, dependency, listener, or trace-state cost of a companion; other packages record the omission reason in their README. - Compositions that mount the diagnostics can disable all checks or select package names without changing their plugin tree. - Explicit companion entries make diagnostic cost and ownership visible in Cordis config and package exports. -- One selected executable contribution adds one child fiber and its listener/state cost; a selected empty contribution has no listener or trace-state cost, while filtered registrations retain only name ownership. +- One selected contribution adds one child fiber and its listener/state cost, while filtered registrations retain only name ownership. - Regex sources are deployment configuration and remain fixed until the service reloads. -- Ordinary Vitest roots install the owning test package's selected companion; one exhaustive topology pays the full child-fiber cost once for repository-wide registration coverage. +- Ordinary Vitest roots install the owning test package's selected companion when published; one exhaustive topology pays the full child-fiber cost once for repository-wide registration coverage. - Session storage validation, snapshotting, freezing, cited source-event validation, and surface acceptance remain always on and are not affected by invariant selection. diff --git a/.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.zh.md b/.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.zh.md index 46b7789034..4fc0fb5d61 100644 --- a/.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.zh.md @@ -10,7 +10,7 @@ Status: implemented 选择启用诊断的部署还需要比“是否加载一个插件”更细的控制。这类组合会携带已知的不变式贡献，同时允许全局关闭或按包选择诊断。包稍后加载或在 HMR（热模块替换）下重载时，选择结果必须保持稳定；被过滤的贡献也不能让两个插件静默占用同一个包名。 -包所有权还必须覆盖完整。若没有机械化的仓库规则，新包可能遗漏伴随插件、依赖或发布配置，并一直不会进入诊断范围，直到维护者发现这一缺口。 +已发布的包所有权必须机械完整。若没有仓库规则，包可能暴露不完整的 companion、依赖或发布映射，并一直保持损坏，直到维护者发现；不发布 companion 的包则必须在 README 中保留可评审的原因。 ## 决策 @@ -18,7 +18,7 @@ Status: implemented `@deepseek-ai/dsh-invariants` 是与产品无关的 Cordis 服务插件，注册 `ctx.invariants`。它只负责配置、注册唯一性、子 fiber 生命周期和带包归属的失败；不导入 session、agent、scope 或 agent-loop 包，也不包含这些包的检查。 -工作区内的每个包都发布 `./invariant` 伴随插件，注册自己完整且准确的 npm 包名。如果所有者具备有意义的事件或可变数据关系，companion 就检查该关系；否则空 installer 必须携带该所有者专属的说明。后续的[运行时约定 Agent Note](2026-07-19-package-invariant-runtime-contracts.zh.md) 禁止生成的所有权占位符和合成 API 形状断言。包的根入口不会隐式导入或注册诊断，因此加载根包不会改变运行时检查，也不要求不变式服务存在。 +只有拥有可独立观察的事件或可变数据关系时，工作区包才发布 `./invariant` 伴随插件；该 companion 会注册自己完整且准确的 npm 包名。没有该关系的包会省略 companion 与发布接线，并在 README 中记录原因；[运行时约定 Agent Note](2026-07-19-package-invariant-runtime-contracts.zh.md) 与[省略决策](../simplification/2026-08-28-omit-unneeded-invariant-companions.zh.md)禁止生成占位符、空 installer 和合成 API 形状断言。包的根入口不会隐式导入或注册诊断，因此加载根包不会改变运行时检查，也不要求不变式服务存在。 ### 配置与选择 @@ -64,9 +64,9 @@ blocklist 匹配优先于 allowlist 匹配。每个条目都是区分大小写 | `@deepseek-ai/dsh-scope/invariant` | `@deepseek-ai/dsh-scope` | 作用域事件载体的存在性与主体一致性 | | `@deepseek-ai/dsh-agent-loop/invariant` | `@deepseek-ai/dsh-agent-loop` | 模型请求重建 | -这四个所有者提供了首批有状态检查。后续运行时约定决策为另外十七个确有事件或可变数据关系的所有者增加检查，并为其余包记录有理由的空 companion。每个伴随入口都是单独打包的 `./invariant` export，具有独立声明和对 Loader 安全的命名空间插件形态；服务包自身的伴随插件导入本地服务类型，避免形成自依赖。 +这四个所有者提供了首批有状态检查。后续所有者会为真实事件或可变数据关系增加 companion，没有该关系的包则省略 companion 并记录原因。每个已发布伴随入口都是单独打包的 `./invariant` export，具有独立声明和对 Loader 安全的命名空间插件形态。 -`verify-package-invariants` 会发现每个工作区包，并拒绝缺失的伴随插件源码、生成标记、没有解释的空 installer、缺少或不使用失败报告器的非空 installer、外部或无法解析的注册名、缺失的 `./invariant` export 或发布文件、缺失的不变式对等依赖（peer dependency）、开发依赖及项目引用，以及遗漏伴随入口的自定义构建配置。 +`verify-package-invariants` 会发现每个工作区包，接受完整省略，并拒绝不完整的 companion 接线、生成标记、空 installer、缺少或不使用失败报告器的 installer、外部或无法解析的注册名、缺失的 `./invariant` export 或发布文件、缺失的不变式对等依赖（peer dependency）、开发依赖及项目引用，以及遗漏已发布伴随入口的自定义构建配置。 ### 作用域事件语义映射 @@ -84,7 +84,7 @@ Workspace 约束识别独立的不变式 bundle；包 exports、项目引用、 组合测试覆盖标准主干转发和生成的 SDK 条目。Loader 测试固定每个伴随命名空间，构建后的纯 Node 冒烟测试覆盖编译子路径 export。作用域事件新鲜度门禁会重新执行语义 Program 分析。 -每个 Vitest 配置都会加载测试宿主；在普通 Cordis 根上下文启动第一个插件之前，宿主会挂载显式启用的服务，并添加当前测试包的伴随插件。一个完整拓扑会一次挂载所有包的伴随插件；服务与所有者的聚焦测试自行构建不变式拓扑，从而在不发生重复所有权冲突的前提下覆盖关闭、过滤、回滚与重载。门禁测试还会执行每个伴随插件的 `apply` 函数，并验证它调用 `register` 时使用 manifest（元数据清单）中的包名，而不是只检查源码文本。 +每个 Vitest 配置都会加载测试宿主；在普通 Cordis 根上下文启动第一个插件之前，宿主会挂载显式启用的服务，并在当前测试包存在伴随插件时添加它。一个完整拓扑会一次挂载所有已发布伴随插件；服务与所有者的聚焦测试自行构建不变式拓扑，从而在不发生重复所有权冲突的前提下覆盖关闭、过滤、回滚与重载。门禁测试还会执行每个已发布伴随插件的 `apply` 函数，并验证它调用 `register` 时使用 manifest（元数据清单）中的包名，而不是只检查源码文本。 ## 考虑过的替代方案 @@ -96,10 +96,10 @@ Workspace 约束识别独立的不变式 bundle；包 exports、项目引用、 ## 后果 - 产品包拥有并测试自己的关系断言，服务保持与产品无关。 -- 每个包都承担 companion 的发布与依赖成本；只有具备有意义运行时关系的所有者才增加 listener 或 trace 状态成本。 +- 只有具备有意义运行时关系的所有者才承担 companion 的发布、依赖、listener 或 trace 状态成本；其他包在 README 中记录省略原因。 - 挂载诊断的组合无需改变插件树即可关闭全部检查或按包名选择。 - 显式伴随条目让诊断成本和所有权在 Cordis 配置与包 export 中可见。 -- 每个选中的可执行贡献增加一个子 fiber 及其 listener/状态成本；选中的空贡献不增加 listener 或 trace 状态成本，被过滤注册则只保留包名占用。 +- 每个选中贡献增加一个子 fiber 及其 listener/状态成本，被过滤注册则只保留包名占用。 - 正则表达式源属于部署配置，在服务重载前保持固定。 -- 普通 Vitest 根上下文会安装当前测试包中被选中的伴随插件；一个完整拓扑只支付一次全部子 fiber 成本，用于覆盖整个仓库的注册。 +- 当前测试包发布伴随插件时，普通 Vitest 根上下文会安装其中被选中的伴随插件；一个完整拓扑只支付一次全部子 fiber 成本，用于覆盖整个仓库的注册。 - 会话存储验证、快照、冻结、引用的源事件验证与 surface 接受规则始终启用，不受不变式选择影响。 diff --git a/.agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.i18n.yaml index f53ede9338..38f55f03e8 100644 --- a/.agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.md -2026-07-19-zstandard-jsonl-session-logs.md: 93fc20f931c75552352834b9340e7d38680d4254 -2026-07-19-zstandard-jsonl-session-logs.zh.md: d58f89430ab91de6beabba83c2a31f43e4a7d275 +2026-07-19-zstandard-jsonl-session-logs.md: 33486251a8b018cda61a2845a55218f13c38072b +2026-07-19-zstandard-jsonl-session-logs.zh.md: 5e6b4a1ed7cf5869e1700c2884898b9899452a30 diff --git a/.agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.md b/.agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.md index 93fc20f931..33486251a8 100644 --- a/.agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.md +++ b/.agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.md @@ -32,7 +32,7 @@ A frame-boundary scanner reads the standard magic, variable header fields, block Listing reads in bounded chunks only until the first complete frame is available, validates and decompresses that header frame, and never reads an event frame. The dedicated header frame therefore preserves metadata-only listing even for very large session logs. -EOF inside the final frame is a recoverable torn tail. After the scanner establishes that boundary, a dedicated prefix decoder uses `finishFlush: ZSTD_e_flush` so Node emits available plaintext without requiring frame or checksum completion; every complete newline-terminated event it emits is retained. Repair truncates from that frame's starting byte and appends one new checksummed frame containing the recovered complete events followed by the coordinator's synthetic tool, step, and turn closers. If the tear occurs before any complete event is decodable, repair drops the partial frame and retains all prior complete frames. +EOF inside the final frame is a torn tail. The frame belongs to an append that never resolved, so none of its records were acknowledged durable: repair truncates from that frame's starting byte, retains all prior complete frames, and appends the coordinator's synthetic tool, step, and turn closers as one new checksummed frame ([export and pre-release trims](../simplification/2026-08-27-persistence-export-and-pre-release-trims.md) owns dropping the earlier partial-plaintext salvage). ### Consumers and verification diff --git a/.agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.zh.md b/.agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.zh.md index d58f89430a..5e6b4a1ed7 100644 --- a/.agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.zh.md @@ -32,7 +32,7 @@ JSONL 持久化后端会逐字保留每个 `SessionEvent`，其中包括数量 列举只按有界分片读取到第一个完整帧可用为止，验证并解压该头部帧，绝不读取事件帧。因此，即使会话日志很大，专用头部帧仍能维持仅元数据列举。 -最终帧内部遇到 EOF 属于可恢复的撕裂尾部。扫描器确定该边界后，专用前缀解码器会使用 `finishFlush: ZSTD_e_flush`，使 Node 不必等到帧结束或读到完整校验和就能产出已有明文；其中每个完整且以换行结束的事件都会保留。修复从该帧起始字节截断，再追加一个新的带校验和帧，其中依次包含恢复出的完整事件，以及协调器生成的工具、步骤与轮次闭合事件。如果撕裂位置尚不足以解码任何完整事件，修复会丢弃该不完整帧并保留此前全部完整帧。 +最终帧内部遇到 EOF 属于撕裂尾部。该帧属于一次从未完成结算的追加，因此其中没有任何记录被确认为持久：修复从该帧起始字节截断，保留此前全部完整帧，并把协调器生成的工具、步骤与轮次闭合事件作为一个新的带校验和帧追加（对早先部分明文抢救路径的移除由[导出与预发布精简](../simplification/2026-08-27-persistence-export-and-pre-release-trims.zh.md)负责）。 ### 消费方与验证 diff --git a/.agents/notes/implemented/architecture/2026-07-24-project-session-directories.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-24-project-session-directories.i18n.yaml index c20fc92181..e1483287f0 100644 --- a/.agents/notes/implemented/architecture/2026-07-24-project-session-directories.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-24-project-session-directories.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-24-project-session-directories.md -2026-07-24-project-session-directories.md: 0aa3f513d5a1bb3e44cf33a0ae1eb791ee3a46c2 -2026-07-24-project-session-directories.zh.md: 932b1d29c41d2a854abfc0bab0e47a0ff8c96fe9 +2026-07-24-project-session-directories.md: a37f9231167822e409308f8da60f6c1e837c74d5 +2026-07-24-project-session-directories.zh.md: 469567764219d7baabea89bd94aecd81bd0e5ab3 diff --git a/.agents/notes/implemented/architecture/2026-07-24-project-session-directories.md b/.agents/notes/implemented/architecture/2026-07-24-project-session-directories.md index 0aa3f513d5..a37f923116 100644 --- a/.agents/notes/implemented/architecture/2026-07-24-project-session-directories.md +++ b/.agents/notes/implemented/architecture/2026-07-24-project-session-directories.md @@ -29,7 +29,7 @@ Case-insensitive filesystems can also make differently cased project keys refer The configured root remains a deployment choice. The layout neither selects a global root nor requires projects to share one. When a deployment does centralize storage, project paths remain recognizable; a project-local root uses the same deterministic structure. -The encoded session id names an ownership directory rather than the transcript itself. `SessionPersistence.locate()` continues to return the fixed transcript path, preserving hook `transcript_path` and `DSH_SESSION_JSONL` semantics. Discovery ignores other entries inside the session directory so the backend can add session-owned artifacts without another layout change. +The encoded session id names an ownership directory rather than the transcript itself. The backend's diagnostics-only `locate` hook resolves the fixed transcript path inside it for format-refusal messages ([export and pre-release trims](../simplification/2026-08-27-persistence-export-and-pre-release-trims.md) owns removing the consumer-facing path query). Discovery ignores other entries inside the session directory so the backend can add session-owned artifacts without another layout change. Lazy materialization remains tied to the transcript: `create()` performs no filesystem I/O, and the first append creates the project/session directories before collision-safe transcript publication. Empty directories are not listed as sessions. The backend rejects flat `/.jsonl*` artifacts with an explicit layout error; the pre-release format provides no automatic data migration. diff --git a/.agents/notes/implemented/architecture/2026-07-24-project-session-directories.zh.md b/.agents/notes/implemented/architecture/2026-07-24-project-session-directories.zh.md index 932b1d29c4..4695677642 100644 --- a/.agents/notes/implemented/architecture/2026-07-24-project-session-directories.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-24-project-session-directories.zh.md @@ -29,7 +29,7 @@ JSONL 后端按可读的项目键存储会话，并为每个会话提供独立 根目录由部署配置决定。这种布局既不选择全局根目录，也不要求项目共享根目录。部署选择集中存储时，目录名仍能让项目路径易于辨认；使用项目本地根目录时，也采用同样的确定性结构。 -编码后的会话 id 用于命名归属目录，而不是 transcript 文件本身。`SessionPersistence.locate()` 仍返回固定的 transcript 路径，从而保持钩子 `transcript_path` 和 `DSH_SESSION_JSONL` 的语义不变。发现过程会忽略会话目录中的其他条目，因此后端以后添加会话自有产物时无需再次改变布局。 +编码后的会话 id 用于命名归属目录，而不是 transcript 文件本身。后端仅供诊断的 `locate` 钩子在其中解析固定的 transcript 路径，供格式拒绝消息使用（移除面向消费者的路径查询由[导出与预发布裁剪](../simplification/2026-08-27-persistence-export-and-pre-release-trims.zh.md)负责）。发现过程会忽略会话目录中的其他条目，因此后端以后添加会话自有产物时无需再次改变布局。 延迟物化仍以 transcript 为界：`create()` 不执行文件系统 I/O，首次追加会先创建项目目录和会话目录，再以无冲突方式发布 transcript。空目录不会被列为会话。后端会显式报告布局错误并拒绝扁平的 `/.jsonl*` 产物；预发布格式不提供自动数据迁移。 diff --git a/.agents/notes/implemented/architecture/2026-07-25-web-client-session-scope-and-provide-channel.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-25-web-client-session-scope-and-provide-channel.i18n.yaml index 05c4325e05..cc913ae925 100644 --- a/.agents/notes/implemented/architecture/2026-07-25-web-client-session-scope-and-provide-channel.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-25-web-client-session-scope-and-provide-channel.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-25-web-client-session-scope-and-provide-channel.md -2026-07-25-web-client-session-scope-and-provide-channel.md: b4566d70c79607bbf736ee02e3e37a79c2391232 -2026-07-25-web-client-session-scope-and-provide-channel.zh.md: 1d1acf00fa6a1efc868c3613715a5ff781e0323a +2026-07-25-web-client-session-scope-and-provide-channel.md: feffb1ac0c5bb91c33e61ea18583202e96bad34d +2026-07-25-web-client-session-scope-and-provide-channel.zh.md: 5d368b4fd810264cb1b451d583de51e1d4cb232b diff --git a/.agents/notes/implemented/architecture/2026-07-25-web-client-session-scope-and-provide-channel.md b/.agents/notes/implemented/architecture/2026-07-25-web-client-session-scope-and-provide-channel.md index b4566d70c7..feffb1ac0c 100644 --- a/.agents/notes/implemented/architecture/2026-07-25-web-client-session-scope-and-provide-channel.md +++ b/.agents/notes/implemented/architecture/2026-07-25-web-client-session-scope-and-provide-channel.md @@ -61,7 +61,7 @@ Session instances share the scope's lifecycle; liveness eligibility = host-liste A session "materialized but with no first prompt" is governed by the summary-derived bit `blank` (a derived column, not a header field; SessionHeader stays immutable): -- The host criterion: `session.events.length === 0` (zero log events = no user message yet). A live session reads `summarize()` straight from memory; a cold session is always `false` — the JSONL provider's lazy-create contract guarantees a never-appended Session never enters `persistence.list()`, so blank never touches disk. +- The host criterion: `session.seq === 0` (zero log events = no user message yet). A live session reads `summarize()` straight from memory; a cold session is always `false` — the JSONL provider's lazy-create contract guarantees a never-appended session never enters `persistence.list()`, so blank never touches disk. - The wire carries it in two places: the required `SessionSummary.blank` column, and the required `blank` field on the `host/session-added` frame (always true at creation, letting other tabs enter the same blank-session state into their mirrors). - The client mirror only lowers, never raises (monotonic), flipped from three sources, all reusing existing wire signals: - The sender's own tab: the **successful response** to the first `prompt()` flips false (acceptance proves the user/message is already in the host log — this flip is confirmation, not optimism; `onEngaged` synchronously updates the list mirror, converting the current `New Session` row in place to an ordinary title, adding no list row). A rejected first prompt keeps the session blank: aligned with host authority, still shown as `New Session`, keeping its connectWorkspace reuse eligibility while it remains a Workspace member. diff --git a/.agents/notes/implemented/architecture/2026-07-25-web-client-session-scope-and-provide-channel.zh.md b/.agents/notes/implemented/architecture/2026-07-25-web-client-session-scope-and-provide-channel.zh.md index 1d1acf00fa..5d368b4fd8 100644 --- a/.agents/notes/implemented/architecture/2026-07-25-web-client-session-scope-and-provide-channel.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-25-web-client-session-scope-and-provide-channel.zh.md @@ -61,7 +61,7 @@ Session 实例与 scope 同生命周期，存活资格 = host listed（一个判 「实体化但无首条提示词」的会话经 summary 派生位 `blank` 治理（派生列而非 header 字段，SessionHeader 保持不可变）： -- host 判据：`session.events.length === 0`（零日志事件 = 尚无用户消息）。live 会话 `summarize()` 内存直读；cold 会话恒 `false`——JSONL provider 的 lazy-create 约定保证 never-appended Session 不进入 `persistence.list()`，所以 blank 从不落盘。 +- host 判据：`session.seq === 0`（零日志事件 = 尚无用户消息）。live 会话 `summarize()` 内存直读；cold 会话恒 `false`——JSONL provider 的 lazy-create 约定保证 never-appended 会话不进入 `persistence.list()`，所以 blank 从不落盘。 - wire 承载两处：`SessionSummary.blank` 必填列；`host/session-added` 帧必填 `blank` 字段（创建时恒 true，供别的 tab 按同一空会话状态入镜像）。 - client 镜像只降不升（单调），三来源翻转，全部复用既有 wire 信号： - 发送方本地：首次 `prompt()` 的**成功响应**翻 false（受理即证明用户消息已入 host 日志——此点翻转是确证而非乐观；`onEngaged` 同步更新列表镜像，当前 `New Session` 行原地转为普通标题，不新增列表行）。首条提示词被拒则会话保持 blank：与 host 权威对齐、继续显示为 `New Session`、在仍为该工作区成员时保持 connectWorkspace 复用资格。 diff --git a/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.i18n.yaml index 3e59e7e702..28ab0da3dd 100644 --- a/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.md -2026-07-25-web-input-machine-and-slash-pipeline.md: 69899efcda42eb1087aaa68d1eba8c08dd14f361 -2026-07-25-web-input-machine-and-slash-pipeline.zh.md: 7e37dd67a2d5a7943a8c601a890d2de7489b227d +2026-07-25-web-input-machine-and-slash-pipeline.md: 200761cc9e648eea80bdae9d7b363246c816e5d1 +2026-07-25-web-input-machine-and-slash-pipeline.zh.md: 673d0ee4bd0916b20ee74226f50240e3904fe06c diff --git a/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.md b/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.md index 69899efcda..200761cc9e 100644 --- a/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.md +++ b/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.md @@ -55,7 +55,7 @@ A trigger/menu/pick pipeline with zero knowledge of "commands": - The hub (trigger/decoration registries + send orchestration) takes the slash/command services as optional `ctx.get()` dependencies: without ui-input-trigger or the command surfaces, input still sends and receives normally — graceful degradation. - Each materialized Session has exactly one `SessionInputShell` (the facade), created and torn down with the session scope; with no session, no input machine is built. `ConversationRoot` is itself the `session-maybe` resident shell, holding HeroShell, the Workspace picker, the composer stack, and the chain-fallback frame. It always owns the same scrollport and composer seat; separate strict-session header and body outlets fill those fixed regions after a Session appears. -- The composer bar is one `session-maybe` slot entry rendered unconditionally: with no session the same InputBar renders inert (machine faces absent, `disabled` owner prop), and once `connectWorkspace` returns a blank session the same instance goes live — the composer surface DOM survives the no-session → blank transition and every later phase flip; `ConversationRoot`, the Hero, and the layout skeleton hold throughout. +- The composer bar is one `session-maybe` slot entry rendered unconditionally: with no session the same InputBar renders inert (machine faces absent, `disabled` owner prop), and once `connectWorkspace` returns a blank session the same instance goes live — the composer surface DOM survives the no-session → blank transition and every later phase flip; `ConversationRoot`, the Hero, and the layout skeleton hold throughout. The memoized InputBar renders its overlay, left, right, and dock child slots after the renderer has bound their standard props; `ConversationRoot` passes only scalar data and callbacks, so an unrelated shell render does not create fresh ReactNode owner props or invalidate the bar. - ConversationRoot's Hero criterion is `sessionId === undefined || (composerPhase === 'blank' && (openState === 'open' || summaryBlank === true))`: a summary-proven blank Session remains Hero in every open state, while an unproven Session settles during loading. The first submit enters engaging synchronously, and a failure keeps the composer and the error context rather than falling back to the blank Hero; the sidebar's blank bit flips false only after a prompt is successfully accepted. - Sending unifies in the hub defaultSink: after an optimistic draft clear it goes only through `session.prompt` with `mode:'queue'` (the Web UI has no steer entry; host-wire `mode:'steer'` remains outside this machine); backfill happens only when it fails and the live draft is still empty — a user who has kept typing is never overwritten. No Draft materialize or attach transaction exists. - When the blank Hero re-picks the Workspace, the shell calls `connectWorkspace`; if the target session differs, the non-empty draft moves from the current shell to the target shell before the new id is opened, and the old blank session survives but is no longer current. @@ -105,6 +105,7 @@ The state machine's entire behavior is covered by pure-JS unit tests (event sequ | Dual draft persistence {text, occurrences} | The mirror writing the clipboard projection adds zero new concepts; chip degradation across refresh is acceptable | | The native textarea undo stack | Unreliable under controlled + programmatic writes; the paste two-step undo semantics can only be self-managed — both sides retired with the textarea itself; Lexical's history owns undo now | | The InputBar receiving a 16-member wiring-callback bundle | The consumption matrix proved 11 members InputBar-exclusive and 1 a dead member; the standard-kit channel lets components fetch their own, with the keyboard surface passed privately in-package | +| `ConversationRoot` rendering InputBar's child slots into owner props | Fresh React elements defeat the bar's memo boundary; the bar already receives `renderSlot` and owns the exact positions | | Space adjudication also claiming execute-kind commands | The misfire defense: after a space the whole line is an ordinary prompt; irreversible side effects keep explicit entry points only | | A generic tokenPattern decoration mechanism | Structured occurrence records replace pattern scanning | | A placeholder select resident in the tool row | Named seats stay empty until registration; a placeholder clashing with the real implementation is two sources of truth | diff --git a/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.zh.md b/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.zh.md index 7e37dd67a2..673d0ee4bd 100644 --- a/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.zh.md @@ -55,7 +55,7 @@ Status: implemented - hub（trigger/decoration 注册表 + 发送编排）对 slash/command 服务是可选 `ctx.get()` 依赖：无 ui-input-trigger/命令面时输入正常收发，优雅降级。 - 每个实体会话只有一个 `SessionInputShell`（facade），随会话作用域创建和拆除；无会话时不造 input machine。`ConversationRoot` 自身是 `session-maybe` 常驻外壳，持有 HeroShell、Workspace picker、composer stack 与 chain fallback 外框。它始终拥有同一个 scrollport 与 composer seat；会话出现后，彼此独立的严格会话 header 和 body outlet 只填入这些固定区域。 -- composer bar 是一个无条件渲染的 `session-maybe` slot entry：无会话时同一个 InputBar 以惰性态渲染（machine face 缺席、`disabled` owner prop），`connectWorkspace` 返回 blank 会话后同一实例转为 live——编辑器表面 DOM 在无会话 → blank 切换及其后每次 phase 翻转中都不重建；`ConversationRoot`、Hero 与布局骨架全程保持。 +- composer bar 是一个无条件渲染的 `session-maybe` slot entry：无会话时同一个 InputBar 以惰性态渲染（machine face 缺席、`disabled` owner prop），`connectWorkspace` 返回 blank 会话后同一实例转为 live——编辑器表面 DOM 在无会话 → blank 切换及其后每次 phase 翻转中都不重建；`ConversationRoot`、Hero 与布局骨架全程保持。memoized InputBar 在 renderer 绑定各 child slot 的标准 props 后自行渲染 overlay、left、right 与 dock；`ConversationRoot` 只传标量数据和回调，因此无关 shell render 不会制造新的 ReactNode owner prop 或使 bar 失效。 - ConversationRoot 的 Hero 判据是 `sessionId === undefined || (composerPhase === 'blank' && (openState === 'open' || summaryBlank === true))`：summary 已证实为空的会话在任何 open state 下都保持 Hero，未经证实的会话则在 loading 期间进入 settling。首次 submit 同步进入 engaging，失败也保留 composer 与错误上下文，不退回 blank Hero；sidebar 的 blank 位只在提示词成功受理后翻 false。 - 发送统一在 hub defaultSink：乐观清稿后只走 `session.prompt` 且固定 `mode:'queue'`（Web UI 无 steer 入口；host 线缆上的 `mode:'steer'` 不经此 machine）；失败且 live draft 仍为空才回填，用户已经继续输入则不覆盖。不存在 Draft materialize 或 attach 事务。 - blank Hero 改选 Workspace 时，外壳调用 `connectWorkspace`；目标会话不同时把非空 draft 从当前 shell 搬到目标 shell，再 open 新 id，旧 blank 会话留存但不再 current。 @@ -105,6 +105,7 @@ skill/@subagent 引用不走占位符 + occurrence 身份链——纯文本引 | draft 双持久化 {text, occurrences} | mirror 写剪贴板投影零新概念；chip 跨刷新降级可接受 | | 原生 textarea undo 栈 | 受控 + 程序化写入下不可靠；粘贴两段 undo 语义只能自管——两侧都随 textarea 一并退役；undo 现归 Lexical history | | InputBar 收 16 员 wiring 回调包 | 消费矩阵实证 11 员 InputBar 独占、1 员死成员；标准件通道让组件自取，键盘面包内私递 | +| 由 `ConversationRoot` 把 InputBar child slot 渲染为 owner prop | 新 React element 会击穿 bar 的 memo 边界；bar 已收到 `renderSlot`，也拥有这些位置 | | 空格裁决也认领即执行型命令 | 误触发防线：空格后整行是普通提示词；不可逆副作用只留显式入口 | | 通用 tokenPattern 装饰机制 | 结构化 occurrence 记录取代模式扫描 | | 占位 select 常驻工具行 | 具名 slot 在注册前保持为空；占位件与真实现冲突时是两个真源 | diff --git a/.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.i18n.yaml index de962f5749..58dcf1ee20 100644 --- a/.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.md -2026-07-26-packed-chunk-rows-by-default.md: 14da6b3cbe650e80118e7c960c96bf618acd1e48 -2026-07-26-packed-chunk-rows-by-default.zh.md: f62a8e52a67adc960ac3150552594b4f061e6205 +2026-07-26-packed-chunk-rows-by-default.md: bd4b3b9f773afbf6aa7e88d51b6e842d6634c222 +2026-07-26-packed-chunk-rows-by-default.zh.md: eafe6632150aadd74395f4d0f09d064fb703a03d diff --git a/.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.md b/.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.md index 14da6b3cbe..bd4b3b9f77 100644 --- a/.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.md +++ b/.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.md @@ -18,7 +18,7 @@ Reading is unconditional and layout-blind. Packed, unpacked, and mixed files loa ### Logical events and physical rows -The JSONL packing path stays at the `dsh-session` storage seam through `packChunkRuns()` and `decodeStorageRecord()`. The encoder recognizes exact delta-event shapes, preserves unrecognized events verbatim, and packs only runs of at least three. A packed row is encoding vocabulary, not a `SessionEventMap` member: it never enters `Session.events` or fires `session/event`. The [packed session-history transport decision](2026-08-15-packed-session-history-transport.md) reuses this vocabulary for a bounded lossless wire interval without changing those event semantics. +The JSONL packing path stays at the `dsh-session` storage seam through `packChunkRuns()` and `decodeStorageRecord()`. The encoder recognizes exact delta-event shapes, preserves unrecognized events verbatim, and packs only runs of at least three. A packed row is encoding vocabulary, not a `SessionEventMap` member: it never enters the Session log or fires `session/event`. The [packed session-history transport decision](2026-08-15-packed-session-history-transport.md) reuses this vocabulary for a bounded lossless wire interval without changing those event semantics. The JSONL backend packs each durable append batch. Raw `compression: 'none'` and default Zstandard framing carry the same logical storage records; selecting raw mode for reviewable fixtures does not disable packing. Repository replay readers and normalizers decode the shared row format instead of maintaining snapshot-specific codecs. diff --git a/.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.zh.md b/.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.zh.md index f62a8e52a6..eafe663215 100644 --- a/.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.zh.md @@ -18,7 +18,7 @@ JSONL 存储 seam 可以在不改变逻辑日志的情况下减少这部分封 ### 逻辑事件与物理行 -JSONL 打包路径保留在 `dsh-session` 的存储 seam，并通过 `packChunkRuns()` 和 `decodeStorageRecord()` 实现。编码器识别精确的增量事件形态，原样保留无法识别的事件，并且只打包至少包含 3 个事件的连续段。打包行属于编码词汇，不是 `SessionEventMap` 成员：它绝不会进入 `Session.events`，也不会触发 `session/event`。[打包会话历史传输决策](2026-08-15-packed-session-history-transport.zh.md)会为有界的无损协议区间复用该词汇，而不改变这些事件语义。 +JSONL 打包路径保留在 `dsh-session` 的存储 seam，并通过 `packChunkRuns()` 和 `decodeStorageRecord()` 实现。编码器识别精确的增量事件形态，原样保留无法识别的事件，并且只打包至少包含 3 个事件的连续段。打包行属于编码词汇，不是 `SessionEventMap` 成员：它绝不会进入 Session 日志，也不会触发 `session/event`。[打包会话历史传输决策](2026-08-15-packed-session-history-transport.zh.md)会为有界的无损协议区间复用该词汇，而不改变这些事件语义。 JSONL 后端会打包每个持久追加批次。原始模式 `compression: 'none'` 与默认 Zstandard 帧承载相同的逻辑存储记录；为使 fixture 便于评审而选择原始模式，不会禁用打包。仓库中的回放读取器和规范化器会解码共享行格式，而不维护快照专用编解码器。 diff --git a/.agents/notes/implemented/architecture/2026-07-29-projected-token-usage-and-request-context.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-29-projected-token-usage-and-request-context.i18n.yaml index 8baf8386d1..2e41d77dd7 100644 --- a/.agents/notes/implemented/architecture/2026-07-29-projected-token-usage-and-request-context.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-29-projected-token-usage-and-request-context.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-29-projected-token-usage-and-request-context.md -2026-07-29-projected-token-usage-and-request-context.md: 75a05e5a0e8f0183fef1e7d80701ce6d81041cd6 -2026-07-29-projected-token-usage-and-request-context.zh.md: 7cce5989d719156f1d66c48937780ff8aed02a42 +2026-07-29-projected-token-usage-and-request-context.md: 7c984012b7a4b387f1ff24279567fe15aa34cf77 +2026-07-29-projected-token-usage-and-request-context.zh.md: b34fb7702898f4a51524a24ad27927c48220bfad diff --git a/.agents/notes/implemented/architecture/2026-07-29-projected-token-usage-and-request-context.md b/.agents/notes/implemented/architecture/2026-07-29-projected-token-usage-and-request-context.md index 75a05e5a0e..7c984012b7 100644 --- a/.agents/notes/implemented/architecture/2026-07-29-projected-token-usage-and-request-context.md +++ b/.agents/notes/implemented/architecture/2026-07-29-projected-token-usage-and-request-context.md @@ -26,7 +26,7 @@ Capacity deliberately stays out of `EpochHeader`. That type is the reconstructio Both units ride the standard projection lifecycle: history tail baselines, `session/projection` live frames, higher-seq-wins client storage, JSON checkpoints, cache recovery, and unit unload. There is no token-specific history field, mux frame, projector, revision counter, or client fence. -The Web `StatsLine` reads both through the standard `useProjection` seat. Window nodes still supply turn and step counts plus LLM and tool wall times — those answer "what is on screen" and are correctly window-scoped. Durable token and context groups remain when compaction leaves no visible assistant step. Cache writes count in billed input and in the cache-hit denominator. A deployment without token-meter drops the token groups; occupancy stays hidden until both pressure and capacity are known. +The Web `StatsLine` reads both through the standard `useProjection` seat. Window nodes still supply turn and step counts plus LLM and tool wall times — those answer "what is on screen" and are correctly window-scoped. Durable token and context groups remain when compaction leaves no visible assistant step. Cache writes count in billed input and in the cache-hit denominator. A deployment without token-meter drops the token groups; occupancy stays hidden until both pressure and capacity are known. The exact-overflow tooltip mounts its measuring child only for a non-empty line and retains one `ResizeObserver` while values change; text changes perform one direct measurement without replacing the observer. ## Context occupancy is approximate, and that is the decision @@ -58,4 +58,4 @@ Token totals stay stable across pagination, compaction, replay, restart, and rec Occupancy is approximate in the ways documented above. It is available immediately after restore or reconnect, since both fields are durable, at the cost of describing the last recorded request rather than an exact current boundary. -Each session log gains one small `request/context` record per route or advertised-capacity change. Token-meter is the canonical owner of durable usage semantics, including retry-attempt separation in the cumulative projection and the reusable exact attempt/Turn fold; Web Chat only selects a complete loaded Turn and renders the fold result. The TUI retains its live per-step map because it does not mount the generic projection seam, and the standalone browser fixture mirrors the unit. Connection and API Gateway carry no token-specific code, own no per-session metrics cache, and perform no measurement. The browser keeps two generic projection values and no connection-local telemetry, and streaming text deltas still do not force the stats line to recompute. +Each session log gains one small `request/context` record per route or advertised-capacity change. Token-meter is the canonical owner of durable usage semantics, including retry-attempt separation in the cumulative projection and the reusable exact attempt/Turn fold; Web Chat only selects a complete loaded Turn and renders the fold result. The TUI retains its live per-step map because it does not mount the generic projection seam, and the standalone browser fixture mirrors the unit. Connection and API Gateway carry no token-specific code, own no per-session metrics cache, and perform no measurement. The browser keeps two generic projection values and no connection-local telemetry; streaming text deltas do not force the stats line to recompute or churn layout-observer subscriptions. diff --git a/.agents/notes/implemented/architecture/2026-07-29-projected-token-usage-and-request-context.zh.md b/.agents/notes/implemented/architecture/2026-07-29-projected-token-usage-and-request-context.zh.md index 7cce5989d7..b34fb77028 100644 --- a/.agents/notes/implemented/architecture/2026-07-29-projected-token-usage-and-request-context.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-29-projected-token-usage-and-request-context.zh.md @@ -26,7 +26,7 @@ token-meter 还拥有在持久事件上运行的共享纯 attempt／Turn fold。 两个单元都沿用标准投影生命周期：历史尾页基线、`session/projection` 实时帧、seq 高者胜的客户端存储、JSON 检查点、缓存恢复和单元卸载。系统没有任何 token 专用的历史字段、mux 帧、投影器、修订计数器或客户端栅栏。 -Web `StatsLine` 通过标准 `useProjection` 席位读取两者。窗口内节点仍提供轮次和步骤计数，以及 LLM（大语言模型）与工具的墙钟时间：它们回答的是「屏幕上有什么」，按窗口作用域正是正确的。压缩使可见 assistant 步骤归零后，持久 token 与上下文分组仍会保留。缓存写入会计入计费输入和缓存命中率分母。未部署 token-meter 时会去掉 token 分组；只有压力与容量都已知时才显示占用率。 +Web `StatsLine` 通过标准 `useProjection` 席位读取两者。窗口内节点仍提供轮次和步骤计数，以及 LLM（大语言模型）与工具的墙钟时间：它们回答的是「屏幕上有什么」，按窗口作用域正是正确的。压缩使可见 assistant 步骤归零后，持久 token 与上下文分组仍会保留。缓存写入会计入计费输入和缓存命中率分母。未部署 token-meter 时会去掉 token 分组；只有压力与容量都已知时才显示占用率。精确 overflow tooltip 只在统计行非空时挂载测量子组件，并在值变化期间保留同一个 `ResizeObserver`；文本变化只直接测量一次，不替换 observer。 ## 上下文占用率是近似值，而这正是决策本身 @@ -58,4 +58,4 @@ token 总量在分页、压缩、回放、重启和重连期间保持稳定， 占用率在上文记录的意义上是近似值。由于两个字段都是持久的，它在恢复或重连后立即可用；代价是它描述的是最后一条已记录的请求，而不是精确的当前边界。 -每个会话日志会为每次路由或已公布容量变化增加一条小型 `request/context` 记录。token-meter 是持久用量语义的正典所有方，包括累计投影中的重试 attempt 分离，以及可复用的精确 attempt／Turn fold；Web Chat 只选择已完整加载的 Turn 并渲染 fold 结果。TUI 未挂载通用投影 seam，因此保留自己的实时逐步骤 map，而独立浏览器 fixture（测试前置数据）会镜像该单元。Connection 与 API Gateway 不携带任何 token 专用代码，不拥有逐会话指标缓存，也不执行测量。浏览器只保留两个通用投影值，不保留连接本地的遥测数据；流式文本增量仍不会迫使统计行重新计算。 +每个会话日志会为每次路由或已公布容量变化增加一条小型 `request/context` 记录。token-meter 是持久用量语义的正典所有方，包括累计投影中的重试 attempt 分离，以及可复用的精确 attempt／Turn fold；Web Chat 只选择已完整加载的 Turn 并渲染 fold 结果。TUI 未挂载通用投影 seam，因此保留自己的实时逐步骤 map，而独立浏览器 fixture（测试前置数据）会镜像该单元。Connection 与 API Gateway 不携带任何 token 专用代码，不拥有逐会话指标缓存，也不执行测量。浏览器只保留两个通用投影值，不保留连接本地的遥测数据；流式文本增量不会迫使统计行重新计算或反复替换布局 observer 订阅。 diff --git a/.agents/notes/implemented/architecture/2026-07-30-session-end-seed-log-boundary.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-30-session-end-seed-log-boundary.i18n.yaml index a995ac778f..b91733813c 100644 --- a/.agents/notes/implemented/architecture/2026-07-30-session-end-seed-log-boundary.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-30-session-end-seed-log-boundary.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-30-session-end-seed-log-boundary.md -2026-07-30-session-end-seed-log-boundary.md: 1c5a8097a6b901f133205dcd52d674a8e3594b28 -2026-07-30-session-end-seed-log-boundary.zh.md: ea3549543229a15d0fba7ad0316a8683c674a558 +2026-07-30-session-end-seed-log-boundary.md: c6ed3911a797480804d064273922d85412664c79 +2026-07-30-session-end-seed-log-boundary.zh.md: 1e9517f9a5aed819fdaff6194ab952c322c85b82 diff --git a/.agents/notes/implemented/architecture/2026-07-30-session-end-seed-log-boundary.md b/.agents/notes/implemented/architecture/2026-07-30-session-end-seed-log-boundary.md index 1c5a8097a6..c6ed3911a7 100644 --- a/.agents/notes/implemented/architecture/2026-07-30-session-end-seed-log-boundary.md +++ b/.agents/notes/implemented/architecture/2026-07-30-session-end-seed-log-boundary.md @@ -40,7 +40,7 @@ The predicate holds for a bracket *this* session inherited, not as a liveness si **A boundary appended at loop start.** The loop calls `resumeWith`, so it covers the resume paths, but it misses `fork()` and `adopt()` entirely, and the event would have to fire on `'startup'` — the source a fork child publishes — so `SessionStartSource` would stop discriminating. It also publishes the session before the marker is appended, so a `session/created` listener could observe a seeded log with no boundary. -**Reusing `header.seedLength`.** It is the durable *fork-lineage* boundary and deliberately keeps the original fork value across a resume, where the constructor seed is the whole stored log. The two facts differ and conflating them would lose both. +**Reusing `Session.inheritedEventCount`.** It is the durable *fork-lineage* cut and deliberately keeps the original fork value across a resume, where the constructor seed is the whole stored log. The two facts differ and conflating them would lose both. **Crash repair closing `compaction/*` alongside turn boundaries.** Rejected: it moves every plugin's bracket semantics into core's repair pass, and core cannot know what closing another package's bracket should record. diff --git a/.agents/notes/implemented/architecture/2026-07-30-session-end-seed-log-boundary.zh.md b/.agents/notes/implemented/architecture/2026-07-30-session-end-seed-log-boundary.zh.md index ea35495432..1e9517f9a5 100644 --- a/.agents/notes/implemented/architecture/2026-07-30-session-end-seed-log-boundary.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-30-session-end-seed-log-boundary.zh.md @@ -40,7 +40,7 @@ Status: implemented **在 loop 启动时追加边界。** loop 调用 `resumeWith`，因此覆盖恢复路径，但完全漏掉 `fork()` 与 `adopt()`，而且事件不得不在 `'startup'` 上触发——那是 fork 子会话发布的来源——于是 `SessionStartSource` 将不再具有区分力。它还会在追加标记之前就发布会话，因此 `session/created` 监听方可能观察到一份没有边界的带种子日志。 -**复用 `header.seedLength`。** 它是持久的 *fork 血缘*边界，并且刻意在恢复时保留原始 fork 取值——而恢复时构造种子是整份存储日志。这两个事实并不相同，混同会同时失去两者。 +**复用 `Session.inheritedEventCount`。** 它是持久的 *fork 血缘* cut，并且刻意在恢复时保留原始 fork 取值——而恢复时构造种子是整份存储日志。这两个事实并不相同，混同会同时失去两者。 **让崩溃修复连同轮次边界一起关闭 `compaction/*`。** 否决：这会把每个插件的括号语义搬进核心的修复流程，而核心无法知道关闭另一个包的括号应该记录什么。 diff --git a/.agents/notes/implemented/architecture/2026-07-30-web-config-plane.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-30-web-config-plane.i18n.yaml index 16238db465..33d4ec3daa 100644 --- a/.agents/notes/implemented/architecture/2026-07-30-web-config-plane.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-30-web-config-plane.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-30-web-config-plane.md -2026-07-30-web-config-plane.md: 81b501db529bf1b2974fd4541045991c5a8bf087 -2026-07-30-web-config-plane.zh.md: 3f02a17e4826bb35ecfd45da25c4b0170be270cb +2026-07-30-web-config-plane.md: a919487ba48cd7735a9f7fbc65a548bc5bfb7114 +2026-07-30-web-config-plane.zh.md: ca9e9f4427bba80a63865e891e42656aeb1b5c64 diff --git a/.agents/notes/implemented/architecture/2026-07-30-web-config-plane.md b/.agents/notes/implemented/architecture/2026-07-30-web-config-plane.md index 81b501db52..a919487ba4 100644 --- a/.agents/notes/implemented/architecture/2026-07-30-web-config-plane.md +++ b/.agents/notes/implemented/architecture/2026-07-30-web-config-plane.md @@ -22,7 +22,7 @@ The request-level configuration seam made LLM adapter configuration restart-free **A hand-written editor over a schema model layer.** `ctx.settingsSchema`, provided by `dsh-client-ui-settings`, rehydrates the wire's `toJSON()` envelope into live schemastery nodes for validation, path resolution, and immutable draft editing — but no generic rendering: the first cut shipped a full schema-driven form renderer, and the resulting page was an unstyled schema dump (every advanced field flattened onto the card, raw field names as labels, the `retryPolicy` unsupported-fallback in the main flow). The hand-written direction won over adding a hint/grouping system, and a further simplification removed the reference input entirely: the card's primary field is one **API key** input, a whole-section provider without a configured key opens as its setup card, and the collapsed 自定义设置 fold carries the curated per-family extras (`baseURL` for both families, `reasoningEffort` for deepseek / `reasoning` for pi-ai, plus direct DeepSeek model rows with `id`, `name`, and `contextWindow`). Existing model fields outside that visible set survive array edits; retry policy, timeouts, and other fields remain owned by `settings.yaml`. Validation still runs the rehydrated schema before writing, while adapter-specific checks reject catalog invariants that the serialized schema cannot express. The card's colors resolve through the `--dsw-alias-*` design tokens; it had named `--border`/`--surface`/`--text-*`, which nothing in this app defines, so it rendered their light-mode fallbacks and stayed light under the dark theme. The model catalog takes the row shape the pi-ai provider form introduces: one bordered entry per model, id and display name on the row, and the capacities behind the row's own disclosure, so the two editors read as one design rather than diverging. Every field keeps the indexed `aria-label` that names it. Both capacities are text fields reading a decimal `K`/`M` suffix (`1M` is 1000K, matching how capacities are quoted) and storing the plain count: a field holds the typed text while it has focus, because re-deriving it from the parsed count on every keystroke would rewrite `1000` to `1K` mid-word, and text that does not parse stays on screen so the save-time rejection names a row the user can still see. The shared class names carry only declared token spellings: `--dsw-alias-border-subtle`, `--dsw-alias-text-tertiary`, and `--dsw-alias-text-primary` are undeclared, so naming them resolves to the light-mode literals in their fallback slots. A styles test now rejects any `--dsw-*` name the token sheet does not declare, so the next editor to name one fails rather than shipping a light-only surface. -**The Models page is a three-domain join with service-shaped apply semantics.** Rows are configured providers; the add card's select is the dormant directory remainder. Route liveness still gates readiness and invalidates the join, but the page does not render it as provider status because configuration presence and runtime availability are distinct. The key path stays reference-shaped without ever showing a reference: a typed key stores **write-only** through `credentials.set` under the profile's `apiKeyEnv`, deriving `_API_KEY` when none exists (the pi-ai profile records the derivation only when a key is entered), so `settings.yaml` never carries a key value; a blank pi-ai key materializes a reference-free profile and preserves provider-native authentication. Profile edits and removals land as minimal path-addressed `settings.mutate` operations against the redacted user section, which never names a secret the page did not receive. Removing a user-layer provider first opens a localized confirmation dialog whose row actions, title, description, and final action identify the same provider; confirmation removes an exact configured+writable derived credential before the profile, while custom, environment, and unidentified targets remain untouched. Both stages are idempotent and a partial failure stays in the dialog for retry. DeepSeek's model list is array-replace configuration: inherited effective rows remain visible until the first edit materializes the complete list in the user layer, and reset unsets the list override. `llm.discoverModels` results stay in picker-local state until **Add selected**; configured ids start unchecked, while **Select all** / **Deselect all** changes only that local set, so bulk selection preserves the same capacity-protection rule. The partial-commit and credential-ownership rationale lives in the [provider credential lifecycle note](../bug-fix/2026-08-06-provider-credential-lifecycle.md). +**The Models page is a three-domain join with service-shaped apply semantics.** Rows are configured providers; the add card's select is the dormant directory remainder. Route liveness still gates readiness and invalidates the join, but the page does not render it as provider status because configuration presence and runtime availability are distinct. The key path stays reference-shaped without ever showing a reference: a typed key stores **write-only** through `credentials.set` under the profile's `apiKeyEnv`, deriving `_API_KEY` when none exists (the pi-ai profile records the derivation only when a key is entered), so `settings.yaml` never carries a key value; a blank pi-ai key materializes a reference-free profile and preserves provider-native authentication. Profile edits and removals land as minimal path-addressed `settings.mutate` operations against the redacted user section, which never names a secret the page did not receive. Removing a user-layer provider first opens a localized confirmation dialog whose row actions, title, description, and final action identify the same provider; confirmation removes an exact configured+writable derived credential before the profile, while custom, environment, and unidentified targets remain untouched. Both stages are idempotent and a partial failure stays in the dialog for retry. DeepSeek's model list is array-replace configuration: inherited effective rows remain visible until the first edit materializes the complete list in the user layer, and reset unsets the list override. `llm.discoverModels` results stay in picker-local state until **Add selected**; configured ids start unchecked, and a localized search filters ids and optional display names without changing hidden selections. **Select all** / **Deselect all** changes only the visible candidates in that local set, so filtered bulk selection preserves the same capacity-protection rule. The partial-commit and credential-ownership rationale lives in the [provider credential lifecycle note](../bug-fix/2026-08-06-provider-credential-lifecycle.md). ## Alternatives considered @@ -36,4 +36,4 @@ The request-level configuration seam made LLM adapter configuration restart-free ## Consequences -The whole loop is pinned keyless in the browser lane (`apps/web/tests/models-settings.e2e.ts`): the add card offers the dormant pi-ai catalog, adding `minimax-cn` with a typed key writes the reference-only profile into `settings.yaml`, stores the value into the harness home's `.env` under the derived `MINIMAX_CN_API_KEY`, registers the route live on the topology frame, and the customized fold merges `reasoning` beside the reference — zero model calls, ARIA goldens for the add-card, configured, model-picker, and identified delete-confirmation states, plus a scaffold `harnessHome` so tests never touch a real `~/.dsh` (the provider under test is one whose derived reference cannot collide with a developer's exported keys). The component suite pins configured-id exclusions and both directions of the bulk toggle. The settings-shell scenario intercepts the pathless native intent; Service Definition, provider, wire, React, and native-opener tests separately pin provider absence, custom-path resolution, absent-file materialization, owner-only permissions, hidden remote/unavailable states, duplicate-click collapse, localized failure, macOS text-editor dispatch, and Linux/Windows desktop dispatch. The removal scenario proves cancellation leaves both profile and key intact, then confirmation removes both the profile and its identified managed credential. The DeepSeek onboarding fixture edits the default catalog into a user-owned list, persists an arbitrary model id/name/context window, removes the active row, and observes the model selector's empty-selection fallback. The rename touched 239 files (fixtures, goldens, docs, python) with no compatibility alias. The renderer replacement needed no wire change: apply semantics, redaction, and the directory join were renderer-agnostic all along. Deferred: a per-row models preview (the picker already lists models) and a page address for live routes that never declared configurability. +The whole loop is pinned keyless in the browser lane (`apps/web/tests/models-settings.e2e.ts`): the add card offers the dormant pi-ai catalog, adding `minimax-cn` with a typed key writes the reference-only profile into `settings.yaml`, stores the value into the harness home's `.env` under the derived `MINIMAX_CN_API_KEY`, registers the route live on the topology frame, and the customized fold merges `reasoning` beside the reference — zero model calls, ARIA goldens for the add-card, configured, model-picker, and identified delete-confirmation states, plus a scaffold `harnessHome` so tests never touch a real `~/.dsh` (the provider under test is one whose derived reference cannot collide with a developer's exported keys). The model-picker path filters the catalog, clears only the visible selection, restores hidden picks when the query clears, and records the localized search control in its ARIA golden; the component suite also pins matching by id and optional name, the no-results state, and visible-only bulk selection. The settings-shell scenario intercepts the pathless native intent; Service Definition, provider, wire, React, and native-opener tests separately pin provider absence, custom-path resolution, absent-file materialization, owner-only permissions, hidden remote/unavailable states, duplicate-click collapse, localized failure, macOS text-editor dispatch, and Linux/Windows desktop dispatch. The removal scenario proves cancellation leaves both profile and key intact, then confirmation removes both the profile and its identified managed credential. The DeepSeek onboarding fixture edits the default catalog into a user-owned list, persists an arbitrary model id/name/context window, removes the active row, and observes the model selector's empty-selection fallback. The rename touched 239 files (fixtures, goldens, docs, python) with no compatibility alias. The renderer replacement needed no wire change: apply semantics, redaction, and the directory join were renderer-agnostic all along. Deferred: a per-row models preview (the picker already lists models) and a page address for live routes that never declared configurability. diff --git a/.agents/notes/implemented/architecture/2026-07-30-web-config-plane.zh.md b/.agents/notes/implemented/architecture/2026-07-30-web-config-plane.zh.md index 3f02a17e48..ca9e9f4427 100644 --- a/.agents/notes/implemented/architecture/2026-07-30-web-config-plane.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-30-web-config-plane.zh.md @@ -22,7 +22,7 @@ Status: implemented **架在 schema 模型层之上的手写编辑器。**`dsh-client-ui-settings` 提供的 `ctx.settingsSchema` 把 wire 的 `toJSON()` 信封还原（rehydrate）为活的 schemastery 节点，用于校验、路径解析与不可变草稿编辑——但不做通用渲染：第一版交付了完整的 schema 驱动表单渲染器，得到的却是一个未加样式、把 schema 原样倾倒出来的页面（每个进阶字段都平铺到卡片上、原始字段名直接充当标签、`retryPolicy` 的「不支持」回退落在主流程里）。手写方向胜过了再加一套提示／分组系统，进一步的简化又把引用输入框整个移除：卡片的主字段是一个 **API 密钥** 输入框，未配置密钥的整分节提供方会以其设置卡片的形式打开，收起的「自定义设置」折叠区承载按家族精选的额外字段（两个家族都有 `baseURL`，deepseek 有 `reasoningEffort`／pi-ai 有 `reasoning`，另有直接 DeepSeek 模型行的 `id`、`name` 和 `contextWindow`）。现有模型字段中不在可见集合内的部分会在数组编辑后保留；重试策略、超时及其他字段仍归 `settings.yaml` 所有。校验仍会在写入前运行还原出的 schema，适配器特有的检查则会拒绝序列化 schema 无法表达的目录不变量。卡片的颜色经 `--dsw-alias-*` 设计 token 解析；它此前引用的 `--border`／`--surface`／`--text-*` 在本应用中无人定义，于是渲染出的是它们的亮色模式回退值，在暗色主题下依旧保持亮色。模型目录采用 pi-ai 提供方表单引入的行形态：每个模型一个带边框的条目，ID 与显示名称落在行上，容量则收在该行自己的折叠区里，使两个编辑器呈现为同一套设计，而不是各自分岔。每个字段都保留那个为其命名的带序号 `aria-label`。两项容量都是文本输入框，读取十进制的 `K`／`M` 后缀（`1M` 即 1000K，与容量的通行标注方式一致）并存储纯数值：字段持有焦点期间保留键入的文本，因为若每次按键都从解析出的数值重新推导该文本，`1000` 会在尚未输完时就被改写成 `1K`；无法解析的文本也会留在屏幕上，因此保存时的拒绝点名的是用户仍能看见的那一行。共用的类名只承载已声明的 token 写法：`--dsw-alias-border-subtle`、`--dsw-alias-text-tertiary` 和 `--dsw-alias-text-primary` 均未声明，写出它们就会解析为各自回退槽位中的亮色模式字面值。现在有一个样式测试会拒绝 token 表未声明的任何 `--dsw-*` 名称，因此下一个写出这类名称的编辑者会当场失败，而不是交付一个只有亮色的界面。 -**Models 页是一次三领域联接，应用语义与服务同形。**每一行是一个已配置的提供方；「新增」卡片的选择框是可配置提供方目录中剩余的休眠条目。路由存活状态仍用于就绪判定，并会使该联接失效，但页面不将其渲染为提供方状态，因为配置存在与运行时可用性是两个不同概念。密钥通道保持引用形态，却从不展示任何引用：键入的密钥经 `credentials.set` **只写**存入 profile 的 `apiKeyEnv` 之下，引用不存在时便派生 `_API_KEY`（仅在输入密钥时，pi-ai profile 才会记录该派生），因此 `settings.yaml` 从不携带密钥值；留空 pi-ai 密钥会具化一个不带引用的 profile，并保留提供方原生认证。profile 的编辑和删除会针对脱敏后的用户分节，以按路径寻址的最小 `settings.mutate` 操作落地，绝不会点名页面未收到的机密。删除用户层提供方时，会先打开本地化确认对话框，其行操作、标题、说明和最终操作都会点名同一个提供方；确认后会先清除与派生目标精确匹配且已配置、可写的凭据，再删除 profile，自定义目标、环境目标和无法识别的目标则保持不变。两个阶段都具备幂等性，部分失败会留在对话框中供重试。DeepSeek 的模型列表是数组替换配置：继承而来的生效模型行会一直显示，直到第一次编辑将完整列表具化到用户层；重置则会取消设置该列表覆盖。`llm.discoverModels` 的结果会留在选择框的本地状态中，直至点击**添加所选**；已配置的 ID 默认不勾选，**全选**／**取消全选**也只会改变这一本地集合，因此批量选择仍遵守同一条容量保护规则。部分提交与凭据所有权的理由记录在[提供方凭据生命周期 note](../bug-fix/2026-08-06-provider-credential-lifecycle.zh.md)中。 +**Models 页是一次三领域联接，应用语义与服务同形。**每一行是一个已配置的提供方；「新增」卡片的选择框是可配置提供方目录中剩余的休眠条目。路由存活状态仍用于就绪判定，并会使该联接失效，但页面不将其渲染为提供方状态，因为配置存在与运行时可用性是两个不同概念。密钥通道保持引用形态，却从不展示任何引用：键入的密钥经 `credentials.set` **只写**存入 profile 的 `apiKeyEnv` 之下，引用不存在时便派生 `_API_KEY`（仅在输入密钥时，pi-ai profile 才会记录该派生），因此 `settings.yaml` 从不携带密钥值；留空 pi-ai 密钥会具化一个不带引用的 profile，并保留提供方原生认证。profile 的编辑和删除会针对脱敏后的用户分节，以按路径寻址的最小 `settings.mutate` 操作落地，绝不会点名页面未收到的机密。删除用户层提供方时，会先打开本地化确认对话框，其行操作、标题、说明和最终操作都会点名同一个提供方；确认后会先清除与派生目标精确匹配且已配置、可写的凭据，再删除 profile，自定义目标、环境目标和无法识别的目标则保持不变。两个阶段都具备幂等性，部分失败会留在对话框中供重试。DeepSeek 的模型列表是数组替换配置：继承而来的生效模型行会一直显示，直到第一次编辑将完整列表具化到用户层；重置则会取消设置该列表覆盖。`llm.discoverModels` 的结果会留在选择框的本地状态中，直至点击**添加所选**；已配置的 ID 默认不勾选，本地化搜索会过滤 id 和可选显示名称而不改变隐藏项的勾选状态。**全选**／**取消全选**只会改变本地集合中的可见候选，因此经过过滤的批量选择仍遵守同一条容量保护规则。部分提交与凭据所有权的理由记录在[提供方凭据生命周期 note](../bug-fix/2026-08-06-provider-credential-lifecycle.zh.md)中。 ## 曾考虑的替代方案 @@ -36,4 +36,4 @@ Status: implemented ## 后果 -整条闭环以无密钥方式固定在浏览器测试通道（`apps/web/tests/models-settings.e2e.ts`）：「新增」卡片提供休眠的 pi-ai catalog，携键入的密钥添加 `minimax-cn` 会把只含引用的 profile 写入 `settings.yaml`、把密钥值存入 harness 家目录 `.env` 中派生的 `MINIMAX_CN_API_KEY` 之下、路由随拓扑帧注册为存活，「自定义设置」折叠区则把 `reasoning` 合并到引用旁边——全程零模型调用，「新增」卡片态、已配置态、模型选择框态与已点名目标的删除确认态各有 ARIA golden，另有脚手架式的 `harnessHome`，测试绝不触碰真实的 `~/.dsh`（受测提供方是派生引用不可能与开发者已导出密钥相撞的那一个）。组件测试固定了已配置 ID 的排除与批量切换的两个方向。设置外壳场景会截获无路径参数的原生意图；Service Definition、提供方、wire、React 与原生打开器测试分别固定了提供方缺失、自定义路径解析、缺失文件创建、仅属主权限、远程／不可用时隐藏、重复点击合并、本地化失败、macOS 文本编辑器分发，以及 Linux/Windows 桌面分发。删除场景证明，取消会保留 profile 和密钥，随后的确认会同时删除 profile 及其已识别的受管凭据。DeepSeek 首次使用 fixture 会把默认目录编辑为用户自有列表、持久化任意模型的 ID／名称／上下文窗口、移除活动模型行，并观察模型选择器的空选择回退。这次重命名触及 239 个文件（fixture（测试前置数据）、golden、文档、python），未保留兼容别名。替换渲染器不需要任何 wire 变更：应用语义、脱敏与目录联接从一开始就与渲染器无关。延后事项：每行的模型预览（选择器已能列出模型）和为从未声明可配置性的存活路由提供页面地址。 +整条闭环以无密钥方式固定在浏览器测试通道（`apps/web/tests/models-settings.e2e.ts`）：「新增」卡片提供休眠的 pi-ai catalog，携键入的密钥添加 `minimax-cn` 会把只含引用的 profile 写入 `settings.yaml`、把密钥值存入 harness 家目录 `.env` 中派生的 `MINIMAX_CN_API_KEY` 之下、路由随拓扑帧注册为存活，「自定义设置」折叠区则把 `reasoning` 合并到引用旁边——全程零模型调用，「新增」卡片态、已配置态、模型选择框态与已点名目标的删除确认态各有 ARIA golden，另有脚手架式的 `harnessHome`，测试绝不触碰真实的 `~/.dsh`（受测提供方是派生引用不可能与开发者已导出密钥相撞的那一个）。模型选择器路径会过滤 catalog、只清除可见项的选择、在查询清空后恢复隐藏勾选，并在 ARIA golden 中记录本地化搜索控件；组件测试还固定了按 id 和可选名称匹配、无结果状态以及仅作用于可见项的批量选择。设置外壳场景会截获无路径参数的原生意图；Service Definition、提供方、wire、React 与原生打开器测试分别固定了提供方缺失、自定义路径解析、缺失文件创建、仅属主权限、远程／不可用时隐藏、重复点击合并、本地化失败、macOS 文本编辑器分发，以及 Linux/Windows 桌面分发。删除场景证明，取消会保留 profile 和密钥，随后的确认会同时删除 profile 及其已识别的受管凭据。DeepSeek 首次使用 fixture 会把默认目录编辑为用户自有列表、持久化任意模型的 ID／名称／上下文窗口、移除活动模型行，并观察模型选择器的空选择回退。这次重命名触及 239 个文件（fixture（测试前置数据）、golden、文档、python），未保留兼容别名。替换渲染器不需要任何 wire 变更：应用语义、脱敏与目录联接从一开始就与渲染器无关。延后事项：每行的模型预览（选择器已能列出模型）和为从未声明可配置性的存活路由提供页面地址。 diff --git a/.agents/notes/implemented/architecture/2026-07-31-code-runtime-portable-identifier-seam.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-31-code-runtime-portable-identifier-seam.i18n.yaml index dd6e455025..d40897535b 100644 --- a/.agents/notes/implemented/architecture/2026-07-31-code-runtime-portable-identifier-seam.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-31-code-runtime-portable-identifier-seam.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-31-code-runtime-portable-identifier-seam.md -2026-07-31-code-runtime-portable-identifier-seam.md: 2011b0f6bc8209e628227ddf486aa1143a63688a -2026-07-31-code-runtime-portable-identifier-seam.zh.md: 36af33366d004fedc6b1077a937d6519de743638 +2026-07-31-code-runtime-portable-identifier-seam.md: e4cf236f62407c9fda42a3e2cdcc5d3ef02a1f92 +2026-07-31-code-runtime-portable-identifier-seam.zh.md: 63fc89eb0d674a381ce7a5a626bd51d5f8b234d3 diff --git a/.agents/notes/implemented/architecture/2026-07-31-code-runtime-portable-identifier-seam.md b/.agents/notes/implemented/architecture/2026-07-31-code-runtime-portable-identifier-seam.md index 2011b0f6bc..e4cf236f62 100644 --- a/.agents/notes/implemented/architecture/2026-07-31-code-runtime-portable-identifier-seam.md +++ b/.agents/notes/implemented/architecture/2026-07-31-code-runtime-portable-identifier-seam.md @@ -25,7 +25,7 @@ The constants live in the Service Definition even though the worker is the only ## Scope -This decision delivers only the Service Definition extension and the worker's adoption of it. The `py-types` renderer and PTC mode language dispatch are owned by the [language-dispatch note](../feature/2026-07-31-ptc-language-dispatch.md); a Python backend does not exist yet. The Service Definition README keeps its worker-only wording for that reason: linking to a `dsh-code-runtime-python` README that does not exist would break the dead-link gate. +This decision delivers the Service Definition extension and the worker-thread backend's adoption of it. The `py-types` renderer and PTC mode language dispatch are owned by the [language-dispatch note](../feature/2026-07-31-ptc-language-dispatch.md). The private experimental CPython subprocess backend (`dsh-experimental-code-runtime-python`) adopts the same portable-identifier contract. `RESERVED_BINDING_GLOBALS` encodes the Python bootstrap's concrete design ahead of the backend itself: it seeds exactly `__builtins__`/`__name__` and wraps the program under `__dsh_main__`. A Python backend that seeds any additional module global (`__doc__`, `__loader__`, `__spec__`, `__file__`, `__package__`, …) MUST widen this set in the same change, exactly as adding a language widens `PORTABLE_RESERVED_WORDS` — a name the bootstrap seeds but the set omits is the portability split this contract exists to prevent. diff --git a/.agents/notes/implemented/architecture/2026-07-31-code-runtime-portable-identifier-seam.zh.md b/.agents/notes/implemented/architecture/2026-07-31-code-runtime-portable-identifier-seam.zh.md index 36af33366d..63fc89eb0d 100644 --- a/.agents/notes/implemented/architecture/2026-07-31-code-runtime-portable-identifier-seam.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-31-code-runtime-portable-identifier-seam.zh.md @@ -25,7 +25,7 @@ Service Definition 同时把可移植标识符子集收窄为 `[A-Za-z_][A-Za-z0 ## Scope -本决策只交付 Service Definition 扩展与 worker 对它的采用。`py-types` 渲染器与 PTC mode 的语言分发归[语言分发 note](../feature/2026-07-31-ptc-language-dispatch.zh.md) 所有；Python 后端尚不存在。Service Definition README 因此保留仅描述 worker 的措辞：链接到一个不存在的 `dsh-code-runtime-python` README 会破坏死链 gate。 +本决策交付 Service Definition 扩展与 worker-thread 后端对它的采用。`py-types` 渲染器与 PTC mode 的语言分发归[语言分发 note](../feature/2026-07-31-ptc-language-dispatch.zh.md)所有。私有的实验性 CPython 子进程后端（`dsh-experimental-code-runtime-python`）采用同一 portable-identifier 契约。 `RESERVED_BINDING_GLOBALS` 先于后端本身编码了 Python bootstrap 的具体设计：它恰好 seed `__builtins__`/`__name__`，并把程序包装在 `__dsh_main__` 之下。任何 seed 额外模块 global（`__doc__`、`__loader__`、`__spec__`、`__file__`、`__package__` 等）的 Python 后端必须在同一改动中扩宽此集合，正如新增一门语言即扩宽 `PORTABLE_RESERVED_WORDS`——bootstrap 会 seed 却不在集合中的名称，正是本约定要防止的可移植性分裂。 diff --git a/.agents/notes/implemented/architecture/2026-07-31-code-runtime-python-fd3-protocol.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-31-code-runtime-python-fd3-protocol.i18n.yaml index 312c609705..06974904b9 100644 --- a/.agents/notes/implemented/architecture/2026-07-31-code-runtime-python-fd3-protocol.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-31-code-runtime-python-fd3-protocol.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-31-code-runtime-python-fd3-protocol.md -2026-07-31-code-runtime-python-fd3-protocol.md: 5572fe58cb1dd8832ff9405670afc7f80a20362c -2026-07-31-code-runtime-python-fd3-protocol.zh.md: 6254e94a7b48b38edfbe23a6ea0b994d04ac21f4 +2026-07-31-code-runtime-python-fd3-protocol.md: cd8a42b509598d4782fc7c0637839e0dfd06f289 +2026-07-31-code-runtime-python-fd3-protocol.zh.md: a6454c17dc23e3f6385fe2dc3b46eabdb241faff diff --git a/.agents/notes/implemented/architecture/2026-07-31-code-runtime-python-fd3-protocol.md b/.agents/notes/implemented/architecture/2026-07-31-code-runtime-python-fd3-protocol.md index 5572fe58cb..cd8a42b509 100644 --- a/.agents/notes/implemented/architecture/2026-07-31-code-runtime-python-fd3-protocol.md +++ b/.agents/notes/implemented/architecture/2026-07-31-code-runtime-python-fd3-protocol.md @@ -2,13 +2,15 @@ Status: implemented +The CPython code runtime now lives at `packages/experimental/code-runtime-python` (private, npm name `@deepseek-ai/dsh-experimental-code-runtime-python`); promotion to a released package follows the experimental-packages decision. + English | [中文](2026-07-31-code-runtime-python-fd3-protocol.zh.md) ## Problem -`@deepseek-ai/dsh-code-runtime-python` owns the wire protocol intended for a CPython code-runtime provider. Such a provider runs each model program in a fresh `python3 -I` subprocess and bridges binding calls and completion values over the child's fd 3. The host cannot trust that channel: model code has full access to fd 3 and can forge any frame, so every inbound frame is hostile input that the host must validate and rebuild before reading. The protocol also has to carry lossless JSON without the depth limit `JSON.stringify` and `json.dumps` impose, because the seam's `CodeJsonValue` is depth-unbounded. +`@deepseek-ai/dsh-experimental-code-runtime-python` owns the wire protocol intended for a CPython code-runtime provider. Such a provider runs each model program in a fresh `python3 -I` subprocess and bridges binding calls and completion values over the child's fd 3. The host cannot trust that channel: model code has full access to fd 3 and can forge any frame, so every inbound frame is hostile input that the host must validate and rebuild before reading. The protocol also has to carry lossless JSON without the depth limit `JSON.stringify` and `json.dumps` impose, because the seam's `CodeJsonValue` is depth-unbounded. -The package ships the protocol independently from a runtime implementation. It exports no `PythonCodeRuntime`, subprocess path, or Python-side JSON codec; those remain work for a future provider. The protocol builds on the [portable identifier seam](2026-07-31-code-runtime-portable-identifier-seam.md). +The private experimental package contains both the protocol and runtime implementation: `PythonCodeRuntime` (the plugin's default export), the `python3 -I` subprocess path, and the Python-side JSON codec all live in `@deepseek-ai/dsh-experimental-code-runtime-python`. The protocol builds on the [portable identifier seam](2026-07-31-code-runtime-portable-identifier-seam.md). ## Decision @@ -20,24 +22,24 @@ The package ships the protocol independently from a runtime implementation. It e `py/protocol.py` mirrors the message shapes as `TypedDict`s and re-declares the two surfaces both sides EXECUTE against — `PROTOCOL_FD = 3` and `log_truncation_marker` — with byte-identical text. -The package remains independently buildable with protocol-only exports. `check-workspace-constraints` reads every `packages///package.json` unconditionally, while the coverage and invariant-topology checks exercise the package as soon as its directory exists. +The package ships the runtime alongside the protocol; it remains independently buildable. `check-workspace-constraints` reads every `packages///package.json` unconditionally, while the coverage and invariant-topology checks exercise the package as soon as its directory exists. ## Wire contract -Frames are JSON-lines on fd 3, one object per line, leaving stdout/stderr free for the program's own output. Child → host: `boot-ack`, `call`, `log`, `done`. Host → child: `boot` (first frame), `run` (after `boot-ack`), and one `reply` per `call`. The `log` frame's `truncated` flag marks the frame that IS the child ledger's own truncation marker, so the host stops capturing at the same point the child did instead of inferring it from its own budget. `done.error.kind` is one of `exception`, `invalid-output`, `output-limit`; wall/CPU budgets, aborts, and substrate death are observed host-side, not carried as frames. +Frames are JSON-lines on fd 3, one object per line, leaving stdout/stderr free for the program's own output. Child → host: `boot-ack`, `call`, `log`, `done`. Host → child: `boot` (first frame), `run` (after `boot-ack`), and one `reply` per `call`. The `log` frame's `truncated` flag marks the frame that IS the child ledger's own truncation marker, so the host stops capturing at the same point the child did instead of inferring it from its own budget. The `log` frame's `open` flag marks an unterminated line committed by an explicit flush: the host holds it and appends the next frame to the same entry, so an explicit flush followed by more text reads back as one line rather than a fake newline. The one exception is truncation: when a later over-budget frame trips the ledger, the already-billed prefix is committed as its own entry and the truncation marker follows it (marker last, no re-charge). The merged entry's wire cost is billed exactly once, split incrementally across its fragments on both sides (O(k) for k fragments, never a re-walk of the whole hold): the FIRST fragment pays the full JSON-string cost plus the separator, each continuation and the closing frame pay only their content; the host's exact-cost caps are `logBudget - 1` for a first fragment (the ledger's reserved byte, matching `admit`) and `logBudget + 2` for a continuation or closing frame (billed without the two quotes), and `jsonStringCostUpTo` returns `undefined` below a 2-byte cap; the child keys its split billing off `_open_started` alone, so a closing frame bills as the merged tail. `done.error.kind` is one of `exception`, `invalid-output`, `output-limit`; wall/CPU budgets, aborts, and substrate death are observed host-side, not carried as frames. ## Mirror alignment -`py/protocol.py` and `src/protocol.ts` agree that `LogMessage` carries `truncated`, `DoneMessage.error` carries `kind`, and `Namespace` may carry `errorClass`. `tests/protocol-mirror.e2e.ts` spawns a real `python3` and asserts `PROTOCOL_FD`, `log_truncation_marker`, and each `TypedDict`'s required and optional wire field sets against `src/protocol.ts`. A renamed or dropped field, or a required/optional mismatch, fails the test. Field *types* are not compared across the language boundary; review and a future provider's real-subprocess suite own that gap. +`py/protocol.py` and `src/protocol.ts` agree that `LogMessage` carries `truncated`, `DoneMessage.error` carries `kind`, and `Namespace` may carry `errorClass`. `tests/protocol-mirror.e2e.ts` spawns a real `python3` and asserts `PROTOCOL_FD`, `log_truncation_marker`, and each `TypedDict`'s required and optional wire field sets against `src/protocol.ts`. A renamed or dropped field, or a required/optional mismatch, fails the test. Field *types* are not compared across the language boundary; review and the runtime's real-subprocess suite (`runtime.spec.ts`) own that gap. ## Alternatives considered -**Require a future Python JSON codec (`_encode_json_plain` / `_decode_json_plain`) to live in `py/protocol.py` for cross-side symmetry with `protocol.ts`.** Rejected. The repository's "prefer symmetry for parallel values" rule points at genuinely parallel values; these are not. The host-side codec in `protocol.ts` validates hostile input and is self-contained. A child-side codec would produce trusted output and belong with bootstrap-owned emission and cost accounting; forcing only its entry points into `protocol.py` would couple the vocabulary mirror to runtime internals or create an import cycle. `protocol.py` remains a pure wire-vocabulary mirror. No Python codec ships in this package. +**Require a future Python JSON codec (`_encode_json_plain` / `_decode_json_plain`) to live in `py/protocol.py` for cross-side symmetry with `protocol.ts`.** Rejected. The repository's "prefer symmetry for parallel values" rule points at genuinely parallel values; these are not. The host-side codec in `protocol.ts` validates hostile input and is self-contained. A child-side codec would produce trusted output and belong with bootstrap-owned emission and cost accounting; forcing only its entry points into `protocol.py` would couple the vocabulary mirror to runtime internals or create an import cycle. `protocol.py` remains a pure wire-vocabulary mirror; the codec (`_encode_json_plain` / `_decode_json_plain`) lives in `bootstrap.py` with the runtime it serves. **Keep the protocol files outside a buildable package until a runtime ships.** Rejected: the workspace-constraint, coverage, and invariant-topology checks require every directory under `packages//` to be a buildable package, and the protocol has independent tests and a public wire vocabulary. ## Consequences -Bought: the fd-3 protocol and its hostile-input codec form a self-contained, fully unit-covered layer, with an executing guard against TypeScript/Python field-set drift. A future runtime can consume a reviewed wire contract. +Bought: the fd-3 protocol and its hostile-input codec form a self-contained, fully unit-covered layer, with an executing guard against TypeScript/Python field-set drift. The runtime built on it (`bootstrap.py`) consumes the reviewed wire contract. -Cost: the package name denotes a Python runtime family while `src/index.ts` exports only the protocol vocabulary. The mirror e2e compares field names and required/optional status across the two sides but not field types; comparing type declarations across TypeScript and Python has no mechanical equivalent, so review and the future runtime's real-subprocess suite retain that responsibility. +Cost: the package name denotes a Python runtime family and `src/index.ts` exports the full `PythonCodeRuntime` implementation, so the protocol vocabulary is only one part of the package surface. The mirror e2e compares field names and required/optional status across the two sides but not field types; comparing type declarations across TypeScript and Python has no mechanical equivalent, so review and the runtime's real-subprocess suite retain that responsibility. diff --git a/.agents/notes/implemented/architecture/2026-07-31-code-runtime-python-fd3-protocol.zh.md b/.agents/notes/implemented/architecture/2026-07-31-code-runtime-python-fd3-protocol.zh.md index 6254e94a7b..a6454c17dc 100644 --- a/.agents/notes/implemented/architecture/2026-07-31-code-runtime-python-fd3-protocol.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-31-code-runtime-python-fd3-protocol.zh.md @@ -2,13 +2,15 @@ Status: implemented +CPython 代码运行时现在位于 `packages/experimental/code-runtime-python`（私有，npm 名 `@deepseek-ai/dsh-experimental-code-runtime-python`）；提升为发布包遵循 experimental-packages 决策。 + [English](2026-07-31-code-runtime-python-fd3-protocol.md) | 中文 ## Problem -`@deepseek-ai/dsh-code-runtime-python` 负责供 CPython code-runtime 提供方使用的 wire protocol。这样的提供方会在全新的 `python3 -I` 子进程中运行每个模型程序，并通过子进程 fd 3 桥接 binding 调用与完成值。Host 不能信任这条通道：模型代码可以完全访问 fd 3 并伪造任意帧，因此 host 必须把每个入站帧视为敌意输入，先校验并重建后才能读取。协议还必须承载无深度限制的 lossless JSON，因为 seam 的 `CodeJsonValue` 深度无界，而 `JSON.stringify` 和 `json.dumps` 都有递归深度限制。 +`@deepseek-ai/dsh-experimental-code-runtime-python` 负责供 CPython code-runtime 提供方使用的 wire protocol。这样的提供方会在全新的 `python3 -I` 子进程中运行每个模型程序，并通过子进程 fd 3 桥接 binding 调用与完成值。Host 不能信任这条通道：模型代码可以完全访问 fd 3 并伪造任意帧，因此 host 必须把每个入站帧视为敌意输入，先校验并重建后才能读取。协议还必须承载无深度限制的 lossless JSON，因为 seam 的 `CodeJsonValue` 深度无界，而 `JSON.stringify` 和 `json.dumps` 都有递归深度限制。 -该包独立交付协议，不包含 runtime 实现。它不导出 `PythonCodeRuntime`、子进程路径或 Python 侧 JSON codec；这些属于未来提供方。协议建立在[可移植标识符 seam](2026-07-31-code-runtime-portable-identifier-seam.zh.md)之上。 +这个私有实验包同时包含协议与 runtime 实现：`PythonCodeRuntime`（插件的默认导出）、`python3 -I` 子进程路径与 Python 侧 JSON codec 都在 `@deepseek-ai/dsh-experimental-code-runtime-python` 中。协议建立在[可移植标识符 seam](2026-07-31-code-runtime-portable-identifier-seam.zh.md)之上。 ## Decision @@ -20,24 +22,24 @@ Status: implemented `py/protocol.py` 用 `TypedDict` 镜像消息形状，并重新声明两侧都会 EXECUTE 的两个面——`PROTOCOL_FD = 3` 与 `log_truncation_marker`——文本逐字节一致。 -该包只导出协议，同时保持独立可构建。`check-workspace-constraints` 会无条件读取每个 `packages///package.json`，coverage 与 invariant-topology 检查则会在包目录存在时立即覆盖该包。 +该包随协议一起交付 runtime，同时保持独立可构建。`check-workspace-constraints` 会无条件读取每个 `packages///package.json`，coverage 与 invariant-topology 检查则会在包目录存在时立即覆盖该包。 ## Wire contract -帧是 fd 3 上的 JSON-lines，每行一个对象，让 stdout/stderr 空出给程序自己的输出。Child → host：`boot-ack`、`call`、`log`、`done`。Host → child：`boot`（首帧）、`run`（在 `boot-ack` 之后）、以及每个 `call` 对应一个 `reply`。`log` 帧的 `truncated` 标志标记那个本身就是子进程 ledger 截断标记的帧，使 host 在与子进程相同的点停止捕获，而不是从自己的预算去推断。`done.error.kind` 是 `exception`、`invalid-output`、`output-limit` 之一；wall/CPU 预算、abort、substrate 死亡都在 host 侧观测，不作为帧携带。 +帧是 fd 3 上的 JSON-lines，每行一个对象，让 stdout/stderr 空出给程序自己的输出。Child → host：`boot-ack`、`call`、`log`、`done`。Host → child：`boot`（首帧）、`run`（在 `boot-ack` 之后）、以及每个 `call` 对应一个 `reply`。`log` 帧的 `truncated` 标志标记那个本身就是子进程 ledger 截断标记的帧，使 host 在与子进程相同的点停止捕获，而不是从自己的预算去推断。`log` 帧的 `open` 标志标记由显式 flush 提交的未结束行：宿主持有它并把下一个帧追加到同一条目，因此显式 flush 后接更多文本读回为一行而不是假换行。唯一例外是截断：当后续超预算帧触发账本时，已计费的前缀作为独立条目先提交，截断 marker 跟在后面（marker 保持末位，无重复计费）。合并条目的线上成本恰好计费一次，在两侧按片段增量分摊（k 个片段 O(k)，绝不对整个持有重走）：首片段付完整 JSON 字符串成本加分隔符，每个续接与闭合帧只付内容；宿主精确成本 cap 是首片段 `logBudget - 1`（账本预留字节，与 `admit` 一致）、续接或闭合帧 `logBudget + 2`（不含两个引号计费），且 `jsonStringCostUpTo` 在低于 2 字节 cap 时返回 `undefined`；子进程按 `_open_started` 单独键控拆分计费，因此闭合帧按合并尾部计费。`done.error.kind` 是 `exception`、`invalid-output`、`output-limit` 之一；wall/CPU 预算、abort、substrate 死亡都在 host 侧观测，不作为帧携带。 ## Mirror alignment -`py/protocol.py` 与 `src/protocol.ts` 一致规定：`LogMessage` 携带 `truncated`，`DoneMessage.error` 携带 `kind`，`Namespace` 可以携带 `errorClass`。`tests/protocol-mirror.e2e.ts` 启动真实 `python3`，对照 `src/protocol.ts` 断言 `PROTOCOL_FD`、`log_truncation_marker` 以及每个 `TypedDict` 的必填和可选 wire 字段集。字段改名、删除或必填／可选性不一致都会使测试失败。字段*类型*不跨语言边界比较；这项缺口由评审和未来提供方的真实子进程套件负责。 +`py/protocol.py` 与 `src/protocol.ts` 一致规定：`LogMessage` 携带 `truncated`，`DoneMessage.error` 携带 `kind`，`Namespace` 可以携带 `errorClass`。`tests/protocol-mirror.e2e.ts` 启动真实 `python3`，对照 `src/protocol.ts` 断言 `PROTOCOL_FD`、`log_truncation_marker` 以及每个 `TypedDict` 的必填和可选 wire 字段集。字段改名、删除或必填／可选性不一致都会使测试失败。字段*类型*不跨语言边界比较；这项缺口由评审和 runtime 的真实子进程套件（`runtime.spec.ts`）负责。 ## Alternatives considered -**要求未来的 Python JSON codec（`_encode_json_plain` / `_decode_json_plain`）放进 `py/protocol.py`，以便与 `protocol.ts` 跨侧对称。**拒绝。仓库的 “prefer symmetry for parallel values” 规则指向真正平行的值；这两者不是。`protocol.ts` 中的 host 侧 codec 校验敌意输入且自包含。Child 侧 codec 会产出受信任输出，应与 bootstrap 拥有的发出逻辑和成本核算放在一起；只把入口强塞进 `protocol.py` 会让 vocabulary 镜像耦合 runtime 内部实现，或制造 import 环。`protocol.py` 保持纯 wire-vocabulary 镜像。本包尚未交付 Python codec。 +**要求未来的 Python JSON codec（`_encode_json_plain` / `_decode_json_plain`）放进 `py/protocol.py`，以便与 `protocol.ts` 跨侧对称。**拒绝。仓库的 “prefer symmetry for parallel values” 规则指向真正平行的值；这两者不是。`protocol.ts` 中的 host 侧 codec 校验敌意输入且自包含。Child 侧 codec 会产出受信任输出，应与 bootstrap 拥有的发出逻辑和成本核算放在一起；只把入口强塞进 `protocol.py` 会让 vocabulary 镜像耦合 runtime 内部实现，或制造 import 环。`protocol.py` 保持纯 wire-vocabulary 镜像；codec（`_encode_json_plain`／`_decode_json_plain`）与它所服务的 runtime 一起位于 `bootstrap.py`。 **在 runtime 交付前把协议文件放在不可构建的包外。**拒绝：workspace-constraint、coverage 与 invariant-topology 检查要求 `packages//` 下的每个目录都是可构建包，而协议本身拥有独立测试与公开 wire vocabulary。 ## Consequences -收获：fd-3 协议及其敌意输入 codec 构成自包含、unit 全覆盖的一层，并由执行中的 guard 防止 TypeScript／Python 字段集漂移。未来 runtime 可以直接消费经过评审的 wire contract。 +收获：fd-3 协议及其敌意输入 codec 构成自包含、unit 全覆盖的一层，并由执行中的 guard 防止 TypeScript／Python 字段集漂移。基于它构建的 runtime（`bootstrap.py`）消费经过评审的 wire contract。 -代价：包名表示 Python runtime 家族，而 `src/index.ts` 只导出协议 vocabulary。mirror e2e 会比较两侧字段名与必填／可选状态，但不比较字段类型；跨 TypeScript 与 Python 比较类型声明没有机械等价物，因此评审与未来 runtime 的真实子进程套件继续负责这项检查。 +代价：包名表示 Python runtime 家族，而 `src/index.ts` 导出完整的 `PythonCodeRuntime` 实现，协议 vocabulary 只是包表面的一部分。mirror e2e 会比较两侧字段名与必填／可选状态，但不比较字段类型；跨 TypeScript 与 Python 比较类型声明没有机械等价物，因此评审与 runtime 的真实子进程套件继续负责这项检查。 diff --git a/.agents/notes/implemented/architecture/2026-08-04-draft-provider-endpoint-interrogation.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-04-draft-provider-endpoint-interrogation.i18n.yaml index e57fda2a7f..277670d745 100644 --- a/.agents/notes/implemented/architecture/2026-08-04-draft-provider-endpoint-interrogation.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-04-draft-provider-endpoint-interrogation.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-04-draft-provider-endpoint-interrogation.md -2026-08-04-draft-provider-endpoint-interrogation.md: 502d9bab15dcb91a59deb26443d869a36b028b48 -2026-08-04-draft-provider-endpoint-interrogation.zh.md: e0605369c7f4707eb682cc1c32d11123140b449a +2026-08-04-draft-provider-endpoint-interrogation.md: d4112d813ad4f5781b74639209d13952e459f7dd +2026-08-04-draft-provider-endpoint-interrogation.zh.md: 1626a34cb3163949d70688cefeec77d328c62caa diff --git a/.agents/notes/implemented/architecture/2026-08-04-draft-provider-endpoint-interrogation.md b/.agents/notes/implemented/architecture/2026-08-04-draft-provider-endpoint-interrogation.md index 502d9bab15..d4112d813a 100644 --- a/.agents/notes/implemented/architecture/2026-08-04-draft-provider-endpoint-interrogation.md +++ b/.agents/notes/implemented/architecture/2026-08-04-draft-provider-endpoint-interrogation.md @@ -17,11 +17,11 @@ The awkward part is that the question is about something that does not exist yet Interrogation is keyed by **settings namespace**, not by provider route: - `ctx.llm.registerModelDiscovery(settingsNs, discover)` lets an adapter plugin offer to interrogate endpoints for the namespace it owns, and `ctx.llm.discoverModels(settingsNs, request)` asks. There is no way to enumerate which namespaces registered: a surface that cannot interrogate learns it from the refusal, and a list nothing consumed would be a required wire field doing nothing. The namespace is the right key because a configuration surface already holds it from the configurable-provider directory, and because a provider being added has no route to name. -- `LlmModelDiscoveryRequest` carries the draft — an optional `provider`, an optional `baseURL`, an optional `api`, an optional `apiKey`, and a signal — and needs at least one of `provider` or `baseURL` to have anything to answer about. `provider` exists because a route the adapter already describes is answered from its own registry with no network call at all; only a route it does not describe reaches an endpoint. Nothing in this path writes settings or credentials. The one read is the credential of a route the request names: a configuration surface holds a redacted descriptor rather than the stored secret, so the draft's `apiKey` is present only while the user is typing one, and without that read an already-configured route would be interrogated unauthenticated and answer 401. The typed key wins, being the one under test. +- `LlmModelDiscoveryRequest` carries the draft — an optional `provider`, an optional `baseURL`, an optional `api`, an optional `apiKey`, and a signal — and needs at least one of `provider` or `baseURL` to have anything to answer about. `provider` exists because a route the adapter already describes is answered from its own registry with no network call at all; only a route it does not describe reaches an endpoint. Nothing in this path writes settings or credentials. A named configured route reads its stored credential and deployment-owned profile `headers` inside the Host: the credential is write-only and the curated Models page does not edit headers, so neither can be reconstructed from that page's draft. The typed key wins over the stored credential, while the profile headers still accompany the request. - `LlmDiscoveredModel` makes every field but `id` optional, because most listings disclose an id and nothing else. The reply is candidates, not a catalog: a surface adopting one still owes the capacities the adapter requires. - `llm.discoverModels` carries the same draft over the wire. Its `apiKey` is the third and last payload on which a secret may ride, alongside `settings.update`/`mutate` and `credentials.set`, and it is never stored or echoed back. It does ride the client's outgoing envelope like every other secret-bearing payload, where a `subscribeEnvelopes()` observer can see it; redacting that tap is a configuration-plane-wide change, not this method's to make alone. Connection authenticates the method with the complete Host API: it makes the host issue a GET to a caller-chosen URL and reports the outcome, which an anonymous caller must not receive. Every refusal folds into `model-discovery-failed`, whose message is the adapter's own text and whose details name the endpoint asked but never the credential offered. -`dsh-llm-pi-ai` implements the wire path as a plain `GET {baseURL}/models`, reading `openai-completions` and `openai-responses`: their `GET /models` shape with bearer auth is the one a gateway, a self-hosted server, and the official endpoints all agree on. Azure is excluded despite its OpenAI lineage — it authenticates with an `api-key` header and requires an `api-version` query — and Codex uses OAuth; both would have reported an authentication failure as a provider with no models. Every other protocol answers `DISCOVERY_UNSUPPORTED`, so the surface falls back to hand-entry rather than reporting a guessed response shape as an empty provider. `baseURL` is treated as a prefix rather than a URL to resolve against, so a deployment path such as `https://gateway.example/openai/v1` keeps its segments. The reply is read under a four-megabyte ceiling enforced on the bytes actually received — the endpoint is a URL the user typed, so a declared `content-length` is checked first as a courtesy but never trusted as the bound, matching `dsh-web-fetch`'s two-stage shape for its own caller-supplied URLs. +`dsh-llm-pi-ai` implements the wire path as a plain `GET {baseURL}/models`, reading `openai-completions` and `openai-responses`: their `GET /models` shape with bearer auth is the one a gateway, a self-hosted server, and the official endpoints all agree on. Profile resolution rejects names and values Fetch cannot represent, so a malformed deployment header is reported as a configuration error before interrogation. Configured profile headers are installed first; the fixed JSON accept header, a typed-or-stored bearer credential, and Harness attribution then win case-insensitive collisions in that order. Azure is excluded despite its OpenAI lineage — it authenticates with an `api-key` header and requires an `api-version` query — and Codex uses OAuth; both would have reported an authentication failure as a provider with no models. Every other protocol answers `DISCOVERY_UNSUPPORTED`, so the surface falls back to hand-entry rather than reporting a guessed response shape as an empty provider. `baseURL` is treated as a prefix rather than a URL to resolve against, so a deployment path such as `https://gateway.example/openai/v1` keeps its segments. The reply is read under a four-megabyte ceiling enforced on the bytes actually received — the endpoint is a URL the user typed, so a declared `content-length` is checked first as a courtesy but never trusted as the bound, matching `dsh-web-fetch`'s two-stage shape for its own caller-supplied URLs. ### Why not pi-ai's own refresh machinery @@ -33,7 +33,7 @@ pi-ai supplies `createProvider({ fetchModels })` plus `Models.refresh()` and a ` **Put the capability on `LlmAdapter`.** Adapters are reached through a route registration, so this has the same problem, plus it would make an adapter instance answer questions about endpoints it does not serve. -**Have the host read the stored profile instead of accepting a draft.** No secret would cross the wire for an already-configured provider. But adding a provider would then require saving an unusable configuration first, and a form whose endpoint was edited but not yet saved would silently interrogate the old one. Accepting the draft keeps what the user sees and what is asked identical — with the credential as the one exception, because it is the one field a surface is never shown and so can never put in the draft. +**Have the host read the entire stored profile instead of accepting a draft.** No secret would cross the wire for an already-configured provider. But adding a provider would then require saving an unusable configuration first, and a form whose endpoint was edited but not yet saved would silently interrogate the old one. The draft remains authoritative for the endpoint and protocol. The narrow Host-side exceptions are the stored credential, which is write-only, and profile headers, which remain deployment configuration rather than Models-page fields. **Interrogate every pi-ai protocol.** Anthropic's listing happens to share OpenAI's envelope, and Google's does not. Supporting the ones that are easy would make coverage arbitrary and, worse, make a wrong guess at a response shape indistinguishable from a provider with no models. A protocol that says it cannot be interrogated sends the user to hand-entry, which is the documented fallback. @@ -41,10 +41,10 @@ pi-ai supplies `createProvider({ fetchModels })` plus `Models.refresh()` and a ` ## Consequences -A person adding a gateway can ask it what it serves instead of hunting through its documentation, and the answer arrives as candidates they choose from rather than as configuration written behind their back. The seam gained a registry that is deliberately small: one offer per namespace, no storage, no lifecycle beyond the fiber. +A person adding a gateway can ask it what it serves instead of hunting through its documentation, and the answer arrives as candidates they choose from rather than as configuration written behind their back. An already-configured enterprise gateway uses the same deployment headers for interrogation and model requests without adding a header injection field to the browser protocol. The seam gained a registry that is deliberately small: one offer per namespace, no storage, no lifecycle beyond the fiber. What it costs: the wire gained a third secret-carrying payload, so the configuration plane's write-only surface is now three methods rather than two. Discovery coverage is protocol-shaped rather than provider-shaped — an Anthropic-compatible gateway must be filled in by hand even though its listing would parse. And because nothing re-runs the question, a model list is still only as current as its last edit; that is the same trade the layer below made deliberately. ## Testing -`packages/llm/llm/tests/topology.spec.ts` covers the registry: one offer per namespace, disposal with the fiber, normalization that drops duplicate and unusable ids without inventing capacities, the `NO_DISCOVERY`/`INVALID_DISCOVERY` refusals, and the `model-discovery-failed` Remote mapping. `packages/llm/llm-pi-ai/tests/discovery.spec.ts` drives the probe against local HTTP servers — a listing with and without disclosed capacities, a preserved deployment path, an absent credential, a configured route supplying its own where the draft has none and a typed key winning over it, a catalog route answering without resolving one at all, dropped rows, 401/403 versus a server fault, a non-listing and a non-JSON body, an unreachable endpoint, caller cancellation, an unsupported protocol, and the size ceiling in both its declared-length and streamed forms. `packages/client/connection/tests/node-half.host.spec.ts` pins the `llm/discoverModels` `/api` carrier registration, while `packages/client/ui-settings-models/tests/provider-form.client.spec.tsx` verifies that the draft reaches the Remote whole, absent fields stay absent, and no settings namespace or credential is written before selection. +`packages/llm/llm/tests/topology.spec.ts` covers the registry: one offer per namespace, disposal with the fiber, normalization that drops duplicate and unusable ids without inventing capacities, the `NO_DISCOVERY`/`INVALID_DISCOVERY` refusals, and the `model-discovery-failed` Remote mapping. `packages/llm/llm-pi-ai/tests/discovery.spec.ts` drives the probe against local HTTP servers — a listing with and without disclosed capacities, a preserved deployment path, an absent credential, a configured route supplying its stored credential and headers while a typed key wins without resolving the stored one, a catalog route answering without resolving one at all, dropped rows, 401/403 versus a server fault, a non-listing and a non-JSON body, an unreachable endpoint, caller cancellation, an unsupported protocol, and the size ceiling in both its declared-length and streamed forms. `packages/llm/llm-pi-ai/tests/loader-composition.spec.ts` boots settings and credentials through the Loader and proves settings-only headers reach `GET /models` with request-owned headers winning collisions. `packages/llm/llm-pi-ai/tests/adapter.spec.ts` rejects profile headers Fetch cannot represent, and `packages/llm/llm-pi-ai/tests/dynamic-config.spec.ts` proves a settings write reports that configuration error while its last good routes keep serving. `packages/client/connection/tests/node-half.host.spec.ts` pins the `llm/discoverModels` `/api` carrier registration, while `packages/client/ui-settings-models/tests/provider-form.client.spec.tsx` verifies that the draft reaches the Remote whole, absent fields stay absent, and no settings namespace or credential is written before selection. diff --git a/.agents/notes/implemented/architecture/2026-08-04-draft-provider-endpoint-interrogation.zh.md b/.agents/notes/implemented/architecture/2026-08-04-draft-provider-endpoint-interrogation.zh.md index e0605369c7..1626a34cb3 100644 --- a/.agents/notes/implemented/architecture/2026-08-04-draft-provider-endpoint-interrogation.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-04-draft-provider-endpoint-interrogation.zh.md @@ -17,11 +17,11 @@ Status: implemented 询问以 **settings namespace** 为键，而不是提供方路由： - `ctx.llm.registerModelDiscovery(settingsNs, discover)` 让适配器插件为自己拥有的 namespace 提供「询问端点」的能力，`ctx.llm.discoverModels(settingsNs, request)` 发起询问。没有任何办法枚举哪些 namespace 注册过：询问不了的界面会从那句拒绝里知道，而一份无人消费的列表只会变成一个什么都不做的必填协议字段。以 namespace 为键是对的，因为配置界面已经从可配置提供方目录里拿到了它，也因为正在新增的提供方没有路由可点名。 -- `LlmModelDiscoveryRequest` 携带草稿——可选的 `provider`、可选的 `baseURL`、可选的 `api`、可选的 `apiKey`，以及一个 signal——且 `provider` 与 `baseURL` 至少要有一个，才有东西可答。`provider` 之所以存在，是因为适配器已经描述过的路由直接由它自己的注册表作答、完全不联网；只有它未描述的路由才会抵达某个端点。这条路径不写 settings 与 credentials。唯一的读取是请求所点名路由的凭据：配置界面拿到的是脱敏描述符而非已存的机密，因此草稿里的 `apiKey` 只在用户正键入时才存在；没有这次读取，已配置好的路由就会被不带认证地询问，只换回一个 401。键入的密钥优先，因为那正是被测试的那一把。 +- `LlmModelDiscoveryRequest` 携带草稿——可选的 `provider`、可选的 `baseURL`、可选的 `api`、可选的 `apiKey`，以及一个 signal——且 `provider` 与 `baseURL` 至少要有一个，才有东西可答。`provider` 之所以存在，是因为适配器已经描述过的路由直接由它自己的注册表作答、完全不联网；只有它未描述的路由才会抵达某个端点。这条路径不写 settings 与 credentials。已配置且具名的路由会在 Host 内读取已存凭据和部署方持有的 profile `headers`：凭据只写，而精选的 Models 页面不编辑 headers，因此页面草稿无法重建两者。键入的密钥优先于已存凭据，profile headers 则仍随请求发送。 - `LlmDiscoveredModel` 除 `id` 外每个字段都可选，因为大多数列表只公布 id。回复是候选而非 catalog：采纳其中一条的界面仍要补上适配器所需的容量。 - `llm.discoverModels` 把同一份草稿送过协议层。它的 `apiKey` 是可承载机密的第三个、也是最后一个载荷（另两个是 `settings.update`/`mutate` 与 `credentials.set`），且绝不被存储或回显。它确实会像其他承载机密的载荷一样随客户端外发信封同行，`subscribeEnvelopes()` 观察者看得到；把那个抽头脱敏是整个配置面的改动，不该由这一个方法独自决定。Connection 用与完整 Host API 相同的会话认证该方法：它让宿主向调用方选定的 URL 发起 GET 并回报结果，匿名调用者绝不能获得这类探测能力。每一种拒绝都折叠为 `model-discovery-failed`，其消息是适配器自己的文本，details 点名被询问的端点，绝不点名所提供的凭据。 -`dsh-llm-pi-ai` 的实现只是一次朴素的 `GET {baseURL}/models`，且仅限 OpenAI 兼容协议。它们的列表形状是网关、自建服务与官方端点三方一致认可的那一种，而这正是该动作存在的场景。其余协议一律以 `DISCOVERY_UNSUPPORTED` 回答，让界面回退到手工填写，而不是把猜错的响应形状报成一个空提供方。`baseURL` 按前缀而非待解析 URL 处理，因此 `https://gateway.example/openai/v1` 这类部署路径会保留其路径段。回复在四兆字节上限下读取，且上限落在实际收到的字节上——端点是用户自己填的 URL，因此会先看声明的 `content-length` 作为善意提示，但绝不把它当作边界；这与 `dsh-web-fetch` 面对自己的调用方提供 URL 时所用的两段式形状一致。 +`dsh-llm-pi-ai` 的实现只是一次朴素的 `GET {baseURL}/models`，且仅限 OpenAI 兼容协议。它们的列表形状是网关、自建服务与官方端点三方一致认可的那一种，而这正是该动作存在的场景。Profile 解析会拒绝 Fetch 无法表示的名称与值，因此格式错误的部署 header 会在询问前以配置错误报告。已配置的 profile headers 最先装入；固定的 JSON accept header、键入或已存的 bearer 凭据以及 Harness attribution 随后依次以大小写不敏感方式赢得冲突。其余协议一律以 `DISCOVERY_UNSUPPORTED` 回答，让界面回退到手工填写，而不是把猜错的响应形状报成一个空提供方。`baseURL` 按前缀而非待解析 URL 处理，因此 `https://gateway.example/openai/v1` 这类部署路径会保留其路径段。回复在四兆字节上限下读取，且上限落在实际收到的字节上——端点是用户自己填的 URL，因此会先看声明的 `content-length` 作为善意提示，但绝不把它当作边界；这与 `dsh-web-fetch` 面对自己的调用方提供 URL 时所用的两段式形状一致。 ### 为什么不用 pi-ai 自己的 refresh 机制 @@ -33,7 +33,7 @@ pi-ai 提供了 `createProvider({ fetchModels })` 加上 `Models.refresh()` 与 **把能力挂在 `LlmAdapter` 上。** 适配器要经由路由注册才能抵达，因此问题相同；而且这会让一个适配器实例去回答它并不服务的端点的问题。 -**让 host 读已存 profile，而不是接受草稿。** 对已配置好的提供方来说，不会有机密跨越协议层。但这样一来新增提供方就必须先保存一份不可用的配置，而端点已改却尚未保存的表单会静默地去询问旧地址。接受草稿让用户看见的与被询问的保持一致——凭据是唯一的例外，因为它是从不向界面展示、因而永远无法放进草稿的那个字段。 +**让 Host 读取整个已存 profile，而不是接受草稿。** 对已配置好的提供方来说，不会有机密跨越协议层。但这样一来新增提供方就必须先保存一份不可用的配置，而端点已改却尚未保存的表单会静默地去询问旧地址。草稿仍是端点和协议的权威来源。Host 侧的狭窄例外是只写的已存凭据，以及仍属部署配置、而非 Models 页面字段的 profile headers。 **询问 pi-ai 的每一种协议。** Anthropic 的列表恰好与 OpenAI 共用同一层信封，而 Google 的不是。只支持容易的那几种会让覆盖范围变得任意；更糟的是，猜错的响应形状会与「该提供方没有模型」无法区分。一个明说自己无法被询问的协议，会把用户送去手工填写——那正是既定的回退路径。 @@ -41,10 +41,10 @@ pi-ai 提供了 `createProvider({ fetchModels })` 加上 `Models.refresh()` 与 ## Consequences -接入网关的人可以直接问它服务什么，而不必去翻它的文档；答案以候选形式抵达，由用户自己挑选，而不是被背着写进配置。seam 因此多了一个刻意保持很小的注册表：每个 namespace 一份、不存储、生命周期不超出 fiber。 +接入网关的人可以直接问它服务什么，而不必去翻它的文档；答案以候选形式抵达，由用户自己挑选，而不是被背着写进配置。已配置的企业网关会为询问与模型请求使用同一组部署 headers，而无需给浏览器协议增加 header 注入字段。seam 因此多了一个刻意保持很小的注册表：每个 namespace 一份、不存储、生命周期不超出 fiber。 代价是：协议层多了第三个承载机密的载荷，配置面的只写接口从两个方法变成三个。发现覆盖范围按协议而非按提供方划分——一个 Anthropic 兼容网关即便其列表能被解析，也仍须手工填写。而且由于没有任何环节会重跑该询问，模型列表的新鲜度依旧只到最近一次编辑为止；这与下层刻意做出的取舍是同一个。 ## Testing -`packages/llm/llm/tests/topology.spec.ts` 覆盖注册表：每个 namespace 一份、随 fiber dispose（资源释放）、丢弃重复与不可用 id 且不凭空补容量的归一化、`NO_DISCOVERY`/`INVALID_DISCOVERY` 两种拒绝，以及 `model-discovery-failed` Remote 映射。`packages/llm/llm-pi-ai/tests/discovery.spec.ts` 针对本地 HTTP 服务器驱动探测——含与不含公布容量的列表、被保留的部署路径、无凭据、草稿没带密钥时已配置路由自行取用凭据且键入的密钥压过它、catalog 路由完全不解析凭据即作答、被丢弃的行、401/403 与服务器故障之别、非列表与非 JSON 响应、不可达端点、调用方取消、不支持的协议，以及尺寸上限的「声明长度」与「流式」两种形态。`packages/client/connection/tests/node-half.host.spec.ts` 固定 `llm/discoverModels` 的 `/api` 承载注册，`packages/client/ui-settings-models/tests/provider-form.client.spec.tsx` 则验证草稿完整抵达 Remote、缺席字段保持缺席，以及选择前没有 settings namespace 或凭据被写入。 +`packages/llm/llm/tests/topology.spec.ts` 覆盖注册表：每个 namespace 一份、随 fiber dispose（资源释放）、丢弃重复与不可用 id 且不凭空补容量的归一化、`NO_DISCOVERY`/`INVALID_DISCOVERY` 两种拒绝，以及 `model-discovery-failed` Remote 映射。`packages/llm/llm-pi-ai/tests/discovery.spec.ts` 针对本地 HTTP 服务器驱动探测——含与不含公布容量的列表、被保留的部署路径、无凭据、已配置路由提供自己的已存凭据与 headers 且键入的密钥无需解析已存凭据便可压过它、catalog 路由完全不解析凭据即作答、被丢弃的行、401/403 与服务器故障之别、非列表与非 JSON 响应、不可达端点、调用方取消、不支持的协议，以及尺寸上限的「声明长度」与「流式」两种形态。`packages/llm/llm-pi-ai/tests/loader-composition.spec.ts` 通过 Loader 启动 settings 与 credentials，并证明仅配置在 settings 中的 headers 会抵达 `GET /models`，且请求所持有的 headers 赢得冲突。`packages/llm/llm-pi-ai/tests/adapter.spec.ts` 拒绝 Fetch 无法表示的 profile headers，`packages/llm/llm-pi-ai/tests/dynamic-config.spec.ts` 证明 settings 写入会报告该配置错误，同时上一组可用路由仍继续服务。`packages/client/connection/tests/node-half.host.spec.ts` 固定 `llm/discoverModels` 的 `/api` 承载注册，`packages/client/ui-settings-models/tests/provider-form.client.spec.tsx` 则验证草稿完整抵达 Remote、缺席字段保持缺席，以及选择前没有 settings namespace 或凭据被写入。 diff --git a/.agents/notes/implemented/architecture/2026-08-05-large-session-jsonl-restore-pipeline.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-05-large-session-jsonl-restore-pipeline.i18n.yaml index 6f00f147f3..18313964ec 100644 --- a/.agents/notes/implemented/architecture/2026-08-05-large-session-jsonl-restore-pipeline.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-05-large-session-jsonl-restore-pipeline.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-05-large-session-jsonl-restore-pipeline.md -2026-08-05-large-session-jsonl-restore-pipeline.md: eab53c683880ef7095233ed8122e532eb5add547 -2026-08-05-large-session-jsonl-restore-pipeline.zh.md: 2cd0d2ca20074d6adb0735db08071638ae2ced88 +2026-08-05-large-session-jsonl-restore-pipeline.md: 309d9dc6bdb5c3160f3e6e76a8318915df58fe59 +2026-08-05-large-session-jsonl-restore-pipeline.zh.md: 28acd3ebe804dca22a0619c257ff3ad9c09500a9 diff --git a/.agents/notes/implemented/architecture/2026-08-05-large-session-jsonl-restore-pipeline.md b/.agents/notes/implemented/architecture/2026-08-05-large-session-jsonl-restore-pipeline.md index eab53c6838..309d9dc6bd 100644 --- a/.agents/notes/implemented/architecture/2026-08-05-large-session-jsonl-restore-pipeline.md +++ b/.agents/notes/implemented/architecture/2026-08-05-large-session-jsonl-restore-pipeline.md @@ -41,7 +41,7 @@ Borrowed seeds used by ordinary creation and fork paths still take a JSON snapsh - **Concatenate all plaintext before scanning** — rejected because it retains the compressed input, complete plaintext, whole-log UTF-8 string, line metadata, and parsed rows at the same time, and it rescans a torn-frame prefix. - **Implement a streaming JSON parser** — rejected because JSONL already provides record boundaries; native newline search plus `JSON.parse` removes the large intermediates without owning another parser or changing JSON semantics. - **Use a shared `WeakSet` while freezing restored events** — rejected because JSON materialization cannot produce cycles, and the set adds a lookup per object while retaining the complete graph during traversal. -- **Skip validation or freezing for restored values** — rejected because durable storage is a runtime boundary and `Session.events` promises immutable accepted history. The optimized path specializes those operations around stronger ownership facts instead of removing them. +- **Skip validation or freezing for restored values** — rejected because durable storage is a runtime boundary and Session read methods promise immutable accepted history. The optimized path specializes those operations around stronger ownership facts instead of removing them. ## Consequences diff --git a/.agents/notes/implemented/architecture/2026-08-05-large-session-jsonl-restore-pipeline.zh.md b/.agents/notes/implemented/architecture/2026-08-05-large-session-jsonl-restore-pipeline.zh.md index 2cd0d2ca20..28acd3ebe8 100644 --- a/.agents/notes/implemented/architecture/2026-08-05-large-session-jsonl-restore-pipeline.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-05-large-session-jsonl-restore-pipeline.zh.md @@ -41,7 +41,7 @@ Zstandard 结构扫描器会在解码前识别完整帧范围。系统单独解 - **扫描前拼接全部明文**：不予采纳，因为该方案会同时保留压缩输入、完整明文、整份日志的 UTF-8 字符串、行元数据和解析记录，并会重新扫描撕裂帧前缀。 - **实现流式 JSON 解析器**：不予采纳，因为 JSONL 已提供记录边界；使用原生换行搜索与 `JSON.parse` 就能移除大型中间结构，无需自行维护另一套解析器或改变 JSON 语义。 - **冻结恢复事件时共享一个 `WeakSet`**：不予采纳，因为 JSON 物化不可能产生循环引用，而该集合会对每个对象增加一次查找，并在遍历期间保留完整对象图。 -- **跳过恢复值的校验或冻结**：不予采纳，因为持久存储属于运行时边界，而 `Session.events` 承诺已接受历史不可变。优化路径利用更强的所有权事实特化这些操作，而不是将其移除。 +- **跳过恢复值的校验或冻结**：不予采纳，因为持久存储属于运行时边界，而 Session 读取方法承诺已接受历史不可变。优化路径利用更强的所有权事实特化这些操作，而不是将其移除。 ## 后果 diff --git a/.agents/notes/implemented/architecture/2026-08-05-session-preparation.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-05-session-preparation.i18n.yaml index 3e3b610f73..e96c1b2e6b 100644 --- a/.agents/notes/implemented/architecture/2026-08-05-session-preparation.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-05-session-preparation.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-05-session-preparation.md -2026-08-05-session-preparation.md: 50f1ea38e671c6aa7b0f4adaf2fecbf83decc23c -2026-08-05-session-preparation.zh.md: cd918d126b56b081bcc1b6aa43a10d662668102d +2026-08-05-session-preparation.md: 040c9f788173a7cedd91be33cbe7ced3ca758a06 +2026-08-05-session-preparation.zh.md: 44b609d488c6f8bb406370f9097eb2a085f7cc2b diff --git a/.agents/notes/implemented/architecture/2026-08-05-session-preparation.md b/.agents/notes/implemented/architecture/2026-08-05-session-preparation.md index 50f1ea38e6..040c9f7881 100644 --- a/.agents/notes/implemented/architecture/2026-08-05-session-preparation.md +++ b/.agents/notes/implemented/architecture/2026-08-05-session-preparation.md @@ -6,63 +6,40 @@ English | [中文](2026-08-05-session-preparation.zh.md) ## Problem -Cold history inspection and Agent resume independently materialized the same persisted session log. For a large compressed log, each operation repeated the full read, decompression, parse, validation, freezing, and Session construction. Pagination could therefore pay the cold-read cost again, while making a history query activate an Agent would couple a read lifecycle to a live Agent with no natural retirement point. +Fresh creation and persisted resume reached the same publication boundary through different construction flows. This obscured the invariant that setup must finish against one unpublished Session before that exact Session and its Agent become visible together. -Fresh creation and persisted resume also reached the same publication boundary through different construction flows. This obscured the invariant that setup must finish against one unpublished Session before that exact Session and its Agent become visible together. +Cold history inspection and Agent resume also independently materialized the same persisted session log, which this note originally answered with a persistence-side prepared-Session cache; that half is superseded below. ## Decision -`SessionPreparation` owns one exact unpublished `Session` until publication or rollback. It is a Session lifecycle object, not an Agent lifecycle or activation object. Fresh creation wraps the result of `SessionStore.prepare()`; persisted resume obtains a preparation from `SessionPersistence.prepare()`. +`SessionPreparation` owns one exact unpublished `Session` until publication or rollback. It is a Session lifecycle object, not an Agent lifecycle or activation object. Fresh creation wraps the result of `SessionStore.prepare()`; persisted resume reads the stored log through the session's write handle, appends `interruptedTurnClosers`, and wraps `SessionStore.prepare(id, { seed, meta, seedSource: 'persistence' })` — the restoration branch that validates and freezes the transferred graphs in place. The Agent loop consumes both forms through one setup-and-publication pipeline: it acquires the preparation, builds the private Agent context around `preparation.session`, awaits optional setup, publishes that exact Session and Agent, and disposes the preparation on every exit. Publication transfers the live lifecycle to the existing Session and Agent stores; `SessionPreparation` itself owns no Agent behavior. This refines the publication boundary from the [Agent lifecycle and ownership decision](2026-06-18-agent-lifecycle-and-ownership-contracts.md) without replacing its ownership model. -## Persisted preparation lifecycle +## Superseded: the persistence-side preparation lifecycle -A coordinator-backed persistence implementation loads one cold source into a prepared Session. The backend transfers fresh, mutually unaliased metadata and events together with the source-qualified revision that identifies those exact values; the Session restore path validates and freezes the graphs in place instead of cloning them. The coordinator computes interrupted-turn closers and constructs the exact unpublished Session once. Its immutable header and balanced logical event log form the `SessionInspection` borrowed by readers, while the revision remains internal to persistence. - -`inspect(id, signal?)` does not mutate storage. Synthetic closers exist only in the prepared in-memory view, and a torn physical tail remains untouched. Same-id callers share an in-flight cold read. Once ready, the preparation may remain in a per-coordinator LRU whose capacity defaults to five and is configurable by first-party backends. Before reusing a retained source, the coordinator reads that id's current revision; a mismatch evicts a ready source and repeats the cold materialization. A source already committing or reserved for resume remains exclusively owned, so concurrent inspection borrows that immutable view until publication or release. - -`prepare(id, signal?)` exclusively reserves the prepared Session. It confirms the retained revision before committing any torn-tail and interrupted-turn repair, establishes the durable cursor, then returns a disposable preparation. A stale source is discarded and reloaded instead of being repaired or published. A successful repair also discards the pre-repair source and materializes the committed log again before reservation, so a newer revision is never associated with an older event graph. Another same-id preparation waits until the reservation is published or released. Publication accepts only the exact reserved Session and attaches the committed cursor without rebuilding its history. Failed setup or cancellation returns an unchanged unpublished Session to the LRU; mutation or attachment consumes the reservation. - -The legacy `load(id)` API uses the same preparation and repair machinery, then discards its reservation and returns the immutable logical view. It remains a compatibility API, not the history-to-resume reuse path. This lifecycle extends the [shared persistence coordinator](2026-06-18-shared-persistence-write-coordinator.md) while preserving the storage and recovery rules owned by the [session persistence decision](2026-06-14-session-persistence.md). - -## History and resume reuse - -History reads use `inspect()`, so repeated pages borrow the same immutable prepared state without activating an Agent. A later resume uses `prepare()` and receives the exact Session retained by inspection; it does not read, decompress, parse, clone, validate, or freeze the complete log again. - -If the durable log changes after inspection, its revision changes. The next history read or resume discards a retained ready Session and materializes the new log, so an old event graph cannot be associated with a newer snapshot revision. A source already claimed by an in-flight resume is not evicted: its exclusive owner keeps it through publication or release, and concurrent history may borrow the same immutable view. - -Cold continuable-subagent access follows the same path. Descriptor authorization first inspects the child, then `ctx.agents.resume()` reserves and publishes the retained Session. This preserves the lifecycle and authorization rules in the [continuable subagent conversation decision](../feature/2026-07-28-continuable-subagent-conversations.md) while removing its duplicate cold read. +This note originally also gave persistence a `prepare(id)`/`inspect(id)` lifecycle: a coordinator-backed bounded LRU of cold unpublished Sessions with exclusive reservations, revision-checked reuse, and repair committed inside `prepare`/`load`, so history pagination and a later resume shared one cold materialization. The [handle-based persistence seam](2026-08-27-handle-based-session-persistence.md) deletes all of it: persistence exposes handles only, resume reads the log through its write handle and owns repair, and read-only observers (session-query) own their cold-Session cache keyed by the `stat().revision` change token. The read-reuse goal survives in that cache; the exclusive-reservation machinery does not, because the write handle's single-writer ownership is the exclusion resume actually needs. Resume pays one whole-log read through the handle where the prepared cache sometimes served a warm Session — an accepted cost recorded in the handle note. ## Boundaries -- `readFrom()` remains a detached physical-suffix API. It neither creates nor consumes a preparation, synthesizes logical closers, or joins the LRU. -- HMR adoption keeps the live Session authoritative and reads the stored prefix directly. It may truncate a torn physical fragment but never closes the live open turn as interrupted. -- The cache belongs to one persistence coordinator, not a process-global Session map. Live Sessions are owned by the existing stores and never occupy preparation capacity. -- A fresh create never claims a cold persisted preparation with the same id. Persistence collisions continue to reject. -- Third-party persistence implementations retain the abstract `prepare()` fallback through `load()`. They receive the same publication interface but gain exact-object reuse only when they override preparation. -- Revision validation establishes freshness at the reuse and repair-commit points; it does not add cross-process writer exclusion to a backend. Retries converge after the durable log remains unchanged for one read/check round trip, so continuous external writers can delay preparation. +- The preparation is one disposable ownership window, not a cache: disposal is synchronous and idempotent, and publication accepts only the exact prepared Session. +- A fresh create never claims a persisted identity implicitly. Persistence collisions continue to reject (`SessionAlreadyExistsError`, `SessionAlreadyOwnedError`). +- Live Sessions are owned by the existing stores; preparations hold only unpublished ones. ## Verification -The shared persistence contract pins non-mutating balanced cold inspection and later repair. `persistence.spec.ts` and `preparations.spec.ts` pin same-id in-flight sharing, exact Session reuse across inspect and prepare, revision-triggered refresh before history and resume, single repair commit, exclusive reservation, release after failed setup, ready-entry LRU eviction, append rejection during reservation, and publication of only the reserved Session. Backend tests pin that full and lightweight reads use the same revision identity. Agent-loop and continuable-subagent tests pin the common publication pipeline and inspection-to-resume path across cancellation and teardown. +Agent-loop tests pin the common publication pipeline across create, `createAgent`, and resume, including rollback on setup failure, cancellation, and teardown, and that disposal releases the write handle (reopening for write succeeds). Session-store tests pin the restoration branch's validate-and-freeze-in-place transfer. ## Alternatives considered -**Activate an Agent for history reads.** Rejected because pagination would keep query-only Agents live and transfer cache retirement into the Agent lifecycle. +**Activate an Agent for history reads.** Rejected because pagination would keep query-only Agents live and transfer cache retirement into the Agent lifecycle. This rationale still guards the session-query cold cache: observation never creates an Agent. -**Cache only `{ meta, events }`.** Rejected because resume would still reconstruct, validate, freeze, and copy a Session from the cached values. The exact unpublished Session is the reusable unit. +**Cache only `{ meta, events }`.** Rejected at the time because resume would still reconstruct a Session from the cached values. Under the handle seam this is exactly what the read side does — session-query caches a cold Session per revision for reads only — while resume rebuilds from the handle read, trading the warm-Session reuse for a single write-ownership door. -**Keep a process-global Session map.** Rejected because it would cross backend and runtime ownership boundaries, retain unbounded identities, and duplicate the live Session store. - -**Add a restore transaction or coordinator to the Agent loop.** Rejected because cold reading, repair, reservation, and cursor attachment are persistence and Session concerns. The Agent loop only needs the uniform `SessionPreparation` ownership boundary. - -**Turn `readFrom()` into logical preparation.** Rejected because watermark consumers need a detached physical suffix and, on seek-capable backends, a bounded read. Recovery balancing and whole-Session reuse have different semantics. +**Add a restore transaction or coordinator to the Agent loop.** Rejected because cold reading and Session construction are persistence and Session concerns. The Agent loop only needs the uniform `SessionPreparation` ownership boundary; the handle seam kept that split while moving repair to the loop's resume path. ## Consequences -One cold materialization can serve history pagination, subagent descriptor inspection, and a later resume. Ownership transfer removes redundant restoration clones, while the bounded per-coordinator LRU limits memory and avoids creating live Agents for queries. Create and resume share one publication protocol without merging Agent and Session responsibilities. - -The first cold inspection now pays the complete validation and Session-construction cost and may retain that unpublished Session until eviction. Persistence must coordinate reservation, append, repair, and publication, and callers must treat inspection values as immutable borrowed state. Backends that rely on the default `prepare()` remain correct but do not receive the reuse optimization. +Create and resume share one publication protocol without merging Agent and Session responsibilities, and every exit path disposes exactly one preparation. The persistence-side reuse consequences originally recorded here (shared cold materialization, LRU bounds, reservation coordination) now belong to the [handle note](2026-08-27-handle-based-session-persistence.md) and the session-query cache that replaced them. diff --git a/.agents/notes/implemented/architecture/2026-08-05-session-preparation.zh.md b/.agents/notes/implemented/architecture/2026-08-05-session-preparation.zh.md index cd918d126b..44b609d488 100644 --- a/.agents/notes/implemented/architecture/2026-08-05-session-preparation.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-05-session-preparation.zh.md @@ -6,63 +6,40 @@ Status: implemented ## 问题 -冷历史检查和 agent（智能体）恢复会分别实体化同一份持久会话日志。对于大型压缩日志，每次操作都会重新完整读取、解压、解析、验证、冻结并构造 Session。因此，历史分页可能反复承担冷读成本；如果改为由历史查询激活 agent，读取生命周期又会与缺少自然退出时机的实时 agent 耦合。 +新建和持久化恢复通过不同构造流程抵达相同的发布边界。这使一项关键不变量不够清楚：设置必须基于一个未发布的 Session 完成，之后系统才能同时公开这个精确 Session 及其 agent。 -新建和持久化恢复也通过不同构造流程抵达相同的发布边界。这使一项关键不变量不够清楚：设置必须基于一个未发布的 Session 完成，之后系统才能同时公开这个精确 Session 及其 agent。 +冷历史检查和 agent（智能体）恢复也曾分别实体化同一份持久会话日志，本 Note 最初以持久化侧的已准备 Session 缓存回答了这一半问题；那一半已在下文中被取代。 ## 决策 -`SessionPreparation` 持有一个精确的未发布 `Session`，直至发布或回滚。它属于 Session 生命周期，不属于 agent 生命周期或激活机制。新建流程包装 `SessionStore.prepare()` 的结果；持久化恢复则从 `SessionPersistence.prepare()` 取得准备对象。 +`SessionPreparation` 持有一个精确的未发布 `Session`，直至发布或回滚。它属于 Session 生命周期，不属于 agent 生命周期或激活机制。新建流程包装 `SessionStore.prepare()` 的结果；持久化恢复通过该会话的写句柄读取已存储的日志、追加 `interruptedTurnClosers`，再包装 `SessionStore.prepare(id, { seed, meta, seedSource: 'persistence' })`——即就地验证并冻结转移对象图的恢复分支。 agent loop（智能体循环）通过同一条设置与发布流水线消费这两种形式：先取得准备对象，围绕 `preparation.session` 构建私有 agent 上下文，等待可选设置完成，再发布该精确 Session 和 agent，并在所有退出路径上对准备对象执行 dispose（资源释放）。发布后，实时生命周期由现有 Session 与 agent 存储接管；`SessionPreparation` 本身不负责任何 agent 行为。 该机制细化了 [agent 生命周期与所有权决策](2026-06-18-agent-lifecycle-and-ownership-contracts.zh.md)中的发布边界，但不替换其所有权模型。 -## 持久化准备生命周期 +## 已被取代：持久化侧的准备生命周期 -使用协调器的持久化实现会将一个冷源加载为准备完成的 Session。后端转移新鲜、彼此无别名的元数据和事件，以及标识这些精确值的来源限定 revision；Session 恢复路径直接验证并冻结这些对象图，不再复制。协调器计算中断轮次的 closer，并且只构造一次精确的未发布 Session。其不可变 header 与已配平的逻辑事件日志构成读取方借用的 `SessionInspection`，revision 则保留在持久化内部。 - -`inspect(id, signal?)` 不修改存储。合成 closer 只存在于准备完成的内存视图中，撕裂的物理尾部保持不变。同 id 调用方共享进行中的冷读。准备完成后，该对象可以进入每个协调器自己的 LRU；第一方后端可配置容量，默认保留五个。协调器复用保留源之前会读取该 id 的当前 revision；如果不匹配，就淘汰处于就绪阶段的源并重新完成冷实体化。已经进入提交或为恢复而预留的源仍由其所有者独占，因此并发检查会借用该不可变视图，直至发布或释放。 - -`prepare(id, signal?)` 独占预留准备完成的 Session。它先确认保留的 revision，再提交撕裂尾部和中断轮次修复、建立持久游标，最后返回可 dispose 的准备对象。陈旧源会被丢弃并重新读取，不会参与修复或发布。修复成功后也会丢弃修复前的源，并在预留前重新实体化已提交日志，以免把较新的 revision 关联到较旧的事件对象图。同 id 的另一个准备请求会等待当前预留发布或释放。发布只接受精确的预留 Session，并直接附接已提交游标，无需重建历史。设置失败或取消时，未发生变化的未发布 Session 会返回 LRU；发生变更或完成附接后，系统会消费该预留。 - -存量 `load(id)` API 使用相同的准备和修复机制，随后丢弃其预留并返回不可变逻辑视图。它保留为兼容 API，不承担历史到恢复的复用路径。该生命周期扩展了[共享持久化协调器](2026-06-18-shared-persistence-write-coordinator.zh.md)，同时继续遵循[会话持久化决策](2026-06-14-session-persistence.zh.md)所规定的存储与恢复规则。 - -## 历史与恢复复用 - -历史读取使用 `inspect()`，因此重复分页可以借用同一份不可变准备状态，而不会激活 agent。后续恢复调用 `prepare()`，直接取得检查阶段保留的精确 Session；系统不会再次完整读取、解压、解析、复制、验证或冻结日志。 - -如果持久日志在检查后发生变化，其 revision 也会变化。下一次历史读取或恢复会丢弃保留且处于就绪阶段的 Session，并实体化新日志，因此旧事件对象图不会被关联到较新的快照 revision。已经由进行中恢复操作取得的源不会被淘汰：其独占所有者会持有它直至发布或释放，并发历史读取可以借用同一个不可变视图。 - -冷 continuable subagent 访问沿用同一路径。系统先检查子会话并完成 descriptor 授权，再由 `ctx.agents.resume()` 预留并发布保留的 Session。这样既遵循 [continuable subagent 会话决策](../feature/2026-07-28-continuable-subagent-conversations.zh.md)中的生命周期与授权规则，也消除了重复冷读。 +本 Note 最初还赋予持久化一个 `prepare(id)`/`inspect(id)` 生命周期：由协调器支撑的、装有冷未发布 Session 的有界 LRU，带独占预留、按 revision 校验的复用，以及在 `prepare`/`load` 内部提交的修复，使历史分页与后续恢复共享一次冷实体化。[基于句柄的持久化 seam](2026-08-27-handle-based-session-persistence.zh.md) 删除了这一切：持久化只暴露句柄，恢复通过其写句柄读取日志并自行负责修复，只读观察方（session-query）拥有自己的冷 Session 缓存，以 `stat().revision` 变更令牌为键。读取复用的目标在该缓存中得以延续；独占预留机制则没有延续，因为写句柄的单写者所有权正是恢复真正需要的排他手段。在已准备缓存有时能提供温 Session 的场景下，恢复要为通过句柄的一次全日志读取付出代价——这是句柄 Note 中记录的、已被接受的成本。 ## 边界 -- `readFrom()` 仍是脱离的物理后缀 API。它不会创建或消费准备对象，不会合成逻辑 closer，也不会进入 LRU。 -- HMR（热模块替换）接管继续以实时 Session 为权威，并直接读取已存储前缀。它可以截断撕裂的物理碎片，但绝不把实时开放轮次关闭为中断状态。 -- 缓存属于单个持久化协调器，而不是进程全局 Session map。实时 Session 由现有存储持有，绝不占用准备容量。 -- 新建流程绝不认领相同 id 的冷持久化准备对象。持久化冲突仍会被拒绝。 -- 第三方持久化实现继续获得通过 `load()` 实现的抽象 `prepare()` 回退。它们使用相同发布接口，但只有覆盖准备流程后才能复用精确对象。 -- Revision 校验在复用点和修复提交点建立新鲜度，但不会为后端增加跨进程 writer 排他。持久日志在一次读取与复核往返内保持不变后，重试才能收敛，因此持续的外部写入可能延迟准备。 +- 准备对象是一个可 dispose 的所有权窗口，而不是缓存：dispose 同步且幂等，发布只接受精确的已准备 Session。 +- 新建流程绝不隐式认领持久化身份。持久化冲突仍会被拒绝（`SessionAlreadyExistsError`、`SessionAlreadyOwnedError`）。 +- 实时 Session 由现有存储持有；准备对象只持有未发布的 Session。 ## 验证 -共享持久化约定规定冷检查不得修改存储且须保持配平，并覆盖后续修复。`persistence.spec.ts` 与 `preparations.spec.ts` 覆盖同 id 进行中读取共享、检查与准备之间的精确 Session 复用、在历史读取与恢复前由 revision 触发刷新、修复只提交一次、独占预留、设置失败后释放、就绪项 LRU 淘汰、预留期间拒绝 append，以及只允许发布预留 Session。后端测试覆盖完整读取与轻量读取使用同一 revision 身份。agent loop 与 continuable subagent 测试覆盖统一发布流水线，以及取消和清理期间从检查到恢复的路径。 +agent loop 测试覆盖 create、`createAgent` 与 resume 之间的统一发布流水线，包括设置失败时的回滚、取消与清理，以及 dispose 会释放写句柄（重新以写模式打开可以成功）。Session store 测试覆盖恢复分支的就地验证并冻结的所有权转移。 ## 考虑过的替代方案 -**由历史读取激活 agent。** 不采用，因为分页会使仅用于查询的 agent 长期保持实时状态，并把缓存退出问题转移到 agent 生命周期。 +**由历史读取激活 agent。** 不采用，因为分页会使仅用于查询的 agent 长期保持实时状态，并把缓存退出问题转移到 agent 生命周期。该理由仍然守护着 session-query 冷缓存：观察绝不创建 agent。 -**只缓存 `{ meta, events }`。** 不采用，因为恢复仍需从缓存值重新构造、验证、冻结并复制 Session。真正可复用的单元是精确的未发布 Session。 +**只缓存 `{ meta, events }`。** 当时不采用，因为恢复仍需从缓存值重新构造 Session。在句柄 seam 下，这恰好是读取侧的做法——session-query 按 revision 为只读用途缓存一个冷 Session——而恢复则从句柄读取重建，以温 Session 复用换取唯一的写所有权之门。 -**维护进程全局 Session map。** 不采用，因为它会跨越后端和运行时所有权边界，无界保留身份，并与实时 Session 存储重复。 - -**在 agent loop 中增加恢复事务或协调器。** 不采用，因为冷读、修复、预留和游标附接都属于持久化与 Session 职责。agent loop 只需要统一的 `SessionPreparation` 所有权边界。 - -**把 `readFrom()` 改成逻辑准备流程。** 不采用，因为水位消费方需要脱离的物理后缀；对于可寻址后端，还需要限制实际读取范围。恢复平衡与完整 Session 复用具有不同语义。 +**在 agent loop 中增加恢复事务或协调器。** 不采用，因为冷读与 Session 构造属于持久化与 Session 职责。agent loop 只需要统一的 `SessionPreparation` 所有权边界；句柄 seam 保留了这一分工，同时把修复移入循环的恢复路径。 ## 后果 -一次冷实体化可以同时服务历史分页、subagent descriptor 检查和后续恢复。所有权转移去除了恢复阶段的冗余复制；每个协调器的有界 LRU 限制内存占用，也避免查询创建实时 agent。新建和恢复共享同一发布协议，同时保持 agent 与 Session 职责分离。 - -首次冷检查需要承担完整验证与 Session 构造成本，并可能保留该未发布 Session 直至淘汰。持久化层必须协调预留、append、修复和发布；调用方必须把检查结果视为借用的不可变状态。依赖默认 `prepare()` 的后端仍然正确，但无法获得复用优化。 +新建和恢复共享同一发布协议，同时保持 agent 与 Session 职责分离，且每条退出路径恰好 dispose 一个准备对象。本 Note 最初记录的持久化侧复用后果（共享冷实体化、LRU 上限、预留协调）如今归属于[句柄 Note](2026-08-27-handle-based-session-persistence.zh.md) 以及取代它们的 session-query 缓存。 diff --git a/.agents/notes/implemented/architecture/2026-08-06-subagent-list-identity-projection.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-06-subagent-list-identity-projection.i18n.yaml index 29613b58c7..cfdc3bd607 100644 --- a/.agents/notes/implemented/architecture/2026-08-06-subagent-list-identity-projection.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-06-subagent-list-identity-projection.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-06-subagent-list-identity-projection.md -2026-08-06-subagent-list-identity-projection.md: aeed828530f615b1bb4958a360b5ba4db543f714 -2026-08-06-subagent-list-identity-projection.zh.md: b2b64eaa7c06b738734a7b975adb5948704465bd +2026-08-06-subagent-list-identity-projection.md: cbb15696314930acfaf20ba8651699c53c5dbde2 +2026-08-06-subagent-list-identity-projection.zh.md: dbb62dbfc6bb6ca3ab63504de1ba8dd35327bbbf diff --git a/.agents/notes/implemented/architecture/2026-08-06-subagent-list-identity-projection.md b/.agents/notes/implemented/architecture/2026-08-06-subagent-list-identity-projection.md index aeed828530..cbb1569631 100644 --- a/.agents/notes/implemented/architecture/2026-08-06-subagent-list-identity-projection.md +++ b/.agents/notes/implemented/architecture/2026-08-06-subagent-list-identity-projection.md @@ -14,16 +14,16 @@ The root cause is that the [durable-subagent-catalog decision](../feature/2026-0 ## Decision -mode and label are folded by the new `subagent` projection unit (pure identity, two arms), and the unit is the sole authority over the fold rules; `listChildren` no longer depends on session-query — enumeration is a subagent-owned live-preferred merge, and value retrieval walks a three-rung compute-and-discard ladder: a live child synchronously reads the registry's existing watermark cache (zero log reads); a cold child first asks the optional `sessionProjectionCache` checkpoint, and a served identity that passes the seq gate is final; otherwise it pays one full `persistence.inspect` read plus a fold through the registered `subagent` unit. No index, no cache of its own, no write-back. +mode and label are folded by the `subagent` projection unit (pure identity, two arms), and the unit is the sole authority over the fold rules. Enumeration uses the shared Session query corpus, while value retrieval walks a three-rung compute-and-discard ladder: a live child synchronously reads the registry's existing watermark cache (zero log reads); an unseeded cold child may use the optional `sessionProjectionCache` checkpoint because its exact inherited cut is known to be zero; every seeded child and every cache miss pays one body-bearing Session observation plus a fold through the registered `subagent` unit. No index, no cache of its own, no list-side write-back. There are three families of escape from the per-child scan: promote mode/label into the header (the write path pays); build a durable derivation for the projection (a checkpoint ladder, or values landed during query-index rebuild with read-side reconciliation); or compute at read time (live from the watermark cache, cold from one full read). This note takes the third. "Values landed with the query index" was retired wholesale: query infrastructure was forced to learn domain vocabulary while the sole consumer is satisfied by read-time computation — the live child's zero reads come for free from session-projection's existing watermark cache, and the cold child's single full read is explicitly accepted as compute-and-discard. The first two routes and the retirement rationale are detailed under Alternatives considered. Key points: -- **The subagent list does not depend on session-query**: enumeration is completed by a subagent-owned live-preferred merge, and mode/label is retrieved through `ctx.sessionProjections`; deployments without a query backend list as usual. -- **Value retrieval is a three-rung compute-and-discard ladder**: a live child reads `sessionProjections.snapshot(session, ['subagent'])` (the registry's existing watermark cache, zero log reads); a cold child first reads the optional `sessionProjectionCache.cachedSnapshot(header, ['subagent'])`, using the non-null identity directly when it passes the seq gate (`seq >= seedLength ?? 0`); otherwise it pays one full Session observation plus a fold through the registered `subagent` unit; beyond that, absent is absent — no cache of its own, no write-back, no index. +- **The subagent list uses the Session query corpus for enumeration and body-bearing observations**: mode/label still comes through `ctx.sessionProjections`, and the list owns no descriptor parser or domain index. +- **Value retrieval is a three-rung compute-and-discard ladder**: a live child reads `sessionProjections.snapshot(session, ['subagent'])` (the registry's existing watermark cache, zero log reads); an unseeded cold child may read `sessionProjectionCache.cachedSnapshot(header, SessionLogOffset(0), ['subagent'])`; a seeded child or cache miss pays one Session observation carrying `inheritedEventCount` plus a fold through the registered `subagent` unit. Beyond that, absent is absent — no cache of its own, no list-side write-back, no index. - **The `subagent` projection unit is the sole authority over the fold rules**: live and cold snapshots both run the one registered unit; no second copy of descriptor-interpretation logic exists. -- **The header, the descriptor (v2), session-persistence, session-projection(-cache), and session-query(-sqlite) are all untouched**; pre-existing data acquires exact values through one `inspect` computation the first time it is listed — no degraded unknown state, no migration. +- **The descriptor (v2) remains untouched**. Session, persistence, projection cache, and query now carry the exact inherited cut separately from the logical header; pre-existing data acquires exact values through one body-bearing observation when listing cannot prove a zero cut — no degraded unknown state and no durable format migration. Relationship to existing notes: @@ -36,8 +36,8 @@ It hangs beside the existing `subagentTiming` ([projection.ts](../../../../packa ```ts ignore-check export type SubagentIdentityProjection = - | { mode: 'one-shot'; label?: string; seq: number } - | { mode: 'continuable'; label: string; seq: number } + | { mode: 'one-shot'; label?: string; seq: SessionSeq } + | { mode: 'continuable'; label: string; seq: SessionSeq } declare module '@deepseek-ai/dsh-session-projection/types' { interface SessionProjectionStateMap { @@ -52,19 +52,19 @@ declare module '@deepseek-ai/dsh-session-projection/types' { - The projection is pure identity, and **the projection system has no failure channel**: a unit never throws; a corrupt payload or an unrecognized version folds exactly like a log with no descriptor at all. The host checkpoint state is the serializable wrapper `{ identity?: SubagentIdentityProjection }`; absence is `{}`. Its client view is the non-optional `SubagentIdentityProjection | null` entry. `null` passes JSON losslessly, so a pushed reset replaces a stale identity instead of being dropped by stringify. The judging discipline: consuming surfaces treat null and an absent client key alike as no value. How "computed to nothing" is presented is the consumer's own business (see the `listChildren` four-state mapping below). - Label strength is decided by the descriptor schema: a continuable's label is mandatory at parse, a one-shot's was always optional; the mode/label discriminant matches the child row's strong contract below exactly (the row carries no `seq` — it is the projection's internal own-suffix proof). -- The identity carries `seq`: the seq of the `subagent/descriptor` event it was folded from, mandatory on both arms and absent on the null sentinel — `seq >= header.seedLength ?? 0` proves the identity was folded from the child's own suffix rather than a fork seed's replayed ancestor descriptor. The unit maps the wrapper's validated identity to its client wire view and is checkpointed like every unit (the `persist` opt-in is gone); its `stateVersion` is 2, bumped when `seq` was added. Existing older checkpoint rows are invalidated by version mismatch per the registry contract, falling to the authoritative refold. +- The identity carries branded `seq`: the seq of the `subagent/descriptor` event it was folded from, mandatory on both arms and absent on the null sentinel. A live Session checks it through `isOwnSeq()`; a cold body-bearing observation compares it with `inheritedEventCount`. Header-only seeded candidates skip the cache because the header intentionally exposes no integer cut; unseeded candidates know the cut is zero. The unit maps the wrapper's validated identity to its client wire view and is checkpointed like every unit (the `persist` opt-in is gone); its `stateVersion` is 2, bumped when `seq` was added. Existing older checkpoint rows are invalidated by version mismatch per the registry contract, falling to the authoritative refold. - Fold rule: `subagent/descriptor` is last-wins, under the same descriptor-reset discipline as `subagentTiming` — ancestor descriptors in the fork prefix are overridden by the session's own descriptor. A corrupt or unrecognized-version payload is last-wins all the same: it resets to the null sentinel rather than keeping the prior identity, so a fork of a healthy ancestor does not inherit an identity its own descriptor cannot stand up. -### Enumeration: subagent-owned live-preferred merge +### Enumeration: query corpus with live preference -`listChildren`'s ([list-children.ts](../../../../packages/subagent/subagent/src/list-children.ts)) enumeration goes through no query service: the two sources `ctx.sessions.list()` and `ctx.get('sessionPersistence')?.list()` merge by id, with a live record overriding the same-id persisted record wholesale and no header consistency check. Everything enumeration needs is header facts: +`listChildren` ([list-children.ts](../../../../packages/subagent/subagent/src/list-children.ts)) asks `sessionQuery.listSessions()` for the canonical live-preferred corpus, then pairs each listed id with `ctx.sessions.get(id)` when a live Session exists. The live header overrides the listed header for that id. Everything enumeration needs is header facts: - Filtering: `header.origin === 'subagent' && header.parentSession === parentSessionId`. - `hasChildren`: the same merged material, looked at one level down — a direct descendant exists with `origin === 'subagent'` whose `parentSession` is that child. - `activity`: a live record is `running`; one present only in persistence is `inactive`. - Ordering: `createdAt` ascending, then child id ascending (matching the old contract). -- **Absent persistence degrades to live-only enumeration, not an error**: in a deployment without persistence, a cold child could not be resumed anyway, and listing live children remains meaningful. (Contrast: the old implementation rejected wholesale when sessionQuery was missing.) -- A persistence listing failure fails the whole enumeration; per-child isolation applies only to the per-child cold reads. +- An absent `sessionQuery` service fails with `SUBAGENT_CONTROL_QUERY_UNAVAILABLE`; the shared query corpus owns whether a deployment can enumerate live-only or persisted Sessions. +- A query-corpus failure fails the whole enumeration; per-child isolation applies only to per-child cold observations. ### Value retrieval: the three-rung compute-and-discard ladder @@ -73,20 +73,20 @@ For each enumerated child, mode/label retrieval walks a three-rung ladder — co | Rung | Read | Cost | | --- | --- | --- | | 1: live child | `ctx.sessionProjections.snapshot(session, ['subagent'])` | Zero log reads — the registry's existing watermark cache, synchronous retrieval | -| 2: cold child, cache hit | The optional `sessionProjectionCache.cachedSnapshot(header, ['subagent'])`, used directly only when the non-null identity satisfies `identity.seq >= header.seedLength ?? 0` — an own descriptor is immutable once appended, and the seq gate proves the value was folded from the child's own suffix, regardless of the row's watermark | Zero log reads | -| 3: cold child, fallback | One full `persistence.inspect(id)` read + a fold through the registered `subagent` unit | One full read computed per listing | +| 2: unseeded cold child, cache hit | The optional `sessionProjectionCache.cachedSnapshot(header, SessionLogOffset(0), ['subagent'])`; every valid seq is owned when the exact cut is zero | Zero log reads | +| 3: seeded child or cold fallback | One body-bearing `sessionQuery.observeSession(id)` + the registered `subagent` projection, with `inheritedEventCount` available for the own-suffix check | One full read computed per listing | -- Error contract: `sessionProjections` is a required injection — `SubagentRuntime` declares it in its inject set, so a deployment without the registry never activates the service (or the loop), and `listChildren` is unreachable rather than served degraded rows ([mandatory-seam note](2026-08-19-session-projection-mandatory-seam.md)); the loud runtime check and `SUBAGENT_CONTROL_PROJECTIONS_UNAVAILABLE` are deleted with it. The session store keeps the explicit posture: an absent `ctx.get('sessions')` (a strict global read, never the caller-scope-bound property proxy) fails with `SUBAGENT_CONTROL_SESSION_STORE_UNAVAILABLE`. apiproxy's dedicated `PROJECTIONS_UNAVAILABLE` wire face is deleted along with the code; `SESSION_STORE_UNAVAILABLE` goes through the generic internal fallback — apiproxy's composition injects `sessions` itself, so that error is unreachable in its deployment, and a dedicated mapping would violate the need principle. `SUBAGENT_CONTROL_SESSION_QUERY_UNAVAILABLE` is deleted along with the session-query dependency. -- The cache is a purely optional acceleration layer: an absent service is skipped on a null check — no error code, no part in configuration validation (in contrast to `sessionProjections`, a required injection). Anything the second rung throws (including a poisoned unit row in the cache detonating `viewCheckpoint`) silently falls to the third rung — the cache is derived data, so its faults never produce a `corrupt` verdict; the final judgment belongs to the authoritative refold. A row whose checkpoint cut predates the descriptor naturally lacks the `subagent` key and falls through automatically, with no special-casing; a null sentinel in the row does not count either — it falls to the third rung for the authoritative refold's verdict. A count/interval checkpoint inside the creation window can land a fork seed's replayed ancestor identity in the row — the ancestor's seq falls inside the seed range, the seq gate rejects it, and it likewise falls to the third rung's verdict. +- Error contract: `sessionProjections`, the Session store, and `sessionQuery` are required runtime services for listing. Their explicit failures are `SUBAGENT_CONTROL_PROJECTIONS_UNAVAILABLE`, `SUBAGENT_CONTROL_SESSION_STORE_UNAVAILABLE`, and `SUBAGENT_CONTROL_QUERY_UNAVAILABLE`; no empty result disguises a missing classification or corpus capability. +- The cache is a purely optional acceleration layer: an absent service is skipped on a null check — no error code, no part in configuration validation (in contrast to `sessionProjections`, a required injection). A seeded header skips this rung because it cannot supply the cache identity's exact cut without a body read. For an unseeded child, anything the second rung throws (including a poisoned unit row detonating `viewCheckpoint`) silently falls to the third rung — the cache is derived data, so its faults never produce a `corrupt` verdict; the final judgment belongs to the authoritative refold. A row whose checkpoint cut predates the descriptor, an absent key, or a null sentinel likewise falls through. - Per-child isolation: a single child's failed cold full read only turns that row into an `unavailable` diagnostic, naturally retried on the next listing, without affecting siblings (see the four-state mapping). -- The cold path's lifecycle witness: preparation's result must still point at the lifecycle that was enumerated — the witness field set is the same seven fields as the old SOURCE_CONFLICT check (version, id, createdAt, cwd, parentSession, seedLength, delegationDepth); a session deleted and republished under the same id degrades to a `corrupt` row in the old parent's catalog, leaking nothing of the new owner's child. +- The cold path's lifecycle witness: the observation must still point at the lifecycle that was enumerated. The witness fields are version, id, createdAt, cwd, parentSession, isSeeded, delegationDepth, origin, and agentPreset; a Session deleted and republished under the same id degrades to a `corrupt` row in the old parent's catalog, leaking nothing of the new owner's child. - Cold-read concurrency is bounded by the constant 4 — it constrains a read-only scan of local media, not deployment behavior; when a networked persistence backend appears, it is promoted to a validated `Config` field. -- The cold-read cost, recorded honestly: only with the cache unmounted or missed does a cold child pay one full read per listing, at a cost proportional to its transcript size; the settled stance is compute-and-discard, and no cache of its own is built. The full read goes through `inspect()` into the [Session preparation](2026-08-05-session-preparation.md) cold read, so short-term repeated reads of the same id can hit its LRU for reuse, but listing does not depend on this. A live child reads zero log throughout. +- The cold-read cost, recorded honestly: every seeded child and every unseeded cache miss pays one full query observation per listing, at a cost proportional to its transcript size; the settled stance is compute-and-discard, and no cache of its own is built. The observation may reuse the query/persistence preparation layer, but listing does not depend on that optimization. A live child reads zero log throughout. - Cancellation: the caller's signal is checked before and after each persistence read, and a read that settles only after abort is rejected, normalized to the stable error code `CANCELLED`. ### Authority model -- The session log is the sole authority; this design adds no derived persistence of any kind — no index values, no checkpoints of its own, no in-process memo; the `sessionProjectionCache` checkpoint the second rung reads is an existing composition item's derived data, which this design only reads and never writes. Values are computed on read and discarded, and a value's freshness is exactly the live state or persisted revision at the moment of the read (an own descriptor is immutable once appended — a cached identity past the seq gate has no staleness problem; the gate guards against seed-replayed ancestor identities). +- The session log is the sole authority; this design adds no domain index, checkpoint of its own, or in-process memo. The `sessionProjectionCache` checkpoint the second rung reads is an existing composition item's derived data, which the list only reads. Values are computed on read and discarded. Seeded candidates use a body-bearing observation to classify the identity against the exact cut; unseeded cached identities need no seq gate because every valid seq is owned. - The Session and persistence write paths are entirely unaware of listing and projection consumption: no event-listener write-back, no fold-on-write. - Enumeration and value retrieval constitute no second authorization source and make no unpublished child visible — the two sources see only published live records and durably written persisted records, consistent with the rule the durable-subagent-catalog note laid down for derived read surfaces. @@ -127,25 +127,24 @@ For each enumerated child, the ladder's result maps to a row through four states Known boundary deviations (deliberately accepted, recorded with this note): -- A fork child that died in its publication window, with an ancestor descriptor in its seed, gets the ancestor identity from last-wins and wrongly surfaces as a child row; resume still fails against the own-suffix fold authority (`NOT_RESUMABLE`). The old implementation omitted it via `seedLength` filtering; the projection unit cannot see the header, and this debris-grade deviation is accepted (`subagentTiming` has the same kind of pre-existing exposure). - Multiple descriptors in the own suffix: the old implementation judged corrupt; last-wins now takes the final one (the provider contract guarantees exactly one anyway). - A live/persisted header conflict: the old implementation made it per-child corrupt; enumeration now prefers live with no consistency check, the conflict goes unnoticed, and the live record forms the row. - A source-read failure on damaged storage (e.g. a bad surface rejected by the cold full read): the old implementation mapped it to per-child `corrupt`; it is now uniformly an `unavailable` row (the read side cannot tell the causes apart). - An unknown parent: the old implementation threw not-found through session-query ('parent session … was not found'); the subagent-owned merge now yields an empty subset for a nonexistent parent, enumeration returns an empty list, and later operations on the wire land as child-level subagent-not-found — a silent change of semantics and wording, recorded as explicitly accepted. -- Rung 2's later-event window: a cache row lands right after the first own descriptor, the log then appends a second own descriptor (or a malformed payload setting the null sentinel), and the process crashes before the next checkpoint — from then on a cold listing's rung 2, admitted by the seq≥seedLength gate, keeps serving the row's old identity (the first own descriptor's value), diverging from the authoritative refold (last-wins, the second), and a rung-2 hit triggers no refold, so nothing notices. Three boundaries: ① the precondition is a second own descriptor on the same child, violating the establishing provider's append-exactly-once contract — corruption-class data, same family and source as the multi-descriptor deviation; ② it takes both "corruption + a crash missing every checkpoint (the two mandatory points, turn/end and disposal, and the count/interval throttle points all unmet)" at once; ③ a healthy child (exactly one own descriptor) is unaffected — what the seq gate admits is precisely the only true identity. Self-healing: any live run of that child (the turn/end mandatory checkpoint) or any moment that triggers cache.write overwrites the whole row with a fresh fold (whole-record replace), and rung 2 serves correctly from then on; the authoritative paths (the rung-3 refold, the live snapshot, the resume fold) are correct from the start, and the divergence exists only in listing reads while the child stays cold and the row is never rewritten. The mechanical fixes were not taken: gate reconciliation would need the log-end seq, unavailable to a zero-read cold path; a cache row carrying the revision is an opaque token, incomparable and a cross-domain schema change — filed as accepted under the "the cache is never authoritative" doctrine. +- Rung 2's later-event window applies only to an unseeded child: a cache row lands right after the first descriptor, the log then appends a second descriptor (or a malformed payload setting the null sentinel), and the process crashes before the next checkpoint. Cold listing can keep serving the old identity until a live run or cache write replaces the row. The precondition violates the provider's append-exactly-once contract and also requires missing every mandatory checkpoint; healthy children are unaffected. Seeded children never take rung 2 without the body-owned cut. -Consuming surfaces: diagnostic handling across wire, tool, and GUI **stays entirely as it was, zero changes** (the `list_agents` description and output schema are untouched; the plugin's load requirement changes — `sessionQuery` dropped from inject, `sessionProjections` added as a required injection). The only behavioral changes are in apiproxy: on the route segment, the `hasSubagentDescriptor()` scan is deleted and `hasSubagentOwner` looks only at `header.origin` — pre-#1569 data without `origin` is no longer recognized as a subagent owner; it never entered the catalog anyway, and the pre-release stance accepts this; and `subagents.history` is aligned with `session.history`'s source — a live child served from in-memory events and the registry's watermark snapshot, a cold child from `inspectServable` reading persistence directly with a detached fold, no query service involved, the SESSION_QUERY_* error arms retired with it, and the wire shape unchanged (the `history` JSDoc wording becomes the live in-memory snapshot / cold persisted log dual arm). +Consuming surfaces keep the same row and diagnostic wire shape. `list_agents` reaches the required query corpus plus projection registry; live identities come from the registry snapshot and cold identities from cache or query observation. Host ownership still uses `header.origin`, and history uses the shared live/cold Session query sources; no consumer parses descriptor events independently. ### Change footprint | Area | Files | Change | | --- | --- | --- | | subagent | projection.ts, projection-types.ts, index.ts | New client-visible `subagent` unit and its registration | -| subagent | list-children.ts and its types | Rewritten as subagent-owned enumeration plus the projection-ladder four-state mapping; the session-query dependency, per-child event reads, and in-place classification machinery deleted; error code `SUBAGENT_CONTROL_SESSION_QUERY_UNAVAILABLE` deleted, and `sessionProjections` becomes a required injection (no projection error code remains); new optional dependency dsh-session-projection-cache (pure read acceleration, skipped when absent) | -| host/apiproxy | api-proxy.ts | `hasSubagentDescriptor` deleted; the owner check looks only at `header.origin`; `subagents.history` shares `session.history`'s source — live from in-memory events and the registry's watermark snapshot, cold from `inspectServable` reading persistence directly with a detached fold, no query service, the SESSION_QUERY_* error arms and the dedicated `PROJECTIONS_UNAVAILABLE` wire face retired with it | -| tool | tool-subagent-control/list-agents.ts | Load requirement narrowed (`sessionQuery` dropped from inject); model-visible schema, description, and rendering unchanged | +| subagent | list-children.ts and its types | Query-corpus enumeration plus the projection-ladder four-state mapping; required projections/query services and optional projection-cache acceleration | +| host/apiproxy | Session controller/query integration | Owner checks use `header.origin`; live/cold history and listing consume the shared query and projection sources | +| tool | tool-subagent-control/list-agents.ts | Model-visible schema, description, and rendering remain unchanged | | wire/client | api/subagents.ts, runtime sessions/service.ts, GUI | Types, row shape, and diagnostic handling **unchanged**; api/subagents.ts only reworded the `history` JSDoc to the dual arm | -| core/session, session-persistence, session-projection(-cache), session-query(-sqlite) | — | **Zero changes** | +| core/session, session-persistence, session-projection(-cache), session-query(-sqlite) | body-bearing cut and branded seq plumbing | Logical headers expose `isSeeded`; Session, persistence observations, cache identity, and query records carry exact `inheritedEventCount` separately | ## Alternatives considered @@ -169,20 +168,20 @@ Consuming surfaces: diagnostic handling across wire, tool, and GUI **stays entir ## Verification -`packages/subagent/subagent/tests/list-children.spec.ts` is rewritten to this contract: live-only listing without persistence, query services, or the continuation runtime; without the registry the service never activates (the mandatory seam — a `setup` variant asserting `ctx.get('subagents')` stays undefined); a live child incurs zero `inspect` throughout while a cold child incurs exactly one per listing; multiple descriptors resolve last-wins to the final one; corrupt payloads and unknown versions fold to `corrupt`; a cold-read failure maps to `unavailable` and retries on the next listing; the ancestor descriptor in a fork seed forms a row under that identity (pinning deviation one); ordinary forks and descendants without a subagent origin neither enter the list nor count toward `hasChildren`; `createdAt`-then-id ordering; an unmounted provider does not affect listing; compacted and uncompacted twins list identically; the three cases of pre-abort, persistence listing, and cold-read cancellation all normalize to `CANCELLED`; the empty list and stable error codes (`SUBAGENT_CONTROL_SESSION_STORE_UNAVAILABLE` for an absent store). Second-rung cases: an own-seq identity used directly with zero `inspect`, a fork seed's ancestor identity (seq inside the seed range) rejected by the gate and falling through, an in-row identity absence (null sentinel or absent key) falling through, an absent cache service falling through, and a poisoned cache row silently falling through to the refold; cold-path lifecycle tampering degrades to `corrupt` field by witness field (`it.each` over the seven). The `tool-subagent-control` list-agents tests are updated for the narrowed load requirement; `optional-session-query.spec.ts` is deleted with the dependency it guarded; the existing keyless snapshots (`subagent-list-agents` among others) are unchanged, pinning that the healthy path's wire and model-visible surfaces did not move; a new keyless snapshot, `subagent-diagnostic` (examples/headless-agent), pins the four-state mapping's diagnostic classification — the model-visible changes such as descriptor-less settled debris becoming a `corrupt` row. +`packages/subagent/subagent/tests/list-children.spec.ts` pins this contract: live identity checks use `Session.isOwnSeq()`; an unseeded cold identity may use the cache at cut zero; seeded candidates skip that cache rung and use an observation carrying `inheritedEventCount`; ancestor identities fail the own-suffix check; absent, null, poisoned, and unavailable cache/observation cases fall through or produce the documented diagnostic; lifecycle tampering degrades to `corrupt` across the complete witness field set. The existing keyless snapshots keep the healthy wire and model-visible surfaces fixed, while `subagent-diagnostic` pins diagnostic classification. ## Consequences - Listing a live child reads zero log throughout; with the cache unmounted or missed, a cold child pays one full `inspect` read per listing, at a cost proportional to its transcript size and repeated with listing frequency — compute-and-discard is the settled stance: no cache of its own is built, nothing is written back, and short-term repeated full reads of the same id can hit the preparation-phase LRU, though listing does not depend on it. -- The subagent list no longer requires a query backend: both pure-live and persistence-less deployments can list; `SUBAGENT_CONTROL_SESSION_QUERY_UNAVAILABLE` is gone, loading the `list_agents` plugin no longer requires `sessionQuery`, and `sessionProjections` becomes a required injection of `SubagentRuntime` — a deployment without the projection registry never activates the service (the mandatory seam). +- The subagent list requires the Session query corpus and projection registry; missing services fail explicitly instead of producing incomplete rows. The optional projection cache changes only the number of body reads. - Identity interpretation exists only in the single unit registered with the registry: the list's three-rung ladder and GUI history's cold read use its live, cached, or observed wire snapshots, and no hand-written bypass fold exists; if some future consuming surface bypasses the unit with a hand-written fold, values will drift across read faces — a discipline this design requires be maintained, not a mechanical guarantee. - Per-child isolation is back: a single child's cold-read failure loses only that row and healthy siblings are unaffected; a persistence listing failure still fails the whole enumeration. -- The diagnostic and enumeration semantics leaves six boundary deviations (a stillborn fork surfacing under its ancestor's identity, multiple descriptors resolving to the last, header conflicts going unnoticed, damaged-source read failures shifting from `corrupt` to `unavailable`, an unknown parent yielding an empty list instead of not-found, and rung 2's later-event window); the full semantics is in the known-boundary-deviations list; the first four are display or classification deviations on debris-grade data, the unknown-parent one is a silent query-semantics change, and the rung-2 window is a self-healing cache-serving divergence under the double condition of corruption plus a crash; resume authorization is unaffected throughout, all explicitly accepted. +- The diagnostic and enumeration semantics leaves five boundary deviations (multiple descriptors resolving to the last, header conflicts going unnoticed, damaged-source read failures changing classification, an unknown parent yielding an empty list instead of not-found, and the unseeded rung-2 later-event window). Seeded ancestor identities are no longer a deviation because body-bearing reads compare them with `inheritedEventCount`; resume authorization remains unaffected. - Pre-#1569 data without `origin` is no longer recognized as a subagent owner; it never entered the catalog anyway, and pre-release carries no compatibility promise. ## Related -- [Durable subagent catalog and list_agents](../feature/2026-07-22-durable-subagent-catalog-and-list-agents.md) — partially superseded by this note: the descriptor remains the durable authority for mode/label and the fold input, while the list's enumeration and value retrieval move to the subagent-owned merge plus the projection ladder. +- [Durable subagent catalog and list_agents](../feature/2026-07-22-durable-subagent-catalog-and-list-agents.md) — partially superseded by this note: the descriptor remains the durable authority for mode/label and the fold input, while value retrieval moves to the projection ladder over the shared query corpus. - [Session projections and command lifecycle logging](../../proposed/architecture/2026-07-27-session-projection-and-command-log.md) — the authority for the registry contract; this note adds the `subagent` identity unit and consumes its live and cold wire snapshots. - [Session projection state and client views](2026-08-19-session-projection-state-and-client-views.md) — the state/client split; both `subagent` and `subagentTiming` provide client wire views. - [Session projections as a required seam](2026-08-19-session-projection-mandatory-seam.md) — `sessionProjections` becomes a required injection; the list's error contract follows it (registry absence is an activation-time failure, and the projection error code is deleted). diff --git a/.agents/notes/implemented/architecture/2026-08-06-subagent-list-identity-projection.zh.md b/.agents/notes/implemented/architecture/2026-08-06-subagent-list-identity-projection.zh.md index b2b64eaa7c..dbb62dbfc6 100644 --- a/.agents/notes/implemented/architecture/2026-08-06-subagent-list-identity-projection.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-06-subagent-list-identity-projection.zh.md @@ -14,16 +14,16 @@ Status: implemented ## 决策 -mode 与 label 由新的 `subagent` projection unit（纯身份两臂）折叠，unit 是折叠规则的唯一权威；`listChildren` 不再依赖 session-query——枚举是 subagent 自管的 live-preferred 合并，取值走三级「算完即止」阶梯：live child 同步读注册表的既有水位缓存（零日志读）；cold child 先问可选的 `sessionProjectionCache` checkpoint，取到过 seq 门的身份即定值；否则一次 `persistence.inspect` 整读加经注册的 `subagent` unit 折叠。无索引、不自建缓存、无回写。 +mode 与 label 由 `subagent` projection unit（纯身份两臂）折叠，unit 是折叠规则的唯一权威。枚举使用共享 Session query corpus，取值则走三级「算完即止」阶梯：live child 同步读注册表的既有水位缓存（零日志读）；unseeded cold child 可以使用可选 `sessionProjectionCache` checkpoint，因为其精确 inherited cut 已知为零；每个 seeded child 与每次 cache miss 都执行一次含正文的 Session observation，再经注册的 `subagent` unit 折叠。无索引、不自建缓存、列表侧无回写。 消除逐 child 扫描的出路有三类：把 mode/label 提升进 header（写路承担）；为投影建持久派生（checkpoint 阶梯，或随查询索引重建落值、读端对账）；读时现算（live 走水位缓存，cold 一次整读）。本记录取第三条。「值随查询索引落库」已整体退役：查询基础设施被迫认识领域词汇，而唯一消费方读时现算即可满足——live child 的零读由 session-projection 既有水位缓存白拿，cold child 的一次整读被「算完即止」显式接受。前两条与退役理由详见考虑过的替代方案一节。 要点： -- **subagent 列表不依赖 session-query**：枚举由 subagent 自管的 live-preferred 合并完成，mode/label 经 `ctx.sessionProjections` 取值；没有 query backend 的部署照常列表。 -- **取值三级「算完即止」阶梯**：live child 读 `sessionProjections.snapshot(session, ['subagent'])`（注册表既有水位缓存，零日志读）；cold child 先读可选 `sessionProjectionCache.cachedSnapshot(header, ['subagent'])`，非 null 身份通过 seq 门（`seq >= seedLength ?? 0`）即直接使用；否则执行一次完整 Session 观察，再经注册的 `subagent` unit 折叠；再没有就没有——不自建缓存、无回写、无索引。 +- **subagent 列表使用 Session query corpus 完成枚举与含正文 observation**：mode/label 仍经 `ctx.sessionProjections` 获取，列表不拥有 descriptor parser 或领域索引。 +- **取值三级「算完即止」阶梯**：live child 读 `sessionProjections.snapshot(session, ['subagent'])`（注册表既有水位缓存，零日志读）；unseeded cold child 可读 `sessionProjectionCache.cachedSnapshot(header, SessionLogOffset(0), ['subagent'])`；seeded child 或 cache miss 执行一次携带 `inheritedEventCount` 的 Session observation，再经注册的 `subagent` unit 折叠。再没有就没有——不自建缓存、列表侧无回写、无索引。 - **`subagent` projection unit 是折叠规则唯一权威**：live 与 cold 快照都运行同一份已注册 unit，不存在第二份描述符解释逻辑。 -- **header、描述符（v2）、session-persistence、session-projection(-cache)、session-query(-sqlite) 全部零改动**；存量数据第一次被列表时一次 `inspect` 现算获得精确值，无 unknown 降级态、无迁移。 +- **描述符（v2）保持不变**。Session、persistence、projection cache 与 query 在 logical header 之外单独携带精确 inherited cut；listing 无法证明 cut 为零时，存量数据经一次含正文 observation 获得精确值——无 unknown 降级态，也无持久格式迁移。 与既有记录的关系： @@ -36,8 +36,8 @@ mode 与 label 由新的 `subagent` projection unit（纯身份两臂）折叠 ```ts ignore-check export type SubagentIdentityProjection = - | { mode: 'one-shot'; label?: string; seq: number } - | { mode: 'continuable'; label: string; seq: number } + | { mode: 'one-shot'; label?: string; seq: SessionSeq } + | { mode: 'continuable'; label: string; seq: SessionSeq } declare module '@deepseek-ai/dsh-session-projection/types' { interface SessionProjectionStateMap { @@ -52,19 +52,19 @@ declare module '@deepseek-ai/dsh-session-projection/types' { - 投影是纯身份，**projection 体系不做失败通道**：unit 永不抛错；载荷损坏、版本不认识与整日志没有描述符一样。host checkpoint 状态使用可序列化的包装 `{ identity?: SubagentIdentityProjection }`，缺席为 `{}`；客户端 view 则是非可选的 `SubagentIdentityProjection | null` 条目。`null` 完好通过 JSON，因此推送 reset 会替换旧身份，而不会被 stringify 丢掉。判定纪律：消费面把 null 与客户端 key 缺席一律视为无值。「算出来没有」如何呈现是消费方自己的事（见下文 `listChildren` 四态映射）。 - label 强度由描述符 schema 决定：continuable 的 label 解析强制必有，one-shot 的本就可选；mode/label 判别与下文 child 行的强约定完全一致（行不携带 `seq`——它是投影内部的 own-suffix 证明）。 -- 身份携带 `seq`：折出该身份的 `subagent/descriptor` 事件 seq，两臂必有、null 哨兵无——`seq >= header.seedLength ?? 0` 证明身份折叠自 child 自身后缀，而非 fork 种子回放的祖先描述符。unit 把包装状态中校验后的身份映射为客户端 wire view，并与其他 unit 一律检查点化（`persist` 选项已删除）；`stateVersion` 为 2，在增加 `seq` 时升版。更早的 checkpoint 行按 registry 约定版本失配失效、落权威重折。 +- 身份携带品牌化 `seq`：折出该身份的 `subagent/descriptor` 事件 seq，两臂必有、null 哨兵无。live Session 通过 `isOwnSeq()` 检查它；cold 含正文 observation 则与 `inheritedEventCount` 比较。仅 header 的 seeded candidate 会跳过 cache，因为 header 有意不暴露整数 cut；unseeded candidate 知道 cut 为零。unit 把包装状态中校验后的身份映射为客户端 wire view，并与其他 unit 一律检查点化（`persist` 选项已删除）；`stateVersion` 为 2，在增加 `seq` 时升版。更早的 checkpoint 行按 registry 约定版本失配失效、落权威重折。 - 折叠规则：`subagent/descriptor` last-wins，与 `subagentTiming` 同一条 descriptor-reset 纪律——fork 前缀里的祖先描述符被自身描述符覆盖。损坏或版本不认识的载荷同样 last-wins：重置为 null 哨兵而非保留先前身份，健康祖先的 fork 不会继承自身描述符立不住的身份。 -### 枚举：subagent 自管 live-preferred 合并 +### 枚举：query corpus 与 live preference -`listChildren`（[list-children.ts](../../../../packages/subagent/subagent/src/list-children.ts)）的枚举不经任何查询服务：`ctx.sessions.list()` 与 `ctx.get('sessionPersistence')?.list()` 两个来源按 id 合并，live 记录整条覆盖同 id 持久化记录、不做 header 一致性校验。枚举所需全部是 header 事实： +`listChildren`（[list-children.ts](../../../../packages/subagent/subagent/src/list-children.ts)）通过 `sessionQuery.listSessions()` 取得 canonical live-preferred corpus，再把每个 listed id 与可能存在的 `ctx.sessions.get(id)` 配对；同 id 存在 live Session 时使用 live header。枚举所需全部是 header 事实： - 过滤：`header.origin === 'subagent' && header.parentSession === parentSessionId`。 - `hasChildren`：同一份合并材料向下看一层——存在 `origin === 'subagent'` 且 `parentSession` 为该 child 的直接后代。 - `activity`：live 记录为 `running`，仅存在于持久化的为 `inactive`。 - 排序：`createdAt` 升序、再按 child id 升序（与旧约定一致）。 -- **persistence 缺席退为 live-only 枚举，不报错**：没有 persistence 的部署，cold child 本就无法 resume，列出 live child 仍然有意义。（对照：旧实现在 sessionQuery 缺失时整体拒绝。） -- persistence 列表失败使整次枚举失败；per-child 隔离只作用于逐 child 的冷读。 +- `sessionQuery` 服务缺席时以 `SUBAGENT_CONTROL_QUERY_UNAVAILABLE` 失败；共享 query corpus 负责决定部署能枚举 live-only 还是持久化 Session。 +- query corpus 失败使整次枚举失败；per-child 隔离只适用于逐 child cold observation。 ### 取值：三级「算完即止」阶梯 @@ -73,20 +73,20 @@ declare module '@deepseek-ai/dsh-session-projection/types' { | 级 | 读法 | 成本 | | --- | --- | --- | | 1：live child | `ctx.sessionProjections.snapshot(session, ['subagent'])` | 零日志读——注册表既有水位缓存，同步取值 | -| 2：cold child，cache 命中 | 可选 `sessionProjectionCache.cachedSnapshot(header, ['subagent'])`，非 null 身份满足 `identity.seq >= header.seedLength ?? 0` 才直接使用——own descriptor 一经追加不可变，seq 门证明该值折叠自 child 自身后缀，无视行水位 | 零日志读 | -| 3：cold child，兜底 | `persistence.inspect(id)` 整读 + 经注册的 `subagent` unit 折叠 | 每次列表一次整读现算 | +| 2：unseeded cold child，cache 命中 | 可选 `sessionProjectionCache.cachedSnapshot(header, SessionLogOffset(0), ['subagent'])`；精确 cut 为零时，每个合法 seq 都归 child 自有 | 零日志读 | +| 3：seeded child 或 cold 兜底 | 一次含正文 `sessionQuery.observeSession(id)` 加已注册的 `subagent` projection，使用 `inheritedEventCount` 做 own-suffix 检查 | 每次列表一次整读现算 | -- 错误约定：`sessionProjections` 是必需注入——`SubagentRuntime` 在 inject 集里声明它，没有 registry 的部署根本无法激活服务（与 loop），`listChildren` 不可达，而不是供出降级行（[mandatory-seam 记录](2026-08-19-session-projection-mandatory-seam.zh.md)）；响亮运行时检查与 `SUBAGENT_CONTROL_PROJECTIONS_UNAVAILABLE` 随之删除。会话存储保留显式姿态：`ctx.get('sessions')`（严格全局读取，不走调用方作用域的属性代理）缺席以 `SUBAGENT_CONTROL_SESSION_STORE_UNAVAILABLE` 失败。apiproxy 为 `PROJECTIONS_UNAVAILABLE` 设的专门 wire 脸随码删除；`SESSION_STORE_UNAVAILABLE` 走通用 internal 兜底——apiproxy 组合自身就 inject `sessions`，该错误在其部署不可达，专门映射违反 need 原则。`SUBAGENT_CONTROL_SESSION_QUERY_UNAVAILABLE` 已随 session-query 依赖删除。 -- cache 是纯可选加速层：服务缺席判空跳过——无错误码、不进配置校验（与 `sessionProjections` 的必需注入相对）。第二级任何抛错（包括缓存内任一 unit 行中毒使 `viewCheckpoint` 引爆）静默落第三级——缓存是派生数据，其故障不产生 `corrupt` 判决，终审归权威重折；checkpoint 切面早于描述符的行，`subagent` key 天然缺席，自动落底，无特判；行里的 null 哨兵同样不作数——一律落第三级，由权威重折裁决。创建窗口内的 count/interval checkpoint 可能把 fork 种子回放的祖先身份落进行——祖先 seq 落在 seed 区间，被 seq 门拒绝，同样落第三级裁决。 +- 错误约定：`sessionProjections`、Session store 与 `sessionQuery` 都是 listing 所需的 runtime service。三者分别以 `SUBAGENT_CONTROL_PROJECTIONS_UNAVAILABLE`、`SUBAGENT_CONTROL_SESSION_STORE_UNAVAILABLE` 与 `SUBAGENT_CONTROL_QUERY_UNAVAILABLE` 显式失败；缺失分类或 corpus 能力不会伪装成空结果。 +- cache 是纯可选加速层：服务缺席判空跳过——无错误码、不进配置校验（与 `sessionProjections` 的必需注入相对）。seeded header 会跳过该级，因为不读取正文就无法提供 cache identity 所需的精确 cut。对 unseeded child，第二级任何抛错（包括中毒 unit 行引爆 `viewCheckpoint`）都会静默落第三级——缓存是派生数据，其故障不产生 `corrupt` 判决，终审归权威重折；checkpoint 早于 descriptor、key 缺席或 null 哨兵也都会落底。 - per-child 隔离：单 child 的 cold 整读失败只使该行成为 `unavailable` diagnostic，下次列表自然重试，不影响 sibling（见四态映射）。 -- 冷路径的生命周期见证：preparation 的结果必须仍指向枚举时的那个生命周期——见证字段集与旧 SOURCE_CONFLICT 检查同款七字段（version、id、createdAt、cwd、parentSession、seedLength、delegationDepth）；同 id 删除后重新发布的会话对旧 parent 的目录降级为 `corrupt` 行，不外漏新 owner 的 child。 +- 冷路径的生命周期见证：observation 必须仍指向枚举时的那个生命周期。见证字段为 version、id、createdAt、cwd、parentSession、isSeeded、delegationDepth、origin 与 agentPreset；同 id 删除后重新发布的 Session 对旧 parent 的目录降级为 `corrupt` 行，不外漏新 owner 的 child。 - 冷读并发以常数 4 有界——它约束的是本地介质的一次只读扫描而非部署行为；出现联网 persistence backend 时提升为验证过的 `Config` 字段。 -- 冷读成本如实记录：cache 未挂载或未命中时，cold child 每次列表才付一次整读，成本与其 transcript 大小成正比；定案「算完即止」，不自建缓存。整读经 `inspect()` 走 [Session 准备阶段](2026-08-05-session-preparation.zh.md)的冷读，同 id 短期重复读取可命中其 LRU 复用，但列表不依赖此。live child 全程零日志读。 +- 冷读成本如实记录：每个 seeded child 与每次 unseeded cache miss 都会在每次列表时支付一次完整 query observation，成本与其 transcript 大小成正比；定案「算完即止」，不自建缓存。observation 可以复用 query／persistence preparation 层，但列表不依赖该优化。live child 全程零日志读。 - 取消：每次 persistence 读前后检查调用方 signal，abort 之后才结算的读拒绝归一化为稳定错误码 `CANCELLED`。 ### 权威模型 -- session log 是唯一权威；本方案不新增任何派生持久化——没有索引值、没有自己的 checkpoint、没有进程 memo；第二级读取的 `sessionProjectionCache` checkpoint 是既有组合项的派生数据，本方案只读不写。取值现算现弃，值的新鲜度就是读取时点的 live 状态或持久化 revision（own descriptor 一经追加不可变——缓存身份过 seq 门后无陈旧性问题，门防的是种子回放的祖先身份）。 +- session log 是唯一权威；本方案不新增领域索引、自有 checkpoint 或进程 memo。第二级读取的 `sessionProjectionCache` checkpoint 是既有组合项的派生数据，列表只读。取值现算现弃；seeded candidate 用含正文 observation 按精确 cut 分类，unseeded cached identity 无需 seq 门，因为每个合法 seq 都归自身所有。 - Session 与 persistence 写路完全不感知列表与投影消费：没有事件监听回写，没有写时折叠。 - 枚举与取值不构成第二个鉴权来源，也不让尚未发布的 child 可见——两个来源只见已发布的 live 记录与已落盘的持久化记录，与 durable-subagent-catalog 记录对派生读面立下的规则一致。 @@ -127,25 +127,24 @@ export type SubagentListEntry = 已知边界偏差（有意接受，随本记录留档）： -- 死于发布窗口的 fork child，seed 里若有祖先描述符，last-wins 会给出祖先身份，误现为 child 行；恢复仍按 own-suffix 折叠权威失败（`NOT_RESUMABLE`）。旧实现靠 `seedLength` 过滤将其 omit；projection unit 看不到 header，接受此残骸级偏差（`subagentTiming` 有同类既有暴露）。 - own suffix 出现多个描述符，旧实现判 corrupt，现 last-wins 取末者（提供方约定本就保证恰一）。 - live/persisted header 冲突，旧实现是 per-child corrupt；现枚举 live 优先、不做一致性校验，冲突不再被察觉，以 live 记录成行。 - 损坏存储的源读失败（如坏 surface 被冷读整读拒收），旧实现映射 per-child `corrupt`，现统一成 `unavailable` 行（读侧无从区分成因）。 - 未知 parent，旧实现经 session-query 抛 not-found（「parent session … was not found」）；现自管合并对不存在的 parent 得到空子集，枚举返回空列表，wire 上后续操作落到 child 级 subagent-not-found——语义与文案的静默变化，显式接受。 -- rung 2 的更晚事件窗口：cache 行恰在首个自有描述符之后落盘，日志随后追加第二个自有描述符（或 malformed 载荷置 null 哨兵），且进程在下一次 checkpoint 前崩溃——此后冷列表的 rung 2 凭 seq≥seedLength 门持续供出行内旧身份（第一个自有描述符的值），与权威重折（last-wins 第二个）分歧，且 rung 2 命中期间不触发重折、无从察觉。边界三条：①前提是同一 child 出现第二个自有描述符，违反建档提供方「恰追加一次」约定，属损坏类数据，与多描述符偏差同族同源；②需「损坏 + 崩溃错过 checkpoint（turn/end 与 disposal 两个 mandatory 点及 count/interval 节流点全部未及）」双条件同时成立；③健康 child（恰一自有描述符）不受影响——seq 门放行的正是唯一真身份。自愈条件：该 child 任一次 live 运行（turn/end mandatory checkpoint）或任何触发 cache.write 的时点，都会以新 fold 整行覆写（whole-record replace），rung 2 随即供正；权威路径（rung 3 重折、live snapshot、resume 折叠）自始正确，分歧只存在于持续冷、行未再更新期间的列表读。机制修法不采：gate 对账需知日志末端 seq，冷路径零读不可得；cache 行携 revision 是 opaque token，无法比较且跨域改 schema——按「cache 永不为权威」总纲归档为接受项。 +- rung 2 的更晚事件窗口只适用于 unseeded child：cache 行恰在首个 descriptor 后落盘，日志随后追加第二个 descriptor（或 malformed 载荷置 null 哨兵），且进程在下一次 checkpoint 前崩溃。cold listing 可能持续供出旧身份，直到一次 live 运行或 cache write 替换该行。其前提违反 provider 的「恰追加一次」约定，并且还需错过所有 mandatory checkpoint；健康 child 不受影响。seeded child 没有 body-owned cut 时绝不进入 rung 2。 -消费面：wire、tool、GUI 的 diagnostic 处理**全部保持原状零改动**（`list_agents` 的 description 与 output schema 未动；该插件的加载要求变化——inject 去掉 `sessionQuery`、新增必需注入 `sessionProjections`）。行为上动的只有 apiproxy：路由段的 `hasSubagentDescriptor()` 扫描已删除，`hasSubagentOwner` 只看 `header.origin`——pre-#1569 的无 `origin` 存量不再被认作 subagent 属主，其本就不进目录，pre-release 立场接受；`subagents.history` 与 `session.history` 同源对齐——live child 用内存事件与注册表水位快照，cold child 用 `inspectServable` 直读持久化并 detached 折叠，不经查询服务，SESSION_QUERY_* 错误臂随之退役，wire 形状不变（`history` 的 JSDoc 措辞改为 live 内存快照／cold 持久日志双臂）。 +消费面保持相同的 row 与 diagnostic wire 形状。`list_agents` 使用必需的 query corpus 与 projection registry；live identity 来自 registry snapshot，cold identity 来自 cache 或 query observation。Host ownership 仍使用 `header.origin`，history 使用共享的 live／cold Session query source；没有消费方独立解析 descriptor event。 ### 改动落点 | 区域 | 文件 | 改动 | | --- | --- | --- | | subagent | projection.ts、projection-types.ts、index.ts | 新客户端可见 `subagent` unit 与注册 | -| subagent | list-children.ts 及类型 | 重写为自管枚举 + 投影阶梯四态映射；删 session-query 依赖、逐 child 事件读取与就地分类机器；错误码 `SUBAGENT_CONTROL_SESSION_QUERY_UNAVAILABLE` 删除，`sessionProjections` 转为必需注入（不再存在投影错误码）；新增可选依赖 dsh-session-projection-cache（纯加速读取，缺席跳过） | -| host/apiproxy | api-proxy.ts | 删 `hasSubagentDescriptor`，属主判定只看 `header.origin`；`subagents.history` 与 `session.history` 同源——live 用内存事件与注册表水位快照，cold 用 `inspectServable` 直读持久化并 detached 折叠，不经查询服务，SESSION_QUERY_* 错误臂与 `PROJECTIONS_UNAVAILABLE` 专门 wire 脸随之退役 | -| tool | tool-subagent-control/list-agents.ts | 加载要求收窄（inject 去 `sessionQuery`）；model-visible schema、描述与渲染零改动 | +| subagent | list-children.ts 及类型 | query-corpus 枚举加 projection 阶梯四态映射；必需 projections／query service 与可选 projection-cache 加速 | +| host/apiproxy | Session controller／query integration | owner 检查使用 `header.origin`；live／cold history 与 listing 消费共享 query 和 projection source | +| tool | tool-subagent-control/list-agents.ts | model-visible schema、描述与渲染保持不变 | | wire/client | api/subagents.ts、runtime sessions/service.ts、GUI | 类型、行形状与 diagnostic 处理**零改动**；api/subagents.ts 仅 `history` 的 JSDoc 措辞改为双臂 | -| core/session、session-persistence、session-projection(-cache)、session-query(-sqlite) | — | **零改动** | +| core/session、session-persistence、session-projection(-cache)、session-query(-sqlite) | 含正文 cut 与品牌化 seq 传递 | Logical header 暴露 `isSeeded`；Session、persistence observation、cache identity 与 query record 单独携带精确 `inheritedEventCount` | ## 考虑过的替代方案 @@ -169,20 +168,20 @@ export type SubagentListEntry = ## 验证 -`packages/subagent/subagent/tests/list-children.spec.ts` 重写为本约定：无 persistence、query 服务与继续运行时的 live-only 列表；registry 缺席时服务根本不激活（mandatory seam——`setup` 变体断言 `ctx.get('subagents')` 保持 undefined）；live child 全程零 `inspect`、cold child 每次列表恰一次；多描述符 last-wins 取末者；损坏载荷与未知版本折为 `corrupt`；冷读失败映射 `unavailable` 且下次列表重试；fork seed 里的祖先描述符按该身份成行（偏差一钉住）；普通 fork 与无 subagent origin 的后代不入列也不计入 `hasChildren`；`createdAt`→id 排序；提供方未挂载不影响列表；压缩与未压缩孪生一致；预中止、持久化列表与冷读取消三例归一 `CANCELLED`；空列表与稳定错误码（存储缺席时 `SUBAGENT_CONTROL_SESSION_STORE_UNAVAILABLE`）。第二级例：own-seq 身份直用零 `inspect`、fork 种子祖先身份（seq 落在 seed 区间）被门拒绝落底、行内无身份（null 哨兵或 key 缺席）落底、cache 服务缺席落底、缓存行中毒静默落底重折；冷路径 lifecycle 篡改按见证七字段逐一（`it.each`）降级为 `corrupt`。`tool-subagent-control` 的 list-agents 测试随加载要求收窄更新；`optional-session-query.spec.ts` 随依赖消失删除；既有无密钥快照（`subagent-list-agents` 等）零变化，钉住健康路径的 wire 与 model-visible 面不变；新增无密钥快照 `subagent-diagnostic`（examples/headless-agent）钉住四态映射的诊断分类——descriptor-less 定局残骸成 `corrupt` 行等模型可见变化。 +`packages/subagent/subagent/tests/list-children.spec.ts` 固定本约定：live identity 通过 `Session.isOwnSeq()` 检查；unseeded cold identity 可在 cut 零时使用 cache；seeded candidate 跳过该 cache rung，转而使用携带 `inheritedEventCount` 的 observation；祖先 identity 无法通过 own-suffix 检查；缺席、null、中毒与不可用的 cache／observation 会按约定落底或产生 diagnostic；lifecycle 篡改按完整见证字段集降级为 `corrupt`。既有无密钥快照保持健康 wire 与 model-visible 面不变，`subagent-diagnostic` 则固定诊断分类。 ## 后果 - live child 的列表全程零日志读；cold child 在 cache 未挂载或未命中时每次列表一次 `inspect` 整读，成本与其 transcript 大小成正比、随列表频率重复——定案「算完即止」，不自建缓存、不回写，同 id 短期重复整读可命中准备阶段 LRU 但列表不依赖它。 -- subagent 列表不再要求 query backend：纯 live 与无 persistence 的部署都能列表；`SUBAGENT_CONTROL_SESSION_QUERY_UNAVAILABLE` 消失，`list_agents` 插件加载不再要求 `sessionQuery`，而 `sessionProjections` 转为 `SubagentRuntime` 的必需注入——没有投影 registry 的部署根本不会激活服务（mandatory seam）。 +- subagent 列表要求 Session query corpus 与 projection registry；服务缺失会显式失败，而不是供出不完整 row。可选 projection cache 只改变正文读取次数。 - 身份解释只存在于 registry 注册的一份 unit：列表三级阶梯与 GUI history 冷读使用其 live、cached 或 observed wire 快照，不存在手写旁路折叠；若未来某消费面绕开该 unit 手写折叠，各读面的值将漂移——这是本设计要求维持的纪律，不是机制保证。 - per-child 隔离回归：单 child 冷读失败只损失该行，healthy sibling 不受影响；persistence 列表失败仍使整次枚举失败。 -- 诊断与枚举语义留下六处边界偏差（stillborn fork 祖先身份误现、多描述符取末者、header 冲突不再被察觉、损坏源读失败由 `corrupt` 转 `unavailable`、未知 parent 由 not-found 改为空列表、rung 2 更晚事件窗口），完整语义见已知边界偏差清单；前四处为残骸级数据的展示或分类偏差，未知 parent 一处是查询语义的静默变化，rung 2 窗口一处是损坏加崩溃双条件下可自愈的缓存供值分歧；恢复鉴权均不受影响，显式接受。 +- 诊断与枚举语义留下五处边界偏差（多描述符取末者、header 冲突不再被察觉、损坏源读失败改变分类、未知 parent 由 not-found 改为空列表、unseeded rung 2 更晚事件窗口）。seeded 祖先 identity 已不再构成偏差，因为含正文读取会把它与 `inheritedEventCount` 比较；恢复鉴权始终不受影响。 - pre-#1569 的无 `origin` 存量不再被认作 subagent 属主；其本就不进目录，pre-release 无兼容承诺。 ## 相关 -- [durable-subagent-catalog 与 list_agents](../feature/2026-07-22-durable-subagent-catalog-and-list-agents.zh.md)——被本记录部分取代：描述符仍是 mode/label 的持久权威与折叠输入，列表的枚举与取值改为自管合并加投影阶梯。 +- [durable-subagent-catalog 与 list_agents](../feature/2026-07-22-durable-subagent-catalog-and-list-agents.zh.md)——被本记录部分取代：描述符仍是 mode/label 的持久权威与折叠输入，取值改为共享 query corpus 上的 projection 阶梯。 - [session projections 与命令生命周期日志](../../proposed/architecture/2026-07-27-session-projection-and-command-log.zh.md)——registry 约定的权威；本记录为其新增 `subagent` 身份 unit，并消费其 live 与 cold wire 快照。 - [session projection 状态与客户端视图](2026-08-19-session-projection-state-and-client-views.zh.md)——state/client 拆分；`subagent` 与 `subagentTiming` 都提供客户端 wire view。 - [session projections 作为必需接缝](2026-08-19-session-projection-mandatory-seam.zh.md)——`sessionProjections` 转为必需注入；列表的错误约定随其变化（registry 缺席是激活期失败，投影错误码删除）。 diff --git a/.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.i18n.yaml index 0ff8938b30..4a93d57022 100644 --- a/.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.md -2026-08-08-bounded-session-persistence-write-batching.md: 20c16991b0be30ffe546a94c257bc65f86cb57eb -2026-08-08-bounded-session-persistence-write-batching.zh.md: ac0384f4e28175922f84d23296dfb13848cf5dd3 +2026-08-08-bounded-session-persistence-write-batching.md: 6fb44e494fc17bde08eb3132afe42ce73b5a4e47 +2026-08-08-bounded-session-persistence-write-batching.zh.md: 576764583dd8c465ae45866f62ef15773735961e diff --git a/.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.md b/.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.md index 20c16991b0..6fb44e494f 100644 --- a/.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.md +++ b/.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.md @@ -20,19 +20,19 @@ The scheduling bound is deterministic. With an immediately resolving sink, the f ## Decision -The JSONL provider exposes `writeBatchMaxDelayMs`, a positive integer no greater than Node's timer limit. Its default is `200`. The provider resolves the value at load and passes it to `PersistenceCoordinator`; the coordinator remains the single owner of batching behavior. +The fixed window is the JSONL provider's constant `LIVE_WRITE_BATCH_MAX_DELAY_MS` (200 ms), an internal scheduling policy rather than configuration: the backend's own session listeners route live events by id into the active write handle's buffer, so batching never crosses the package boundary ([handle note](2026-08-27-handle-based-session-persistence.md)). -Each live Session receives a package-private `SessionWriteBehind`. When its pending queue changes from empty to non-empty, the controller starts one fixed window. Later events join that batch without resetting the deadline: this is bounded coalescing, not debounce. When the deadline expires, the controller hands the complete pending prefix to the existing per-id serialization and `appendBatch` path. At most one write for a Session is active. Events admitted during that write form a new pending prefix with their own fixed deadline; if that deadline expires before the active write completes, the new prefix starts immediately after it. +Each active write handle owns its buffer directly. A routed event lands in the handle's pending array, and the first event of an idle buffer arms one fixed timer. Later events join that batch without resetting the deadline: this is bounded coalescing, not debounce. When the deadline expires, a single-flight drain persists the pending prefix through the handle's mutation chain, which already serializes it against explicit appends. Events admitted during a drain pass coalesce into the next chained batch, in order. -`writeBatchMaxDelayMs` bounds only the controller's intentional batching wait. Event-loop scheduling, initialization, an earlier serialized operation, and backend I/O can delay durable completion, so the option is not a hard fsync or crash-loss SLA. +The window bounds only the controller's intentional batching wait. Event-loop scheduling, initialization, an earlier serialized operation, and backend I/O can delay durable completion, so the option is not a hard fsync or crash-loss SLA. -`session/flush` cancels any remaining wait and becomes a shared quiescence barrier. It drains the active attempt and every event admitted while the barrier is running before it resolves. Session retirement and backend disposal use that same barrier, so lifecycle teardown never waits for the batching timer. The checkpoint policy continues to place mandatory barriers before model requests and top-level tool side effects. +`session/flush` cancels any remaining wait and becomes a shared quiescence barrier. It drains the active attempt and every event admitted while the barrier is running before it resolves. Session retirement (`session/disposed`), the handle's close, and backend teardown's close sweep use that same barrier, so lifecycle teardown never waits for the batching timer. The checkpoint policy continues to place mandatory barriers before model requests and top-level tool side effects. Every event remains durable in its original order and shape. The controller copies each event on admission; no `assistant/chunk`, `seq`, `time`, surface metadata, or storage record is removed or rewritten. JSONL can therefore encode more events in one append frame without changing its on-disk format. -A failed background append restores its complete batch before any newer pending events, reports the failure once, and pauses automatic retry. The next newly admitted event opens a fresh fixed window; an explicit flush, retirement, or disposal retries immediately and surfaces a repeated failure to its caller. This avoids a timer-driven failure loop while preserving the existing recoverable flush boundary. +A failed background drain retains its complete batch in order ahead of newer pending events, reports the failure once, and pauses the automatic timer. The next explicit drain — a `session/flush` barrier, service-level `flush()`, or close — retries immediately and surfaces a repeated failure to its caller. This avoids a timer-driven failure loop while preserving the existing recoverable flush boundary. -This decision supersedes only the immediate scheduling cadence in [Collapse live persistence into one flush controller](../simplification/2026-07-23-collapse-persistence-flush-state.md). That note remains authoritative for one controller per live Session, retained failed batches, per-id serialization, retirement, and quiescent disposal. The [shared persistence coordinator](2026-06-18-shared-persistence-write-coordinator.md) remains the owner of the backend hook boundary. +This decision supersedes only the immediate scheduling cadence in [Collapse live persistence into one flush controller](../simplification/2026-07-23-collapse-persistence-flush-state.md). That note remains authoritative for one buffer owner per live Session, retained failed batches, retirement, and quiescent disposal. The coordinator and the separate write-behind controller that first hosted this behavior are deleted; the buffer, timer, and drain live on the provider's handle, and the [handle-based seam](2026-08-27-handle-based-session-persistence.md) owns the storage boundary they write through. ## Alternatives considered @@ -42,11 +42,11 @@ This decision supersedes only the immediate scheduling cadence in [Collapse live **Debounce from the latest event.** Rejected: a continuously streaming response could postpone its first write indefinitely. A fixed window from the first pending event provides a real upper bound on intentional coalescing wait. -**Implement the timer inside JSONL.** Rejected: scheduling, failure retention, flush races, and teardown are provider-neutral lifecycle concerns that belong in `PersistenceCoordinator`; an out-of-tree provider can reuse the same behavior. +**A shared provider-neutral controller component.** Rejected after one iteration shipped it: the handle's mutation chain already serializes writes, so a separate controller duplicated that ordering machinery. Each provider implements the buffer on its own handle, and the shared live-write contract suite pins the equivalent observable behavior for any provider. ## Verification -The controller tests use a fake clock to prove the fixed, non-resetting 200 ms window; immediate and shared flush barriers; events admitted during a barrier; an over-budget tail behind an active write; ordered failure retention; paused automatic retry; and explicit retry of an overlapping background failure. Coordinator tests run the controller through Session notifications, retirement, collision reclamation, and teardown. The JSONL suite retains storage-format, recovery, and shared persistence-contract coverage. +The shared live-write contract suite (`runLiveWritePathContract`) uses a fake clock to prove the fixed, non-resetting 200 ms window; the `session/flush` barrier and its loud failure surfacing; ordered failure retention with exactly-once recovery; the service-level `flush()` sweep with per-session failure aggregation; and the disposed/close/teardown drains. The JSONL suite retains its storage-format, recovery, and shared persistence-contract coverage. ## Consequences @@ -54,6 +54,6 @@ High-frequency event bursts normally produce fewer durable append operations whi This decision does not cap pending event count or bytes behind a slow provider, and it does not reduce the decoded logical log. A demonstrated memory bound or logical-retention policy would require its own failure and replay contract rather than another hidden timer rule. -An admitted event can remain only in memory during the configured window, and then while scheduling or backend work is outstanding. Deployments choose a smaller value for a narrower ordinary loss window or a larger value for stronger batching. Explicit durability boundaries remain unchanged and bypass the wait. +An admitted event can remain only in memory during the fixed window, and then while scheduling or backend work is outstanding. Explicit durability boundaries remain unchanged and bypass the wait. -The deep module gives the timer, active write, pending prefix, retry pause, and barrier one owner. `PersistenceCoordinator` retains initialization and identity serialization; the provider retains only durable storage primitives. `SESSION_FORMAT_VERSION` remains unchanged. +The handle gives the timer, active drain, pending prefix, retry pause, and barrier one owner; the backend's listeners own routing and lifecycle-driven drains. `SESSION_FORMAT_VERSION` remains unchanged. diff --git a/.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.zh.md b/.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.zh.md index ac0384f4e2..576764583d 100644 --- a/.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.zh.md @@ -20,19 +20,19 @@ JSONL 每个持久化追加批次会写入一个 Zstandard 帧并执行一次 fs ## 决策 -JSONL provider 公开 `writeBatchMaxDelayMs`，其值必须是一个不超过 Node 计时器上限的正整数，默认值为 `200`。provider 在加载时解析该值，再传给 `PersistenceCoordinator`；批处理行为仍只由协调器负责。 +固定窗口是 JSONL provider 的常量 `LIVE_WRITE_BATCH_MAX_DELAY_MS`（200 ms），它是内部调度策略而非配置：后端自己的会话监听器按 id 把实时事件路由进活跃写句柄的缓冲，因此批处理绝不跨越包边界（[句柄 Note](2026-08-27-handle-based-session-persistence.zh.md)）。 -每个活跃的会话都有一个包私有 `SessionWriteBehind`。当其待处理队列从空变为非空时，控制器会启动一个固定窗口。后续事件加入该批次但不会重置截止时间：这属于有界合并，而不是防抖。截止时间到达后，控制器会把完整的待处理前缀交给现有的按 id 串行化机制，并沿 `appendBatch` 路径写入。同一会话同时最多有一个活跃写入。该写入期间接纳的事件会形成新的待处理前缀，并拥有自己的固定截止时间；如果该截止时间在活跃写入完成前到期，新前缀会在前一次写入完成后立即开始写入。 +每个活跃写句柄直接拥有自己的缓冲。被路由的事件落入句柄的待处理数组，空闲缓冲收到的第一个事件会启动一个固定计时器。后续事件加入该批次但不会重置截止时间：这属于有界合并，而不是防抖。截止时间到达后，一次 single-flight 排空会把待处理前缀经由句柄的修改链持久化，该链本就将其与显式 append 串行化。排空进行期间接纳的事件会按顺序合并进下一个链上的批次。 -`writeBatchMaxDelayMs` 只限制控制器为批处理而主动等待的时间。事件循环调度、初始化、此前的串行化操作和后端 I/O 都可能延后持久化完成时间，因此该选项并不对 fsync 完成时间或崩溃数据丢失提供硬性 SLA。 +该窗口只限制控制器为批处理而主动等待的时间。事件循环调度、初始化、此前的串行化操作和后端 I/O 都可能延后持久化完成时间，因此该选项并不对 fsync 完成时间或崩溃数据丢失提供硬性 SLA。 -`session/flush` 会取消剩余等待，并充当共享的完全停稳屏障。它会在完成前等待活跃写入尝试，并排空屏障运行期间接纳的每个事件。会话退役与后端 dispose（资源释放）共用该屏障，因此生命周期 teardown 绝不会等待批处理计时器。检查点策略仍会在模型请求与顶层工具副作用之前设置强制屏障。 +`session/flush` 会取消剩余等待，并充当共享的完全停稳屏障。它会在完成前等待活跃写入尝试，并排空屏障运行期间接纳的每个事件。会话退役（`session/disposed`）、句柄的 close 与后端 teardown 的关闭清扫共用该屏障，因此生命周期 teardown 绝不会等待批处理计时器。检查点策略仍会在模型请求与顶层工具副作用之前设置强制屏障。 每个事件仍会按原有顺序和形态持久化。控制器会在接纳时复制每个事件；任何 `assistant/chunk`、`seq`、`time`、surface 元数据或存储记录都不会被删除或重写。因此，JSONL 可以在一个追加帧中编码更多事件，而无需改变其磁盘格式。 -后台追加失败后，控制器会把完整批次恢复到所有较新的待处理事件之前，报告一次该失败，并暂停自动重试。随后新接纳的第一个事件会开启新的固定窗口；显式 flush、退役或 dispose 会立即重试，如果故障再次发生，则会向调用方暴露该故障。这可以避免计时器驱动的失败循环，同时保留现有可恢复的 flush 边界。 +后台排空失败后，其完整批次会按顺序保留在所有较新的待处理事件之前，该失败被报告一次，自动计时器随之暂停。下一次显式排空——`session/flush` 屏障、服务级 `flush()` 或 close——会立即重试，如果故障再次发生，则会向调用方暴露该故障。这可以避免计时器驱动的失败循环，同时保留现有可恢复的 flush 边界。 -本决策仅取代[将实时持久化归并到单个刷新控制器](../simplification/2026-07-23-collapse-persistence-flush-state.zh.md)中的即时调度节奏。对于每个活跃会话使用一个控制器、保留失败批次、按 id 串行化、退役和完全停稳的 dispose，原 Agent Note 仍是权威记录。后端钩子边界仍由[共享持久化协调器](2026-06-18-shared-persistence-write-coordinator.zh.md)定义。 +本决策仅取代[将实时持久化归并到单个刷新控制器](../simplification/2026-07-23-collapse-persistence-flush-state.zh.md)中的即时调度节奏。对于每个活跃会话使用一个缓冲所有者、保留失败批次、退役和完全停稳的 dispose，原 Agent Note 仍是权威记录。最初承载该行为的协调器与独立的 write-behind 控制器均已删除；缓冲、计时器和排空落在 provider 的句柄上，它们写入所经过的存储边界由[基于句柄的 seam](2026-08-27-handle-based-session-persistence.zh.md) 定义。 ## 备选方案 @@ -42,11 +42,11 @@ JSONL provider 公开 `writeBatchMaxDelayMs`，其值必须是一个不超过 No **按最新事件重置防抖窗口。** 不采纳：持续不断的流式响应可能无限期推迟首次写入。由第一个待处理事件启动的固定窗口，为主动合并等待提供了真正的上界。 -**在 JSONL 内实现计时器。** 不采纳：调度、失败保留、flush 竞态和 teardown 都是 provider 无关的生命周期问题，属于 `PersistenceCoordinator`；仓库外 provider 可以复用同一行为。 +**共享的 provider 无关控制器组件。** 曾在一次迭代中交付，随后不采纳：句柄的修改链本就串行化写入，独立控制器重复了这套排序机制。每个 provider 在自己的句柄上实现该缓冲，共享的实时写入约定测试套件为任何 provider 钉住等价的可观察行为。 ## 验证 -控制器测试使用假时钟证明固定且不会重置的 200 ms 窗口、即时且可共享的 flush 屏障、屏障运行期间接纳的事件、在活跃写入之后已超过窗口时限的尾部批次、有序保留失败批次、暂停自动重试，以及对重叠发生的后台失败进行显式重试。协调器测试会在会话通知、退役、冲突回收和 teardown 路径中验证该控制器。JSONL 测试套件继续覆盖存储格式、恢复和共享持久化约定。 +共享的实时写入约定测试套件（`runLiveWritePathContract`）使用假时钟证明固定且不会重置的 200 ms 窗口、`session/flush` 屏障及其失败的响亮暴露、有序保留失败批次并恰好恢复一次、带逐会话失败聚合的服务级 `flush()` 清扫，以及 disposed/close/teardown 的排空。JSONL 测试套件继续覆盖存储格式、恢复和共享持久化约定。 ## 后果 @@ -54,6 +54,6 @@ JSONL provider 公开 `writeBatchMaxDelayMs`，其值必须是一个不超过 No 本决策不会限制因 provider 缓慢而积压的待处理事件数量或字节数，也不会减少解码后的逻辑日志。若要建立经过验证的内存上界或逻辑保留策略，就必须为其另行定义失败与回放约定，而不是再引入一条隐式计时器规则。 -接纳后的事件在配置窗口内可能只存在于内存中，此后在等待调度或后端工作完成期间也可能如此。部署可以选择较小的值以缩短普通丢失窗口，也可以选择较大的值以加强批处理。显式持久性边界保持不变，并会绕过等待。 +接纳后的事件在固定窗口内可能只存在于内存中，此后在等待调度或后端工作完成期间也可能如此。显式持久性边界保持不变，并会绕过等待。 -deep 模块统一负责计时器、活跃写入、待处理前缀、重试暂停和屏障。`PersistenceCoordinator` 继续负责初始化和按标识串行化；provider 仍只负责持久存储原语。`SESSION_FORMAT_VERSION` 保持不变。 +句柄统一负责计时器、活跃排空、待处理前缀、重试暂停和屏障；后端的监听器负责路由和生命周期驱动的排空。`SESSION_FORMAT_VERSION` 保持不变。 diff --git a/.agents/notes/implemented/architecture/2026-08-08-client-tool-presentation-ownership.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-08-client-tool-presentation-ownership.i18n.yaml index c32026ebc7..4c904a894f 100644 --- a/.agents/notes/implemented/architecture/2026-08-08-client-tool-presentation-ownership.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-08-client-tool-presentation-ownership.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-08-client-tool-presentation-ownership.md -2026-08-08-client-tool-presentation-ownership.md: 1daad1559a6c8ef15fadb8e7c8dfeb2874ae3f9a -2026-08-08-client-tool-presentation-ownership.zh.md: f980db28e1174aa95b29defb8b0a36fc0ba4cf2e +2026-08-08-client-tool-presentation-ownership.md: bf8568150cc173f0dc46ba0a125ca9c784d18f2e +2026-08-08-client-tool-presentation-ownership.zh.md: 015a48a0499b84bf034dc905780eca0dcef99c10 diff --git a/.agents/notes/implemented/architecture/2026-08-08-client-tool-presentation-ownership.md b/.agents/notes/implemented/architecture/2026-08-08-client-tool-presentation-ownership.md index 1daad1559a..bf8568150c 100644 --- a/.agents/notes/implemented/architecture/2026-08-08-client-tool-presentation-ownership.md +++ b/.agents/notes/implemented/architecture/2026-08-08-client-tool-presentation-ownership.md @@ -22,6 +22,8 @@ A business Tool plugin receives one standard `ToolCallBlock`, identity, workspac The details panel is a second Tool presentation point, not the call-tree owner. `ui-conversation` locates the selected call and delegates its output body through `'conversation.details.tool'`; `ui-tool` reuses the card model, while the conversation fallback retains raw result text when the plugin is absent. +Generic row models retain the original argument string as `bodyRaw` and expose no preformatted body. `ToolRow` and the Bash fallback format it only while an expanded generic input section is visible; closing the row removes the formatted text, and rows rendering a structured card skip generic-body formatting. + ## Runtime and render path ```text @@ -33,6 +35,8 @@ Session Event window -> tool.call.toolview(entryKey = toolName) |- registered atomic view `- GenericToolCard fallback + |- collapsed or structured card: retain argsRaw only + `- expanded generic input: format argsRaw ``` ## Ownership boundary @@ -42,12 +46,12 @@ Session Event window | Client Runtime Conversation engine | Context identity, Location, history replay, view Node publication | Tool event meaning, call tree, Tool renderer | | `ui-conversation` Tool Definition | call/result pairing, Code Dispatch topology, running/settled/interrupted `ToolCallBlock`, Chat ordering anchor | Tool-name dispatch, card models, recursive React structure | | `ui-conversation` Chat view | keyed Node order, scroll anchors, selection, and host actions | Tool lifecycle, subcall composition, atomic Tool renderers | -| `ui-tool` | root/subcall recursive rendering, atomic keyed dispatch, fallback, card models, and details output | Session Event fold, Chat ordering | +| `ui-tool` | root/subcall recursive rendering, atomic keyed dispatch, fallback, card models, expansion-time argument formatting, and details output | Session Event fold, Chat ordering | | Business Tool plugin | atomic renderers for one or more wire Tool names | root/subcall placement, lifecycle pairing, Session projectors | ## Verification -`ui-conversation` tests pin the Tool Definition's call/result pairing, Code Dispatch, interruption, and running-to-settled keyed identity without importing production `ui-tool` renderers. `ui-tool` tests mount the real conversation host and pin root/subcall recursion, keyed dispatch, Generic fallback, selection, details, and concrete Tool cards. Assembled Web tests cover the path with both plugins loaded. +`ui-conversation` tests pin the Tool Definition's call/result pairing, Code Dispatch, interruption, and running-to-settled keyed identity without importing production `ui-tool` renderers. `ui-tool` tests mount the real conversation host and pin root/subcall recursion, keyed dispatch, Generic fallback, selection, details, concrete Tool cards, and expansion-only generic-body formatting. Assembled Web tests cover the path with both plugins loaded. ## Alternatives considered @@ -61,8 +65,12 @@ Session Event window **Let `ui-conversation` import `ui-tool` components directly.** Rejected: this would reverse the feature dependency and make Tool presentation mandatory. Slots preserve independent loading, lifecycle, and fallback behavior. +**Keep a preformatted body on the row model for compatibility.** Rejected: every collapsed row would retain a second full argument string, and the compatibility field would let future consumers restore eager formatting. The model exposes only `bodyRaw`, making expansion-time formatting the only generic path. + ## Consequences `ui-conversation` no longer depends on presentation for concrete Tool names, and root and subcalls cannot drift onto different dispatch paths. Business packages can independently own atomic Tool renderers; if `ui-tool` is absent, Conversation data assembly remains valid, Chat Nodes use the generic fallback, and details retain raw results. +Collapsed Tool rows retain the existing `argsRaw` reference without a pretty-printed copy or its formatting call. Expanding a generic input performs that work for the visible row, and closing it permits the derived text to be collected; repeated expansion trades bounded recomputation for lower retained memory. + The cost is an explicit dependency from `ui-tool` on the business Node slot and locale namespace declared by conversation, plus one Tool-specific child slot. Tool Definition remains in `ui-conversation` because this change does not split packages; it can later move through the Conversation registry seam without changing the presentation ownership recorded here. diff --git a/.agents/notes/implemented/architecture/2026-08-08-client-tool-presentation-ownership.zh.md b/.agents/notes/implemented/architecture/2026-08-08-client-tool-presentation-ownership.zh.md index f980db28e1..015a48a049 100644 --- a/.agents/notes/implemented/architecture/2026-08-08-client-tool-presentation-ownership.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-08-client-tool-presentation-ownership.zh.md @@ -22,6 +22,8 @@ Conversation 数据组装遵循后续的 [Conversation 业务节点决策](2026- details panel 是第二个工具展示点，但不是调用树所有者。`ui-conversation` 定位 selected call，并通过 `'conversation.details.tool'` 委托 output body；`ui-tool` 复用 card model，插件缺席时 conversation fallback 保留 raw result text。 +Generic row model 保留原始参数字符串 `bodyRaw`，不暴露预格式化 body。`ToolRow` 与 Bash fallback 只在展开后的 generic input section 可见时格式化该字符串；收起行会移除格式化文本，渲染结构化卡片的行则跳过 generic body 格式化。 + ## 运行时与渲染路径 ```text @@ -33,6 +35,8 @@ Session Event window -> tool.call.toolview(entryKey = toolName) |- registered atomic view `- GenericToolCard fallback + |- collapsed or structured card: retain argsRaw only + `- expanded generic input: format argsRaw ``` ## 所有权边界 @@ -42,12 +46,12 @@ Session Event window | Client 运行时 Conversation engine | 上下文 identity、Location、历史回放、view Node 发布 | 工具事件含义、调用树、工具 renderer | | `ui-conversation` 工具 Definition | call/result 配对、Code Dispatch 拓扑、running/settled/interrupted `ToolCallBlock`、Chat 排序 anchor | 工具名称分发、card model、递归 React 结构 | | `ui-conversation` Chat view | keyed Node 顺序、scroll anchor、selection 与宿主动作 | 工具 lifecycle、subcall 组合、原子工具 renderer | -| `ui-tool` | root/subcall 递归渲染、原子 keyed dispatch、fallback、card model 与 details output | 会话事件 fold、Chat 排序 | +| `ui-tool` | root/subcall 递归渲染、原子 keyed dispatch、fallback、card model、展开时参数格式化与 details output | 会话事件 fold、Chat 排序 | | 业务工具插件 | 一个或多个 wire 工具名称的原子 renderer | root/subcall 位置、生命周期配对、会话 projector | ## 验证 -`ui-conversation` 测试固定工具 Definition 的 call/result 配对、Code Dispatch、interruption 和 running-to-settled keyed identity，不导入 `ui-tool` 的生产 renderer。`ui-tool` 测试挂载真实 conversation 宿主，固定 root/subcall 递归、keyed dispatch、Generic fallback、selection、details 和具体工具 card。组装后的 Web 测试覆盖两个插件共同装载的路径。 +`ui-conversation` 测试固定工具 Definition 的 call/result 配对、Code Dispatch、interruption 和 running-to-settled keyed identity，不导入 `ui-tool` 的生产 renderer。`ui-tool` 测试挂载真实 conversation 宿主，固定 root/subcall 递归、keyed dispatch、Generic fallback、selection、details、具体工具 card 与只在展开时执行的 generic body 格式化。组装后的 Web 测试覆盖两个插件共同装载的路径。 ## 考虑过的替代方案 @@ -61,8 +65,12 @@ Session Event window **让 `ui-conversation` 直接导入 `ui-tool` 组件。** 拒绝：这会反转功能依赖并把工具展示变成必选能力。slot 保留独立装载、生命周期和 fallback。 +**为兼容性在 row model 上保留预格式化 body。** 拒绝：每个折叠行都会保留第二份完整参数字符串，而且兼容字段会让后续消费方恢复 eager 格式化。model 只暴露 `bodyRaw`，使展开时格式化成为唯一 generic 路径。 + ## 后果 `ui-conversation` 不再依赖工具名称对应的业务展示，root 与 subcall 也不会漂移到不同分发路径。业务包可以独立拥有原子工具 renderer；`ui-tool` 缺席时，Conversation 数据组装仍然成立，Chat Node 使用通用 fallback，details 保留 raw result。 +折叠的工具行只保留既有 `argsRaw` 引用，不创建 pretty-print 副本，也不执行对应的格式化调用。展开 generic input 时才为当前可见行完成这项工作，收起后派生文本可以被回收；重复展开以有界重算换取更低的常驻内存。 + 代价是 `ui-tool` 明确依赖 conversation 声明的业务 Node slot 和 locale namespace，并拥有一个工具专属子 slot。工具 Definition 暂时位于 `ui-conversation`，因为本次没有拆包；它以后可以沿 Conversation 注册表 seam 移动，而不会改变本记录规定的展示所有权。 diff --git a/.agents/notes/implemented/architecture/2026-08-09-client-conversation-node-assembly.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-09-client-conversation-node-assembly.i18n.yaml index 10370530e9..23f634a3fe 100644 --- a/.agents/notes/implemented/architecture/2026-08-09-client-conversation-node-assembly.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-09-client-conversation-node-assembly.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-09-client-conversation-node-assembly.md -2026-08-09-client-conversation-node-assembly.md: e37f2ab3bdf9f3943cb1bcd193d5ea578a0b2c19 -2026-08-09-client-conversation-node-assembly.zh.md: be23b321b710f2f56e3e7af870ba554adb355cef +2026-08-09-client-conversation-node-assembly.md: 0aac5056e2cbe22359f8064b1bf4aa0b015a35c8 +2026-08-09-client-conversation-node-assembly.zh.md: b06a92113f91e6297da986866dce097b11bab45f diff --git a/.agents/notes/implemented/architecture/2026-08-09-client-conversation-node-assembly.md b/.agents/notes/implemented/architecture/2026-08-09-client-conversation-node-assembly.md index e37f2ab3bd..0aac5056e2 100644 --- a/.agents/notes/implemented/architecture/2026-08-09-client-conversation-node-assembly.md +++ b/.agents/notes/implemented/architecture/2026-08-09-client-conversation-node-assembly.md @@ -18,6 +18,8 @@ Client Runtime provides a target-neutral Conversation Node assembly engine. Busi This Note retains the derivation, business-by-business validation, responsibilities, algorithms, and trade-offs that remain relevant after implementation. +Chat registers an Inbox Definition only for `next-step`, because message classification is its sole consumer; `next-turn` splices remain durable Session inputs but create no Chat Context. Chat and Trajectory each keep target-owned next-step state. Every insertion stores only message IDs in an immutable splice node. A successful claim materializes the pending chain once, replaces the previous claimed set with that batch, and lets later Contexts share the set until another claim. The AgentLoop appends every message admitted from that claim before it can claim another batch; a rejected claim appends no `user/message`, so later classification needs only the current batch. Historical Contexts therefore retain linear ID state instead of cumulative array and Set snapshots. + ### Responsibility layers | Layer | Durable responsibility | Explicitly does not own | @@ -123,16 +125,16 @@ The Assembler does not use State reference equality to decide publication or pro | Return value | Behavior | |---|---| | `immediate` | Request a notification and flush in the current microtask | -| `animation-frame` | Coalesce high-frequency updates into materialization on the next frame | +| `animation-frame` | Coalesce high-frequency updates into materialization after three browser animation frames | | `none` | Do not schedule a flush for this Match; retain its State and dirty marker | Omitting `publication()` means `immediate`. Assistant token deltas and packed runs use `animation-frame`, invisible Inbox Contexts use `none`, and finals, dependency replays, and Location boundaries publish the latest result through an immediate path. -Every live delta within a frame still executes `update()`, while one historical packed run executes one batch `update()`. Only `buildViewNode()`, View Builder work, and React snapshot notification are coalesced; no fragments are lost. +Every live delta during the three-frame interval still executes `update()`, while one historical packed run executes one batch `update()`. Location-data publication, `buildViewNode()`, View Builder work, and React snapshot notification are coalesced; no fragments are lost. An immediate publication cancels a pending frame interval and flushes the latest State without delay. #### `buildLocationData(context, scope)` -`buildLocationData()` lets a Definition publish a read-only value derived from its State onto an engine-owned Step or Turn without exposing another business's mutable State. The Assembler always materializes `step` before `turn`, so Turn-level aggregation can read Step data updated in the same flush; it calls `buildViewNode()` only after all Location data is ready. +`buildLocationData()` lets a Definition publish a read-only value derived from its State onto an engine-owned Step or Turn without exposing another business's mutable State. The Assembler passes the preceding publication back to its owner, which returns that exact value when its business data is unchanged. The Assembler always materializes `step` before `turn`, so Turn-level aggregation can read Step data updated in the same flush; it calls `buildViewNode()` only after all Location data is ready. A Definition receives the `step` and `turn` scopes separately and may return one value or `null` in either phase. A value must identify the exact turn/step coordinates and use the Definition's `kind` as its key. The Assembler owns replacement and removal and rejects another Context that claims the same Location key. @@ -257,8 +259,7 @@ Page size, record packing, the number of history loads, and RAF coalescing affec | Business / `kind` | Stable ID | Start Match | Update Matches | State and cross-Context reads | |---|---|---|---|---| -| Next-turn Inbox / `inbox-next-turn` | Splice Event seq | Each `agent/inbox/spliced` targeting next-turn | None | Apply the current splice to the pending/claimed instantaneous state from `reader.previous(ownKind)` | -| Next-step Inbox / `inbox-next-step` | Splice Event seq | Each `agent/inbox/spliced` targeting next-step | None | Build the same per-instruction instantaneous state; Message reads its claimed set | +| Next-step Inbox / `inbox-next-step` | Splice Event seq | Each `agent/inbox/spliced` targeting next-step | None | Append message IDs to persistent splice state; materialize once per claim and expose the shared current claimed batch to Message | | Message / `input-message` | Message ID | Append-surface `user/message` | None | Use source for a context message, or read the nearest next-step Inbox to distinguish user from steering | | Request Prompt / `request-prompt` | Header Event seq | Each `request/header` | None | Read the preceding Request Prompt through Reader, retain the full prompt state, and classify system/tool changes | | Assistant / `assistant-step` | `turn:step` | `step/start` | Scalar or packed `assistant/chunk`, final `assistant/message`, and same-step Retry | Aggregate blocks, usage, first-token time, final evidence, and retry-hidden state, then publish same-key Step data | @@ -275,7 +276,7 @@ Page size, record packing, the number of history loads, and RAF coalescing affec | Business | `publication()` | Chat output | History and runtime behavior | |---|---|---|---| -| Inbox | `none` | No Node | Recompute instantaneous states along the Reader chain when prepend supplies earlier splices | +| Inbox | `none` | No Node | Recompute next-step ID state along the Reader chain when prepend supplies earlier splices; next-turn creates no Chat Context | | Message | Immediate by default | `user`, `steering`, or `context` | Window-gap repair can reclassify the same message key | | Request Prompt | Immediate by default | One `system-prompt` for every header carrying a non-empty system field | A step's first header anchors before its request messages; a later same-step series anchors after its surface rewrite; prepend of the preceding header can correct a partial-window anchor | | Assistant | RAF for scalar chunks and packed runs, immediate for final, none for pure usage/finish | Same-key `assistant-step` with running/settled/interrupted status | Scalar and packed reducers are equivalent; Matches support fallback without `step/start`; Location close produces interruption presentation | @@ -288,7 +289,7 @@ Page size, record packing, the number of history loads, and RAF coalescing affec | Deliverables | Immediate by default | No Node | Tool settlement incrementally updates Turn data; the Turn Tail extension slot reads produced files | | Fallback | Immediate by default | `unknown` JSON row | Covers only append-surface Events; an ordinary business that claimed but has not rendered an Event does not duplicate it | -Inbox demonstrates that every Event can be a start-only instantaneous-state Context; not every business requires a start/update pair. Reader links each state to the prior same-kind Context instead of inventing a lifecycle ID for the entire Inbox. +Inbox demonstrates that every Event can be a start-only instantaneous-state Context; not every business requires a start/update pair. Reader links each next-step state to the prior same-kind Context instead of inventing a lifecycle ID for the entire Inbox. The state itself shares immutable pending splice nodes and one current claimed-batch Set, while unconsumed next-turn input remains outside Conversation because no Chat or Trajectory classification reads it. Request Prompt demonstrates shared pure interpretation without shared target State: Chat and Trajectory call `inspectRequestPrompt()` from their own Definitions. The function canonicalizes the full header and classifies model-visible system/tool differences; each target then chooses its own output. Chat materializes every header carrying a non-empty system field, including `series` snapshots that repeat an unchanged header for an explicitly declared series or a post-replacement request, while Trajectory retains the complete request fact and its change classification. Ordinary append-only later Turns do not write another unchanged header. The first header in a Step follows the provider envelope rather than the header Event position: step one uses the owning Turn start and later steps use their Step start, placing the system field before the request's user-role messages; a later header in the same Step stays at its own Event after the surface rewrite that began the new series. When the preceding header is outside a partial window, a non-`initial` header stays at its own Event until prepend supplies that predecessor. Every header is a full snapshot, so a first loaded `resume`, `change`, or `series` header can render its system field without fabricating a comparison to unloaded history. @@ -306,25 +307,29 @@ Unknown fallback demonstrates Registry ownership: it handles only append-surface ## View Builder and React identity -[`ConversationViewRegistry`](../../../../packages/client/ui-conversation/src/client/conversation/view-registry.ts) creates an independent per-Session builder for each target. The Registry stores factories and shares no Session's ordering or caches. +[`ConversationViewRegistry`](../../../../packages/client/ui-conversation/src/client/conversation/view-registry.ts) stores an independent builder factory for each target and shares no Session's ordering or caches. -The Assembler calls `replace({ nodes, timeline })` on low-frequency complete replacements and `apply({ upserts, timeline })` for ordinary prepend/append flushes. Builders receive only final target Nodes already constructed by Definitions. +A shell selection or a target source's first subscriber adds that target to the Session's monotonic active-target set. The Assembler indexes each Context under its sole target but creates no builder, Node, or snapshot for an inactive target. First activation flushes pending target-neutral work, creates the builder, and calls `replace({ nodes, timeline })` once from that target's current Contexts. -[`ChatSnapshotBuilder`](../../../../packages/client/ui-chat/src/client/conversation-nodes/chat-snapshot-builder.ts) maintains `order`, a keyed `nodes` store, the turn/step `locations` index, `timeline`, and the `legacy` slice used by StatsLine and mirrored into top-level public compatibility fields. +The shell synchronously resolves the persisted selection when a Session binding becomes available, when a cached binding becomes current, or when the View roster changes, then explicitly activates that registered View or the Chat fallback. Tab and focus actions activate their resolved target before updating selection state. A blank Session does not render the View slot, and `ConversationSnapshot.activeTargets` derives only from materialized active snapshots without querying inactive target Contexts for activity. -Only a new key or a change to `anchorSeq`, visibility, or Location identity makes a Chat update structural. An ordinary content change does not rebuild `order`; the keyed Node store replaces only that key's value. +Ordinary prepend and append flushes call `apply({ upserts, timeline })` only for active targets. Complete window replacement and Registry rebuild call `replace()` only for active targets. Unsubscription does not remove a target, so returning to an opened View does not rebuild it. + +[`ChatSnapshotBuilder`](../../../../packages/client/ui-chat/src/client/conversation-nodes/chat-snapshot-builder.ts) maintains `order`, a keyed `nodes` store with identity-stable Node and Turn-process sources, the turn/step `locations` index, `timeline`, and the `legacy` slice used by StatsLine and mirrored into top-level public compatibility fields. + +Only a new key or a change to `anchorSeq`, visibility, or Location identity makes a Chat update structural. An ordinary content change does not rebuild `order`; the keyed Node store replaces that key's value and publishes only its source. The Turn-process projector recalculates cross-Node presentation only for a Turn whose structure, specification, or status changed, then publishes only that Turn's process sources. For a structural change, the Builder computes visible order from current store values and reuses unchanged index arrays by reference. Prepend may add earlier history keys, append may add a key at the tail or its business anchor, and ordering never renames existing keys. -[`ChatView`](../../../../packages/client/ui-chat/src/client/chat/ChatView.tsx) only traverses `order`. Each [`ChatNodeSeat`](../../../../packages/client/ui-chat/src/client/chat/ChatNodeSeat.tsx) remains in the same parent list under its Context key and dispatches the `'conversation.chat.node'` keyed slot by `node.kind`. +[`ChatView`](../../../../packages/client/ui-chat/src/client/chat/ChatView.tsx) only traverses `order` and resolves the two stable sources for each key. Each [`ChatNodeSeat`](../../../../packages/client/ui-chat/src/client/chat/ChatNodeSeat.tsx) remains in the same parent list under its Context key, subscribes only to its Node and Turn-process sources, and dispatches the `'conversation.chat.node'` keyed slot by `node.kind`. [`ChatNodeDataMap`](../../../../packages/client/ui-chat/src/client/contract/chat-nodes.ts) is a declaration-merged renderer payload registry. Each business module registers its own Definition and keyed renderer; `registerConversationNodes()` and `registerChatNodeRenderers()` only assemble those independent contributions and do not interpret business through a closed union or central switch. Built-ins live in `ui-chat`, and this type and registration boundary allows a business to move into an independent package without changing the Chat dispatcher. -The Chat entry in `conversation.view` registers `ChatNodeTurnDataInjected` once when it declares the `conversation.chat.node` child slot. `ChatNodeSeat` passes only the stable Node key as `hookContext`; the Slot renderer combines that key with `useSession` from the official standard props to construct `useTurnData(businessKey)`. Every keyed Chat renderer therefore reads strongly typed, read-only data from its own Node's Turn, and the Assistant renderer has no special injection authority. +The Chat entry in `conversation.view` registers `ChatNodeTurnDataInjected` once when it declares the `conversation.chat.node` child slot. `ChatNodeSeat` passes the Node's stable Turn data store as `hookContext`; the Slot renderer binds `useTurnData(businessKey)` directly to that store. Every keyed Chat renderer therefore reads strongly typed, read-only data from its own Node's Turn, and the Assistant renderer has no special injection authority. -Slot-level contextual Hooks and entry-owned `inject.hooks` remain independent paths. The latter continues to bind only registration-owned Observables. The former caches definitions by stable slot-inject-face identity and binds its factory and Hook per stable render occurrence. The selector inside `useTurnData()` returns only the current Node's `turn.data.get(key)`, so selector equality filters unrelated Session publications. +Slot-level contextual Hooks and entry-owned `inject.hooks` remain independent paths. The latter continues to bind only registration-owned Observables. The former caches definitions by stable slot-inject-face identity and binds its factory and Hook per stable render occurrence. `useTurnData()` subscribes to `turn.data.source(key)`, so another Location-data key or Session snapshot publication does not notify it. -The standard `useSession` remains available to every session-scoped slot renderer. `useTurnData()` narrows the common read path rather than acting as a permission sandbox. Whole-window statistics or arbitrary object indexes may still read the Session snapshot explicitly, but they are not modeled as current-Node Turn data. +The standard `useSession` remains available to every session-scoped slot renderer, although `ChatNodeSeat` needs neither it nor aggregate `useChat`. `useTurnData()` narrows the common read path rather than acting as a permission sandbox. Whole-window statistics or arbitrary object indexes may still read the Session snapshot explicitly, but they are not modeled as current-Node Turn data. Assistant streaming to final and Tool running to settled stay in one Seat while updating its data and necessary ordering properties. Settlement therefore does not reset component-local State through a parent move. @@ -345,15 +350,15 @@ SessionEventLike window -> Context matches + State + Location -> Definition.buildLocationData(step -> turn) -> StepLocation.data / TurnLocation.data - -> Definition.buildViewNode() for its declared target - -> target View Builder + -> Definition.buildViewNode() for each active target + -> active target View Builder -> chat: ChatSnapshotBuilder -> ChatView -> keyed ChatNodeSeat -> trajectory: TrajectorySnapshotBuilder -> stages/layout/table ``` ## Verification -Runtime tests pin Definition lifecycle registration, exact-ID append, update-before-start collection followed by forward replay after start, prepend identity, Reader window-gap repair, transitive dependencies, Location closure, Step→Turn data phase order, Location data replacement, publication cadence, illegal withdrawal, and per-target Builders. +Runtime tests pin Definition lifecycle registration, exact-ID append, update-before-start collection followed by forward replay after start, prepend identity, Reader window-gap repair, transitive dependencies, Location closure, Step→Turn data phase order, Location data replacement, publication cadence, illegal withdrawal, first-subscription activation, monotonic active targets, and per-target Builders. Conversation tests cover every built-in Chat Definition, Assistant Step data, Turn Tail and Deliverables Turn data, Chat ordering and structural sharing, selector isolation, Assistant and Tool running-to-settled identity, nested Code Dispatch, steering, Compaction, Retry, interruption, load-older anchoring, and slot dispatch. Trajectory tests cover its independently registered Message, Assistant, Tool, Compaction, Request-header, and boundary Definitions together with the preserved stage-oriented view model. @@ -385,10 +390,14 @@ History-path tests cover complete replace, non-overlapping prepend, complete-ran **Let a Location-data consumer read the provider's Context State directly.** Rejected: the consumer would depend on another business's mutable internal shape and could not express which Turn/Step owns the value. Declaration-merged data maps expose only the provider-selected read-only value and engine-owned coordinates. +**Cache every Definition's Location data by State identity.** Rejected because a Definition may mutate and return the same State object, and its Location data may also depend on Match Locations or values published by another Definition. Each Definition instead decides whether its business value changed and returns the preceding publication unchanged when it did not. + **Add generic `end()`, prepared, or window-reset lifecycles.** Rejected: businesses have different completion conditions, and a pagination gap is not a business lifecycle. Business Events update State, Location close triggers replay/build, and Reader dependencies own pagination invalidation. **Reuse one Event Definition across Chat and Trajectory by branching in `buildViewNode(target)`.** Rejected: the views require different business State and intermediate records, so a shared Definition would make each package carry the other's conditions and payloads. Separate target-owned Definitions keep those choices local while sharing the Assembler's ingestion and lifecycle contracts. +**Deactivate a target when its last subscriber leaves.** Rejected: returning to the View would repeatedly rebuild its complete snapshot. Subscription establishes first use; the target then stays incremental for the remaining Session lifetime. + **Add a generic layout model above final business Nodes.** Rejected: activity, tail candidacy, and layout enums would centralize current Chat business semantics in the engine again. Final Nodes carry renderer-required data directly and share only identity, ordering, and Location facts. **Register the Turn-data Hook only on the Assistant renderer.** Rejected: current-Node Location access is a common capability of the `conversation.chat.node` slot, not one business renderer. The parent Chat entry registers common inject once, and every keyed renderer shares the same strongly typed contract. @@ -405,9 +414,13 @@ Initial tail, older prepend, and live append share one set of Context invariants Append does not scan historical Contexts; prepend replays only Contexts whose Matches, Locations, or Reader answers actually changed. A structural Chat change may still recompute visible order and indexes, but does not rerun unrelated business folds or replace unchanged Node identity. -Separating State updates from publication cadence folds every live Assistant delta and each historical packed run while materializing at most once per animation frame. Step or Turn close and final Events can immediately publish the latest State. +Separating State updates from publication cadence folds every live Assistant delta and each historical packed run while materializing at most once per three animation frames. The Assistant view reads the same projection that the preceding Step Location phase installed. Turn Process returns its existing open data and Node for continuing Assistant chunks without deriving or encoding them again, and Turn Tail defers its complete-Match scan until `turn/end`. Step or Turn close and final Events immediately publish the latest State. -Steps and Turns are stable homes for cross-business aggregates. Turn Tail and Deliverables derive their values without renderer scans of global Nodes; slot-level `useTurnData()` narrows common reads to the current Node's Turn and uses selector equality to isolate unrelated updates. +An inactive target retains Definition State and a target Context index but no builder, materialized Nodes, or snapshot. The mounted built-in or third-party View activates its own target through normal subscription; previously opened targets continue receiving incremental updates. + +Steps and Turns are stable homes for cross-business aggregates. Turn Tail and Deliverables derive their values without renderer scans of global Nodes; slot-level `useTurnData()` narrows common reads to the current Node's Turn, and keyed Location sources isolate unrelated updates. + +Inbox Context retention grows with splice count and claimed message count rather than their cumulative prefixes. This removes duplicate state growth but does not deduplicate message content in durable Session events or bound the loaded event window. The cost is new Runtime contracts for Registry, Assembler, Location data, dependency replay, and per-target Builders, plus parent-owned common inject and per-occurrence `hookContext` in UI Slots. Definitions that consume Assistant deltas also maintain equivalent scalar and packed update branches. Definition authors must understand stable IDs, unique scalar starts, forward replay, Step→Turn publication order, read-only Reader access, and the prohibition on Node withdrawal. diff --git a/.agents/notes/implemented/architecture/2026-08-09-client-conversation-node-assembly.zh.md b/.agents/notes/implemented/architecture/2026-08-09-client-conversation-node-assembly.zh.md index be23b321b7..b06a92113f 100644 --- a/.agents/notes/implemented/architecture/2026-08-09-client-conversation-node-assembly.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-09-client-conversation-node-assembly.zh.md @@ -18,6 +18,8 @@ Client Runtime 提供 target-neutral 的 Conversation Node 组装引擎，业务 本 Note 保留实现后仍有价值的方案推导、逐业务适配、职责、算法和取舍。 +Chat 只注册 `next-step` Inbox Definition，因为消息分类是其唯一消费方；`next-turn` splice 仍是持久 Session input，但不会创建 Chat Context。Chat 与 Trajectory 各自维护 target 专属 next-step state。每次插入只把消息 ID 写入不可变 splice 节点。成功 claim 时只 materialize 一次 pending 链，以当前批次替换上一个 claimed Set，并让后续 Context 共享该 Set，直到下一次 claim。AgentLoop 会在领取下一批消息之前追加当前 claim 接纳的全部消息；被拒绝的 claim 不追加 `user/message`，因此后续分类只需当前批次。历史 Context 因而只保留线性 ID state，不再保留累计数组和 Set 快照。 + ### 责任分层 | 层 | 长期职责 | 明确不负责 | @@ -123,16 +125,16 @@ Assembler 不以 State 引用相等判断是否需要发布或传播。每次成 | 返回值 | 行为 | |---|---| | `immediate` | 请求当前 microtask 通知与 flush | -| `animation-frame` | 把多条高频更新合并到下一帧 materialize | +| `animation-frame` | 跨过三个浏览器 animation frame 后，把多条高频更新合并为一次 materialization | | `none` | 本 Match 不主动安排 flush，State 和 dirty 标记仍被保留 | 省略 `publication()` 等于 `immediate`。Assistant token delta 与 packed run 使用 `animation-frame`，不可见 Inbox Context 使用 `none`，final、依赖 replay 和 Location 边界会以 immediate 路径发布最新结果。 -一帧内的每条 live delta 仍执行 `update()`，一个历史 packed run 则执行一次 batch `update()`；合并的只是 `buildViewNode()`、View Builder 和 React snapshot 通知，不会丢失 fragment。 +三帧间隔内的每条 live delta 仍执行 `update()`，一个历史 packed run 则执行一次 batch `update()`；Location-data publication、`buildViewNode()`、View Builder 与 React snapshot 通知会合并执行，不会丢失 fragment。immediate publication 会取消等待中的帧间隔，并立即发布最新 State。 #### `buildLocationData(context, scope)` -`buildLocationData()` 让 Definition 把 State 的只读派生值发布到 Engine-owned Step 或 Turn，而不把另一个业务的可变 State 暴露出去。Assembler 在每次 materialize 中固定先处理 `step`、再处理 `turn`，因此 Turn 级聚合可以读取同一轮已经更新的 Step data；全部 Location data 就绪后才调用 `buildViewNode()`。 +`buildLocationData()` 让 Definition 把 State 的只读派生值发布到 Engine-owned Step 或 Turn，而不把另一个业务的可变 State 暴露出去。Assembler 会把前一次 publication 传回它的 owner；业务数据未变时，owner 原样返回该值。Assembler 在每次 materialize 中固定先处理 `step`、再处理 `turn`，因此 Turn 级聚合可以读取同一轮已经更新的 Step data；全部 Location data 就绪后才调用 `buildViewNode()`。 Definition 分别收到 `step` 和 `turn` scope，可以在任一阶段返回一个值或 `null`。返回值必须声明准确的 turn/step 坐标，并使用与 Definition `kind` 相同的 key；Assembler 拥有替换和移除，并拒绝另一个 Context 占用同一 Location key。 @@ -257,8 +259,7 @@ Chat `order` 的结构性变化仍可能重排当前可见 key；纯 data 更新 | 业务 / `kind` | 稳定 ID | start Match | update Matches | State 与跨 Context 读取 | |---|---|---|---|---| -| Next-turn Inbox / `inbox-next-turn` | splice Event seq | 每条目标为 next-turn 的 `agent/inbox/spliced` | 无 | 从 `reader.previous(ownKind)` 的 pending/claimed 瞬间态应用当前 splice | -| Next-step Inbox / `inbox-next-step` | splice Event seq | 每条目标为 next-step 的 `agent/inbox/spliced` | 无 | 同样形成逐指令瞬间态，claimed 集合供 Message 读取 | +| Next-step Inbox / `inbox-next-step` | splice Event seq | 每条目标为 next-step 的 `agent/inbox/spliced` | 无 | 把消息 ID 追加到持久 splice state；每次 claim 只 materialize 一次，并向 Message 暴露共享的当前 claimed batch | | Message / `input-message` | message ID | append-surface `user/message` | 无 | 根据 source 生成 context message，或读取最近 next-step Inbox 判断 user/steering | | Request Prompt / `request-prompt` | header Event seq | 每条 `request/header` | 无 | 通过 Reader 读取前一条 Request Prompt，保留完整 prompt 状态，并判定 system/tool 变化 | | Assistant / `assistant-step` | `turn:step` | `step/start` | scalar 或 packed `assistant/chunk`、final `assistant/message`、同 step Retry | 聚合 blocks、usage、首 token 时间、final 和 retry 隐藏状态，并发布同 key Step data | @@ -275,7 +276,7 @@ Chat `order` 的结构性变化仍可能重排当前可见 key；纯 data 更新 | 业务 | `publication()` | Chat 产物 | 历史分页与运行时行为 | |---|---|---|---| -| Inbox | `none` | 不生成 Node | prepend 补前序 splice 时沿 Reader 链重算瞬间态 | +| Inbox | `none` | 不生成 Node | prepend 补前序 splice 时沿 Reader 链重算 next-step ID state；next-turn 不创建 Chat Context | | Message | 默认 immediate | `user`、`steering` 或 `context` | window gap 修复可让同一 message key 重新分类 | | Request Prompt | 默认 immediate | 每条带非空 system 字段的 header 都生成一个 `system-prompt` | Step 首条 header 锚定在请求消息之前；同 step 后续序列锚定在表层改写之后；prepend 补入前序 header 后可纠正部分窗口的锚点 | | Assistant | scalar chunk 与 packed run 为 RAF，final immediate，纯 usage/finish 为 none | 同 key `assistant-step`，状态为 running/settled/interrupted | scalar 与 packed reducer 等价；缺 `step/start` 可先用 Matches fallback；Location close 生成中断表现 | @@ -288,7 +289,7 @@ Chat `order` 的结构性变化仍可能重排当前可见 key；纯 data 更新 | Deliverables | 默认 immediate | 不生成 Node | Tool 结算增量更新所属 Turn data，Turn Tail 扩展槽读取 produced files | | Fallback | 默认 immediate | `unknown` JSON row | 只兜底 append surface，普通业务已认领但暂不可见时不会重复生成 | -Inbox 展示了“每条 Event 都是一个 start-only 瞬间态 Context”，不是所有业务都需要 start/update 配对。它通过 Reader 与前一个同 kind Context 形成连续 fold，而非给整个 Inbox 人工制造生命周期 ID。 +Inbox 展示了“每条 Event 都是一个 start-only 瞬间态 Context”，不是所有业务都需要 start/update 配对。每个 next-step state 通过 Reader 与前一个同 kind Context 形成连续 fold，而非给整个 Inbox 人工制造生命周期 ID。state 自身共享不可变 pending splice 节点和一个当前 claimed-batch Set；未消费的 next-turn input 不进入 Conversation，因为 Chat 与 Trajectory 都不读取它来分类。 Request Prompt 展示了如何在不共享 target State 的前提下共用纯解释逻辑：Chat 与 Trajectory 各自在自己的 Definition 中调用 `inspectRequestPrompt()`。该函数规范化完整 header，并判定面向模型的 system/tool 差异；随后每个 target 自行选择产物。Chat 会物化每条带非空 system 字段的 header，包括为显式声明的序列或表层替换后的请求重复未变 header 的 `series` 快照；Trajectory 则保留完整请求事实及其变化分类。普通的仅追加后续 Turn 不会再次写入未变 header。一个 Step 中的首条 header 遵循提供方信封，而不是 header Event 位置：step one 使用所属 Turn start，后续 step 使用各自的 Step start，把 system 字段放到该请求的 user-role 消息之前；同一 Step 的后续 header 保留在开启新序列的表层改写之后。部分窗口未包含前序 header 时，非 `initial` header 会保留在自身 Event，直到 prepend 补入该前序 header。每条 header 都是完整快照，因此已加载窗口中的首条 `resume`、`change` 或 `series` header 无需凭空构造与未加载历史的比较，也能渲染其 system 字段。 @@ -306,25 +307,29 @@ Unknown fallback 展示了 Registry ownership：fallback 只处理没有任何 ## View Builder 与 React identity -[`ConversationViewRegistry`](../../../../packages/client/ui-conversation/src/client/conversation/view-registry.ts) 为每个 target 创建独立的 per-Session builder。Registry 保存 factory，不共享某个 Session 的排序或缓存。 +[`ConversationViewRegistry`](../../../../packages/client/ui-conversation/src/client/conversation/view-registry.ts) 为每个 target 保存独立的 builder factory，不共享某个 Session 的排序或缓存。 -Assembler 低频完整替换时调用 `replace({ nodes, timeline })`；普通 prepend/append flush 调用 `apply({ upserts, timeline })`。Builder 只接收 Definition 已构造完成的 target Nodes。 +shell 选择或 target source 的首个 subscriber 会把该 target 加入 Session 单调增长的 active-target set。Assembler 按唯一 target 索引每个 Context，但不会为 inactive target 创建 builder、Node 或 snapshot。首次激活会 flush 尚未发布的 target-neutral 工作、创建 builder，并从该 target 的当前 Context 调用一次 `replace({ nodes, timeline })`。 -[`ChatSnapshotBuilder`](../../../../packages/client/ui-chat/src/client/conversation-nodes/chat-snapshot-builder.ts) 维护 `order`、keyed `nodes` store、turn/step `locations` index、`timeline`，以及由 StatsLine 使用并镜像到顶层公共兼容字段的 `legacy` slice。 +Session binding 可用、缓存的 binding 成为 current 或 View roster 变化时，shell 会同步解析持久化选择，再显式激活已注册的偏好 View 或 Chat fallback。Tab 与 focus action 在更新选择状态前先激活解析出的 target。blank Session 不渲染 View slot；`ConversationSnapshot.activeTargets` 只从已物化的 active snapshot 派生，不查询 inactive target Context 的 activity。 -Chat 结构变化只由新 key、`anchorSeq`、visibility 或 Location identity 变化触发。普通内容变化不重建 `order`；keyed Node store 只替换该 key 的 value。 +普通 prepend 与 append flush 只对 active target 调用 `apply({ upserts, timeline })`。完整 window replace 与 Registry rebuild 只对 active target 调用 `replace()`。取消订阅不会移除 target，因此返回已打开的 View 不会重建。 + +[`ChatSnapshotBuilder`](../../../../packages/client/ui-chat/src/client/conversation-nodes/chat-snapshot-builder.ts) 维护 `order`、带身份稳定 Node 与 Turn-process source 的 keyed `nodes` store、turn/step `locations` index、`timeline`，以及由 StatsLine 使用并镜像到顶层公共兼容字段的 `legacy` slice。 + +Chat 结构变化只由新 key、`anchorSeq`、visibility 或 Location identity 变化触发。普通内容变化不重建 `order`；keyed Node store 只替换该 key 的 value 并发布其 source。Turn-process projector 仅为结构、规格或状态发生变化的 Turn 重算跨 Node 呈现，再只发布该 Turn 的 process source。 Builder 遇到结构变化时从 store 的当前 values 计算 visible order，并按未变化引用复用索引数组。Prepend 可以增加前部历史 key，append 可以增加尾部或按业务 anchor 落位，既有 key 不因排序变化而重命名。 -[`ChatView`](../../../../packages/client/ui-chat/src/client/chat/ChatView.tsx) 只遍历 `order`。每个 [`ChatNodeSeat`](../../../../packages/client/ui-chat/src/client/chat/ChatNodeSeat.tsx) 以 Context key 固定在同一个父列表中，并按 `node.kind` 分发 `'conversation.chat.node'` keyed slot。 +[`ChatView`](../../../../packages/client/ui-chat/src/client/chat/ChatView.tsx) 只遍历 `order`，并为每个 key 解析两份稳定 source。每个 [`ChatNodeSeat`](../../../../packages/client/ui-chat/src/client/chat/ChatNodeSeat.tsx) 以 Context key 固定在同一个父列表中，只订阅自身的 Node 与 Turn-process source，并按 `node.kind` 分发 `'conversation.chat.node'` keyed slot。 [`ChatNodeDataMap`](../../../../packages/client/ui-chat/src/client/contract/chat-nodes.ts) 是 declaration-merged 的 renderer payload registry。每个业务模块分别注册自己的 Definition 和 keyed renderer；`registerConversationNodes()` 与 `registerChatNodeRenderers()` 只负责装配这些独立贡献，不通过 closed union 或中心 switch 解释业务。内建实现位于 `ui-chat`，且该类型和注册边界允许业务迁入独立 package 而不修改 Chat dispatcher。 -`conversation.view` 的 Chat entry 在声明 `conversation.chat.node` child slot 时统一注册 `ChatNodeTurnDataInjected`。`ChatNodeSeat` 只把稳定 Node key 作为 `hookContext` 传给 slot；Slot renderer 用官方 standard props 中的 `useSession` 和该 key 构造 `useTurnData(businessKey)`，因此每个 keyed Chat renderer 都能读取自己 Node 所属 Turn 的强类型只读 data，Assistant renderer 不拥有特殊注入权限。 +`conversation.view` 的 Chat entry 在声明 `conversation.chat.node` child slot 时统一注册 `ChatNodeTurnDataInjected`。`ChatNodeSeat` 把 Node 所属 Turn 的稳定 data store 作为 `hookContext` 传给 slot；Slot renderer 直接在该 store 上绑定 `useTurnData(businessKey)`，因此每个 keyed Chat renderer 都能读取自己 Node 所属 Turn 的强类型只读 data，Assistant renderer 不拥有特殊注入权限。 -Slot-level contextual Hook 与 entry-owned `inject.hooks` 是两条独立路径。后者继续只绑定 registration-owned Observable；前者按稳定 slot inject face 缓存定义，并按稳定 render occurrence 绑定 factory 和 Hook。`useTurnData()` 内部 selector 只返回当前 Node 的 `turn.data.get(key)`，无关 Session publication 会被 selector equality 截断。 +Slot-level contextual Hook 与 entry-owned `inject.hooks` 是两条独立路径。后者继续只绑定 registration-owned Observable；前者按稳定 slot inject face 缓存定义，并按稳定 render occurrence 绑定 factory 和 Hook。`useTurnData()` 订阅 `turn.data.source(key)`，其他 Location-data key 或 Session snapshot 的发布不会通知它。 -标准 `useSession` 仍属于所有 session-scoped slot renderer 的公开能力，`useTurnData()` 是收窄常见读取方式而不是权限沙箱。全窗口统计或任意对象索引仍可显式使用 Session snapshot；它们不能伪装成“当前 Node 的 Turn data”。 +标准 `useSession` 仍属于所有 session-scoped slot renderer 的公开能力，但 `ChatNodeSeat` 不再需要它或聚合 `useChat`。`useTurnData()` 是收窄常见读取方式而不是权限沙箱。全窗口统计或任意对象索引仍可显式使用 Session snapshot；它们不能伪装成“当前 Node 的 Turn data”。 Assistant streaming 到 final、Tool running 到 settled 始终留在同一个 Seat，只更新 data 和必要的排序属性。结算不会因跨 parent 移动而重置组件内部 State。 @@ -345,15 +350,15 @@ SessionEventLike window -> Context matches + State + Location -> Definition.buildLocationData(step -> turn) -> StepLocation.data / TurnLocation.data - -> Definition.buildViewNode() for its declared target - -> target View Builder + -> Definition.buildViewNode() for each active target + -> active target View Builder -> chat: ChatSnapshotBuilder -> ChatView -> keyed ChatNodeSeat -> trajectory: TrajectorySnapshotBuilder -> stages/layout/table ``` ## 验证 -Runtime tests 固定 Definition 生命周期注册、exact-ID append、update-before-start 收集与 start 后正序 replay、prepend identity、Reader window-gap 修复、传递依赖、Location closure、Step→Turn data phase order、Location data replacement、publication cadence、非法撤回和 per-target Builder。 +Runtime tests 固定 Definition 生命周期注册、exact-ID append、update-before-start 收集与 start 后正序 replay、prepend identity、Reader window-gap 修复、传递依赖、Location closure、Step→Turn data phase order、Location data replacement、publication cadence、非法撤回、首次订阅 activation、单调 active target 和 per-target Builder。 Conversation tests 覆盖全部内建 Chat Definition、Assistant Step data、Turn Tail 与 Deliverables Turn data、Chat 排序和结构共享、selector isolation、Assistant/Tool running-to-settled identity、nested Code Dispatch、steering、Compaction、Retry、interruption、load-older anchoring 和 slot dispatch。Trajectory tests 则覆盖它独立注册的 Message、Assistant、Tool、Compaction、Request-header 与 boundary Definition，以及继续保留的 stage-oriented view model。 @@ -385,10 +390,14 @@ Assembled Web snapshot、GUI 和浏览器场景覆盖真实 plugin graph。浏 **让 Location data 消费者直接读取提供方 Context State。** 拒绝：消费者会依赖另一个业务的可变内部形状，也无法表达值属于哪个 Turn/Step。declaration-merged data map 只公开提供方选择发布的只读值和 Engine-owned 坐标。 +**按 State identity 缓存每个 Definition 的 Location data。** 拒绝：Definition 可以原地修改并返回同一个 State 对象，其 Location data 也可能依赖 Match Location 或其他 Definition 发布的 value。各 Definition 改为自行判断业务值是否变化；未变化时原样返回前一次 publication。 + **增加通用 `end()`、prepared 或 window reset 生命周期。** 拒绝：不同业务完成条件不同，分页缺口也不是业务生命周期。业务 Event 更新 State，Location close 触发 replay/build，Reader dependency 负责补页失效。 **在同一个 Event Definition 内通过 `buildViewNode(target)` 为 Chat 与 Trajectory 分支。** 拒绝：两种视图需要不同的业务 State 与中间记录，共用 Definition 会迫使每个 package 携带另一边的条件与 payload。target 自有的 Definition 把这些选择留在本地，同时复用 Assembler 的摄入与生命周期约定。 +**最后一个 subscriber 离开时停用 target。** 拒绝：返回该 View 会反复重建完整 snapshot。订阅只确认首次使用；随后 target 在 Session 剩余生命周期中保持增量更新。 + **在最终业务 Node 上再叠一层通用 layout model。** 拒绝：activity、tail candidacy 和 layout enum 会把当前 Chat 的业务语义重新集中到引擎。最终 Node 直接携带 renderer 所需 data，只共享 identity、排序和 Location 事实。 **只在 Assistant renderer 注册 Turn data Hook。** 拒绝：访问当前 Node Location 是 `conversation.chat.node` slot 的公共能力，不属于某个业务 renderer。父 Chat entry 注册一次 common inject，所有 keyed renderer 共享同一强类型约定。 @@ -405,9 +414,13 @@ Host 业务 package 把自己的持久 Event 成员 declaration-merge 到 `@deep Append 不扫描历史 Context；prepend 只 replay Match、Location 或 Reader 答案真正受影响的 Context。Chat 结构变化仍可能重算 visible order 和索引，但不会重跑无关业务 fold 或替换未变化 Node identity。 -State 更新与发布频率分离后，Assistant 的每条 live delta 与每个历史 packed run 都会被 fold，同时每 animation frame 最多 materialize 一次。step/turn close 和 final 可立即发布最新 State。 +State update 与 publication cadence 分离后，Assistant 的每条 live delta 与每个历史 packed run 都会被 fold，同时每三个 animation frame 最多 materialize 一次。Assistant view 读取前置 Step Location 阶段刚写入的同一 projection。Turn Process 对持续 Assistant chunk 直接返回已有 open data 和 Node，不再重复派生或编码；Turn Tail 到 `turn/end` 才执行完整 Match 扫描。Step/Turn close 与 final Event 会立即发布最新 State。 -Step/Turn 是业务间共享聚合的稳定宿主。Turn Tail 和 Deliverables 无需由 renderer 扫描全局 Nodes 即可派生值；Slot-level `useTurnData()` 把常见读取限制到当前 Node 所属 Turn，并通过 selector equality 隔离无关更新。 +inactive target 会保留 Definition State 和 target Context 索引，但不保留 builder、已物化 Node 或 snapshot。已挂载的内建或第三方 View 通过正常订阅激活自己的 target；已经打开的 target 则继续接收增量更新。 + +Step/Turn 是业务间共享聚合的稳定宿主。Turn Tail 和 Deliverables 无需由 renderer 扫描全局 Nodes 即可派生值；Slot-level `useTurnData()` 把常见读取限制到当前 Node 所属 Turn，并通过 keyed Location source 隔离无关更新。 + +Inbox Context 的保留量随 splice 数和已 claim 消息数增长，不再随其累计前缀增长。该结构消除了重复 state 增长，但不会对持久 Session event 中的消息正文去重，也不会限制已加载 event window。 代价是 Runtime 新增 Registry、Assembler、Location data、依赖重放和 per-target Builder 契约，UI Slots 也新增 parent-owned common inject 与 per-occurrence `hookContext`。消费 Assistant delta 的 Definition 还需要维护等价的 scalar 与 packed update 分支。Definition 作者必须理解稳定 ID、唯一 scalar start、正序 replay、Step→Turn 发布顺序、只读 Reader 和 Node 不撤回规则。 diff --git a/.agents/notes/implemented/architecture/2026-08-10-fork-children-stay-one-shot.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-10-fork-children-stay-one-shot.i18n.yaml index 19a6df185d..667d5cf2de 100644 --- a/.agents/notes/implemented/architecture/2026-08-10-fork-children-stay-one-shot.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-10-fork-children-stay-one-shot.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-10-fork-children-stay-one-shot.md -2026-08-10-fork-children-stay-one-shot.md: ba1e99c4d78d14230a2199cd7fb3c3eb9d3ad754 -2026-08-10-fork-children-stay-one-shot.zh.md: 0d4e214be952f5a4d96c296f38a5b0dbfbd85c86 +2026-08-10-fork-children-stay-one-shot.md: 0edf6ca9347dcf57313365e570daed22589f69fd +2026-08-10-fork-children-stay-one-shot.zh.md: e832846d4018cf328e02c693bf21fe47a401d320 diff --git a/.agents/notes/implemented/architecture/2026-08-10-fork-children-stay-one-shot.md b/.agents/notes/implemented/architecture/2026-08-10-fork-children-stay-one-shot.md index ba1e99c4d7..0edf6ca934 100644 --- a/.agents/notes/implemented/architecture/2026-08-10-fork-children-stay-one-shot.md +++ b/.agents/notes/implemented/architecture/2026-08-10-fork-children-stay-one-shot.md @@ -1,4 +1,4 @@ -# Agent Note: Forked children stay one-shot +# Agent Note: Forked children preserve the parent request prefix Status: implemented @@ -6,44 +6,36 @@ English | [中文](2026-08-10-fork-children-stay-one-shot.zh.md) ## Problem -Fork's only difference from spawn is that the child Session is seeded with the parent's completed-turn prefix ([subagent-fork-in-process](../../../../packages/subagent/subagent-fork-in-process/README.md)). That seed costs real tokens — the inherited history is re-sent in every child request — and its one concrete payoff is provider-side prefix reuse: under the same provider and model, a child request whose leading bytes are identical to the parent's re-prefills none of the shared span. Anything a child scope adds *ahead* of the inherited history spends that payoff, because reuse stops at the first differing byte. +Fork differs from spawn by seeding the child Session with the parent's completed-turn prefix. That seed costs tokens, and its intended payoff is provider-side prefix reuse: under the same provider and model, a child request whose leading bytes match the parent's does not prefill the shared span again. A child-only system-prompt section or tool schema ahead of the inherited history defeats that payoff. -The child-scoped `report` return channel is now the largest such addition, and since [the report obligation](../feature/2026-08-06-continuable-child-report-obligation.md) it is two deltas rather than one: the `report` tool schema and the `tool:report` system-prompt section. Both live in the request head — the system block and the tool block precede every message — so a continuable forked child invalidates reuse before the first inherited turn and re-prefills the whole transcript it was forked to reuse. That composition pays fork's duplication cost and collects none of its benefit, while the parent still holds a reusable prefix the child could have shared. +The earlier shipped composition avoided this mismatch by keeping forked children one-shot. That restriction was a consequence of the former child-only return tool, not an intrinsic property of continuable fork. ## Decision -Every shipped composition inherits the fork delegation tool's `backgroundMode: one-shot` from the [base bundle](../../../../packages/bundle/base/cordis.patch.yml). The base bundle leaves `run_in_background` available because it also mounts the task service needed to settle background work. +The model-facing `send_message` tool is registered globally for every Agent in a composition. A continuable forked child therefore receives the same tool name, description, schema, and ordering as its parent. Its initial task is appended after the inherited Session seed, and the task includes the direct parent id plus guidance to return results with `send_message({ agent_id, message })` when that tool is visible to the child. -One-shot children — foreground and background alike — are created through `SubagentRuntime.start()`, which never enters the continuable activation-setup registry, so neither `report` nor its prompt section is installed. A forked one-shot child's system prompt and tool schemas therefore equal its parent's, apart from the `persona` and `toolFilter` deltas a deployment opts into per delegation tool. +The base and headless compositions retain one-shot fork as their conservative lifecycle policy. The `cordis`, `standard`, and `ptc` CLI presets may bind fork to the continuable lifecycle because that binding no longer inserts child-only request-head fields. `ForkInProcessProvider.prepareContinuable()` and `ctx.subagents.startContinuable()` remain the implementation seam for those presets. -`spawn` keeps `backgroundMode: continuable`. Continuable children and the report obligation ship unchanged for the provider whose child starts with no inherited prefix to protect, so this decision costs the report channel nothing. - -### The restriction is composition, not code - -`ForkInProcessProvider.prepareContinuable` stays implemented and `ctx.subagents.startContinuable()` still accepts `fork`; only the shipped `cordis.yml` rows changed. `tool-subagent` knows both the provider's `inheritsParentContext` and its own `backgroundMode` at mount, so a load-time rejection of the pair was available and is deliberately not added: the pair is not wrong in general. It is wrong only while a child-scope delta precedes inherited history, and the package that creates that delta — [`dsh-tool-subagent-report`](../../../../packages/subagent/tool-subagent-report/README.md) — is separately installable and, by its own design, invisible to `tool-subagent`. A deployment that omits the report package can run continuable forked children with the prefix intact. Encoding one roster's consequence as a delegation-tool invariant would make the tool assert something it cannot observe. - -The reintroduction condition is recorded as a `TODO(fork-continuable-prefix-reuse)` marker on `prepareContinuable` itself, the one method the shipped compositions do not call, and tracked as issue #2124: continuable fork reopens when a child's system prompt and tool schemas can match its parent's byte for byte. +Byte-identical prefix reuse is qualified by explicit deployment choices. A fork delegation that applies a child persona or `toolFilter` may still change the request head. In particular, filtering out `send_message` removes both the schema and the return guidance from the child; the runtime does not bypass an explicit allow-list. ## Alternatives considered -**Reject `inheritsParentContext` + `continuable` at mount.** A loud load-time failure would prevent silent reintroduction, which is what the configuration change cannot do. Rejected because the delegation tool cannot see the report package and the combination is legitimate without it; the invariant would be false for a deployment that never installs a child-scope delta, and `tool-subagent` would be asserting a fact owned by the roster. +**Keep every fork one-shot.** This preserves the prefix but unnecessarily gives up durable, multi-turn forked children after the child-only schema difference is gone. -**Stop mounting the fork provider at all.** This was the broader form of the restriction. Rejected because foreground fork *is* the prefix-reusing case and is untouched by the report channel, so a full ban gives up the capability without buying anything the one-shot binding does not already buy — and would leave no shipped composition exercising session seeding. +**Install a child-only return alias.** A recipient-free alias would make child calls shorter, but it would recreate a tool-schema and prompt delta before inherited history and duplicate the adjacent-Agent operation. -**Ship continuable forked children and accept the loss.** Rejected because the loss is total rather than marginal: reuse breaks ahead of the inherited history, so the child pays full prefill on a transcript it duplicated for the sole purpose of not paying it. A deployment that wants a long-lived child with no inherited context already has `spawn`. +**Add the return instruction to the system prompt.** This would place child-only bytes ahead of inherited messages. Appending it to the initial user task preserves the inherited prefix and keeps the parent id next to the task that needs it. -**Make `report` visible to every Agent.** A global registration would restore byte-identical prefixes by giving parent and child the same schema and section. Rejected because roots, one-shot children, remote children, and agentless callers would advertise a tool with no derivable recipient, and execution-time rejection would make schema visibility disagree with authority — the scope-local decision the [report tool Agent Note](../feature/2026-07-30-continuable-subagent-report-tool.md) already settled. - -**Install the child-scope deltas after the inherited history.** Rejected as unrepresentable: the system prompt and the tool schemas are request-head structures in every provider's wire format, so no ordering within them can place a child-only addition behind the message list. +**Ignore an explicit child `toolFilter`.** Structural return tools previously bypassed the child allow-list. Rejected because a declared tool restriction must determine both schema visibility and guidance; hidden authority would make the model-facing roster inaccurate. ## Consequences -- No shipped composition creates a continuable forked child; `subagent_fork` returns a result to its caller's turn, and `send_message` addresses only spawned children. -- A forked child's request prefix stays byte-identical to its parent's unless the deployment configures `persona` or `toolFilter` on the fork delegation tool, so the token cost of seeding buys provider-side reuse again. -- The fork provider's continuable path has no production caller and no assembled-composition coverage. It keeps its package-level tests, and the seam still accepts it, so a bundle or `--patch` overlay can reintroduce it with no code change and no warning. -- `subagent_fork`'s model-visible schema changes: the continuable background wording is replaced by the one-shot task wording in the base bundle, and disappears entirely from the two examples. The affected keyless snapshot tool-schema sidecars are re-recorded in the same change. -- The report obligation's reach narrows to spawned children in shipped deployments. Its default `next-step` scheduling, authority model, and coverage remain independent of fork composition. +- Parent and continuable-fork child expose byte-identical ordered tool schemas when the delegation does not request a persona or tool filter. +- The inherited Session seed precedes the child's initial task and return guidance. +- The base and headless profiles keep one-shot fork, while selected CLI presets exercise continuable fork without a child-only request-head addition. +- A child sends zero or more messages to its direct parent explicitly; its final answer is not implicitly copied. The manager-owned settlement notice remains unconditional and separate. +- Keyless snapshots and package tests pin schema equality, inherited-history ordering, parent-id guidance, and child-to-parent delivery through the same `send_message` operation used in the other direction. ### Accepted risks -The constraint lives in three configuration files and a code comment, not in a gate. A future bundle row or profile patch can set `backgroundMode: continuable` on a fork tool and silently reintroduce the prefix loss; nothing fails loud. That is the accepted cost of not encoding one roster's consequence into `tool-subagent`. +Provider-side prefix reuse still depends on the selected provider and model and on the absence of explicit persona or tool-filter differences. The harness proves equality of its assembled request-head inputs, not a provider's cache behavior. diff --git a/.agents/notes/implemented/architecture/2026-08-10-fork-children-stay-one-shot.zh.md b/.agents/notes/implemented/architecture/2026-08-10-fork-children-stay-one-shot.zh.md index 0d4e214be9..e832846d40 100644 --- a/.agents/notes/implemented/architecture/2026-08-10-fork-children-stay-one-shot.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-10-fork-children-stay-one-shot.zh.md @@ -1,49 +1,41 @@ -# Agent Note: fork 出的 child 保持 one-shot +# Agent Note：Fork child 保留 parent 请求前缀 -Status: implemented +状态：已实现 [English](2026-08-10-fork-children-stay-one-shot.md) | 中文 ## 问题 -fork 与 spawn 的唯一区别是 child 的 Session 会以 parent 已完成轮次的前缀作为初始内容（见 [subagent-fork-in-process](../../../../packages/subagent/subagent-fork-in-process/README.zh.md)）。这份初始内容有实打实的 token 成本——继承的历史会在 child 的每次请求中重新发送——而它唯一确定的回报是提供方侧的前缀复用：在提供方与模型相同的前提下，起始字节与 parent 逐字节相同的 child 请求，无需为这段共享区间重新预填充。任何由 child 作用域添加在继承历史*之前*的内容都会消耗掉这份回报，因为复用在第一个不同字节处即告停止。 +fork 与 spawn 的差异在于：fork 会用 parent 已完成轮次的前缀作为 child Session 的种子。该种子会消耗 token，其预期收益是提供方侧的前缀复用：使用相同提供方和模型时，如果 child 请求的开头字节与 parent 相同，共享区段就无需再次预填充。任何位于继承历史之前、仅属于 child 的系统提示词 section 或工具 schema 都会破坏这项收益。 -作用域局部的 `report` 返回通道现在是此类添加中最大的一项，而自[report 义务](../feature/2026-08-06-continuable-child-report-obligation.zh.md)起它是两项而非一项增量：`report` 工具 schema，以及 `tool:report` 系统提示词 section。两者都位于请求头部——系统块与工具块先于所有消息——因此一个可继续的 fork child 会在第一条继承轮次之前就使复用失效，并重新预填充它当初 fork 就是为了复用的整份 transcript（文本记录）。这种组合付出了 fork 的复制成本却收不到它的收益，而 parent 手上仍握着一份 child 本可共享的可复用前缀。 +先前的随附组合通过把 fork child 保持为 one-shot 来避开这种不匹配。该限制源于原来的 child-only 返回工具，并非可继续 fork 的固有属性。 ## 决策 -所有交付组合都从 [base bundle](../../../../packages/bundle/base/cordis.patch.yml)继承 fork 委派工具的 `backgroundMode: one-shot`。base bundle 保留 `run_in_background`，因为它也挂载了结算后台工作所需的 task 服务。 +面向模型的 `send_message` 工具在组合中的每个 Agent 上全局注册。因此，可继续 fork child 获得与 parent 相同的工具名称、描述、schema 和顺序。其初始任务追加在继承的 Session 种子之后；当 child 可以看到该工具时，任务还会包含直接 parent id，以及使用 `send_message({ agent_id, message })` 返回结果的指引。 -one-shot child——前台与后台皆然——经由 `SubagentRuntime.start()` 创建，该路径从不进入可继续的 activation setup 注册表，因此 `report` 与它的提示词 section 都不会被安装。于是一个 fork 出的 one-shot child 的系统提示词与工具 schema 与其 parent 相同，只差部署逐个委派工具主动选择的 `persona` 与 `toolFilter` 增量。 +base 与 headless 组合保留 one-shot fork 作为其保守生命周期策略。`cordis`、`standard` 和 `ptc` CLI preset 可以把 fork 绑定为可继续生命周期，因为该绑定不再插入 child-only 请求头字段。`ForkInProcessProvider.prepareContinuable()` 与 `ctx.subagents.startContinuable()` 仍是这些 preset 使用的实现 seam。 -`spawn` 保持 `backgroundMode: continuable`。对于 child 起步时本就没有继承前缀需要保护的那个提供方，可继续 child 与 report 义务随附行为不变，因此本决策没有让 report 通道付出任何代价。 +逐字节相同的前缀复用受显式部署选择约束。配置 child persona 或 `toolFilter` 的 fork 委派仍可能改变请求头。尤其是过滤掉 `send_message` 时，child 会同时失去该 schema 与返回指引；运行时不会绕过显式 allow-list。 -### 该限制在于组合，不在于代码 +## 考虑过的替代方案 -`ForkInProcessProvider.prepareContinuable` 仍然实现完好，`ctx.subagents.startContinuable()` 也仍接受 `fork`；改动的只有随附的 `cordis.yml` 行。`tool-subagent` 在挂载时同时知道提供方的 `inheritsParentContext` 与自身的 `backgroundMode`，因此一个加载期拒绝该组合的检查是可行的，而这里刻意不加：该组合并非普遍错误。它只在某个 child 作用域增量位于继承历史之前时才是错的，而产生该增量的包——[`dsh-tool-subagent-report`](../../../../packages/subagent/tool-subagent-report/README.zh.md)——是独立安装的，并且按其自身设计对 `tool-subagent` 不可见。一个不安装 report 包的部署可以在前缀完好的前提下运行可继续的 fork child。把某一份插件清单的后果写成委派工具的不变量，会让该工具断言它无法观察到的事实。 +**让所有 fork 保持 one-shot。** 这能保留前缀，但 child-only schema 差异消失后，继续放弃持久且多轮的 fork child 已无必要。 -重新开放的条件记录为 `prepareContinuable` 方法上的 `TODO(fork-continuable-prefix-reuse)` 标记——随附组合不调用这个方法——并由 issue #2124 跟踪：当 child 的系统提示词与工具 schema 能与其 parent 逐字节一致时，可继续 fork 即可重新开放。 +**安装 child-only 返回别名。** 无需填写接收方的别名可以缩短 child 调用，但会在继承历史之前重新产生工具 schema 与提示词差异，并重复相邻 Agent 操作。 -## 备选方案 +**把返回指令放进系统提示词。** 这会在继承消息之前加入 child-only 字节。将其追加到初始用户任务，可以保留继承前缀，并让 parent id 紧邻需要它的任务。 -**在挂载时拒绝 `inheritsParentContext` 与 `continuable` 的组合。** 一次响亮的加载期失败可以阻止悄然的重新引入，而配置改动做不到这一点。否决的原因是委派工具看不到 report 包，且在没有它时该组合是合法的；对于从不安装任何 child 作用域增量的部署，这个不变量是假的，而 `tool-subagent` 会去断言一件由插件清单拥有的事实。 - -**干脆不挂载 fork 提供方。** 这是该限制更彻底的形式。否决的原因是前台 fork *正是*复用前缀的那种情形，且不受 report 通道影响，因此全面禁用会在不换来任何 one-shot 绑定尚未换来的东西的同时放弃该能力——并且随附组合将没有任何一个演练 session 初始内容。 - -**照常随附可继续的 fork child 并接受这份损失。** 否决的原因是这份损失是全额而非边际的：复用在继承历史之前就已中断，于是 child 为一份自己复制过来、目的恰恰是不必付费的 transcript 付了全额预填充。想要一个没有继承上下文的长期 child 的部署，本来就有 `spawn`。 - -**让 `report` 对每个 Agent 可见。** 全局注册会通过让 parent 与 child 拥有相同的 schema 与 section 来恢复逐字节相同的前缀。否决的原因是根 agent、one-shot child、远端 child 与无 agent 调用方都会宣告一件推导不出收件方的工具，而执行期拒绝会让 schema 可见性与权限彼此矛盾——这正是[report 工具 Agent Note](../feature/2026-07-30-continuable-subagent-report-tool.zh.md)已经定下的作用域局部决策。 - -**把 child 作用域增量安装到继承历史之后。** 否决的原因是它无法表达：在每个提供方的协议格式中，系统提示词与工具 schema 都是请求头部结构，因此它们内部的任何排序都无法把仅属于 child 的添加放到消息列表之后。 +**忽略显式 child `toolFilter`。** 结构性返回工具过去会绕过 child allow-list。否决该方案，因为声明的工具限制必须同时决定 schema 可见性与指引；隐藏权限会让面向模型的工具清单失真。 ## 后果 -- 没有任何随附组合会创建可继续的 fork child；`subagent_fork` 把结果返回给调用方的轮次，而 `send_message` 只寻址 spawn 出的 child。 -- 除非部署在 fork 委派工具上配置了 `persona` 或 `toolFilter`，fork child 的请求前缀与其 parent 逐字节相同，因此初始内容的 token 成本重新换来了提供方侧的复用。 -- fork 提供方的可继续路径没有生产调用方，也没有整体组装层面的覆盖。它保留自己的包内测试，seam 也仍然接受它，因此某个组合包或 `--patch` 覆盖层可以无需改动代码、也不会有任何警告地把它重新引入。 -- `subagent_fork` 面向模型的 schema 发生变化：base 组合包中可继续的后台措辞被 one-shot 的 task 措辞取代，在两个示例中则完全消失。受影响的无密钥快照工具 schema 伴随文件在同一次改动中重新记录。 -- 在随附部署中，report 义务的覆盖范围收窄到 spawn 出的 child。它的 `next-step` 默认调度、权限模型与覆盖仍独立于 fork 组合。 +- 未请求 persona 或工具过滤时，parent 与可继续 fork child 暴露逐字节相同且顺序一致的工具 schema。 +- 继承的 Session 种子位于 child 初始任务及其返回指引之前。 +- base 与 headless profile 保持 one-shot fork；选定的 CLI preset 会在没有 child-only 请求头增量的前提下使用可继续 fork。 +- child 显式向直接 parent 发送零条或多条消息；最终回答不会被隐式复制。管理器负责的结算通知仍然无条件执行，并且与 Agent 消息分离。 +- keyless snapshot 与包测试固定 schema 相等性、继承历史顺序、parent-id 指引，以及通过同一个 `send_message` 操作完成的 child-to-parent 投递。 -### 已接受的风险 +### 已接受风险 -该限制存在于三个配置文件与一处代码注释中，而不在门禁里。未来某个组合包行或 profile 补丁可以在 fork 工具上设置 `backgroundMode: continuable`，从而悄然重新引入前缀损失；没有任何东西会失败得很响亮。这就是不把某一份插件清单的后果写入 `tool-subagent` 所接受的代价。 +提供方侧前缀复用仍取决于选定的提供方和模型，以及是否不存在显式 persona 或工具过滤差异。harness 证明的是自己组装出的请求头输入相等，而不是提供方的缓存行为。 diff --git a/.agents/notes/implemented/architecture/2026-08-10-message-feedback-sidecar.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-10-message-feedback-sidecar.i18n.yaml index 884aa27385..44c31bc1cd 100644 --- a/.agents/notes/implemented/architecture/2026-08-10-message-feedback-sidecar.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-10-message-feedback-sidecar.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-10-message-feedback-sidecar.md -2026-08-10-message-feedback-sidecar.md: eb1f8786f9ec0363b3c98d5b80a796b9c0fc0b4c -2026-08-10-message-feedback-sidecar.zh.md: 573d3b30e3bd14b492925f4a00424db588736943 +2026-08-10-message-feedback-sidecar.md: d047bebf47f844a6d88932c7e19a3952f43d94cc +2026-08-10-message-feedback-sidecar.zh.md: 4ee1b861a8013dacfd5eba40d9e30b23237be2e5 diff --git a/.agents/notes/implemented/architecture/2026-08-10-message-feedback-sidecar.md b/.agents/notes/implemented/architecture/2026-08-10-message-feedback-sidecar.md index eb1f8786f9..d047bebf47 100644 --- a/.agents/notes/implemented/architecture/2026-08-10-message-feedback-sidecar.md +++ b/.agents/notes/implemented/architecture/2026-08-10-message-feedback-sidecar.md @@ -16,9 +16,9 @@ A sidecar keyed only by `SessionId` can outlive the log lifecycle it describes w Every usable row is bound to the inspected Session header identity `{createdAt, cwd}`, not merely its `SessionId`. A lifecycle mismatch is treated as absence: `list` returns no items, and `put` may replace the stale row with one bound to the current identity. An id reused with a different header identity therefore cannot inherit stale feedback. A fork receives its own Session identity and no sidecar copy: even when the fork seed contains the same assistant messages, feedback remains attached to the Session in which the human recorded it. -`put` accepts a target only when `SessionPersistence.inspect()` observes a non-empty, append-origin `assistant/message` with that `MessageId`. Replacement-origin messages, empty usage-only assistant records, and non-assistant targets are rejected. Inspection is the cold-safe authority: it neither publishes or resumes an Agent nor commits cold-log repair merely to validate feedback. A cold `listSnapshots()` preflight classifies definite absence; inspection failure for a catalogued Session remains an infrastructure failure. A request in the narrow live-detach-to-header-materialization interval can therefore return `session-not-found`, and the caller retries after retirement materialization. +`put` accepts a target only when the observed log — a live owner's in-memory events, else the durable log through a persistence read handle — contains a non-empty, append-origin `assistant/message` with that `MessageId`. Replacement-origin messages, empty usage-only assistant records, and non-assistant targets are rejected. Observation is cold-safe: it neither publishes or resumes an Agent nor commits cold-log repair merely to validate feedback. A cold `stat()` preflight classifies definite absence; a read failure for a catalogued Session remains an infrastructure failure. A request in the narrow live-detach-to-header-materialization interval can therefore return `session-not-found`, and the caller retries after retirement materialization. -Before `put` commits a sidecar row, it puts the target log behind a durability barrier. A matching live Session passes through the canonical `ctx.sessions.flush` checkpoint, then both live and cold paths are physically read from sequence zero through `SessionPersistence.readFrom`. The resulting observation's header identity and target are checked again. A missing flush participant, changed identity, vanished target, or physical-read failure prevents the sidecar write, so a committed feedback item never precedes the durable assistant message it references. +Before `put` commits a sidecar row, it puts the target log behind a durability barrier. A matching live Session passes through the canonical `ctx.sessions.flush` checkpoint, then both live and cold paths are physically re-read from sequence zero through a fresh persistence read handle. The resulting observation's header identity and target are checked again. A missing flush participant, changed identity, vanished target, or physical-read failure prevents the sidecar write, so a committed feedback item never precedes the durable assistant message it references. Each message item carries its own opaque version plus Host-assigned `createdAt` and `updatedAt` timestamps. `put` compares the caller's `ifVersion` only with the addressed item, so editing one message does not invalidate another. The comparison is strict even when the desired value already matches, preventing a stale request from crossing an ABA value cycle; a conflict returns the authoritative current item so callers can reconcile without a second read. A matching-version no-op preserves the version and timestamps, while a material update preserves `createdAt`, replaces the version, and keeps `updatedAt` from moving backward. An already-absent delete is likewise successful. Versions are tokens for equality, not counters callers may order or synthesize. diff --git a/.agents/notes/implemented/architecture/2026-08-10-message-feedback-sidecar.zh.md b/.agents/notes/implemented/architecture/2026-08-10-message-feedback-sidecar.zh.md index 573d3b30e3..4ee1b861a8 100644 --- a/.agents/notes/implemented/architecture/2026-08-10-message-feedback-sidecar.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-10-message-feedback-sidecar.zh.md @@ -16,9 +16,9 @@ Status: implemented 每条可用记录都绑定到经检查的 Session header 身份 `{createdAt, cwd}`，而不只是其 `SessionId`。生命周期不匹配按不存在处理：`list` 返回空条目，`put` 可以用绑定当前身份的新记录替换陈旧行。因此，以不同 header 身份复用的 id 不会继承陈旧反馈。fork 拥有自己的 Session 身份，且不复制伴随记录：即使 fork 种子包含相同的 assistant 消息，反馈仍只属于人类记录它的那个 Session。 -`put` 只接受由 `SessionPersistence.inspect()` 观测到的非空、append-origin `assistant/message`，且其 `MessageId` 必须与目标相同。replacement-origin 消息、仅承载 usage 的空 assistant 记录以及非 assistant 目标都会被拒绝。检查使用 cold-safe 权威路径：它不会仅为验证反馈而发布或恢复 Agent，也不会提交 cold 日志修复。cold 路径由 `listSnapshots()` 预检明确不存在；已进入目录的 Session 若检查失败，仍按基础设施故障处理。因此，请求若恰落在 live detach 到 header materialization 的极短窗口，可能返回 `session-not-found`，调用方在 retirement materialization 后重试。 +`put` 只在被观测的日志——live 持有者的内存事件，否则是经由持久化读句柄读取的持久日志——包含非空、append-origin 且 `MessageId` 与目标相同的 `assistant/message` 时才接受该目标。replacement-origin 消息、仅承载 usage 的空 assistant 记录以及非 assistant 目标都会被拒绝。观测是 cold-safe 的：它不会仅为验证反馈而发布或恢复 Agent，也不会提交 cold 日志修复。cold 路径由 `stat()` 预检明确不存在；已进入目录的 Session 若读取失败，仍按基础设施故障处理。因此，请求若恰落在 live detach 到 header materialization 的极短窗口，可能返回 `session-not-found`，调用方在 retirement materialization 后重试。 -`put` 提交伴随记录前，会先让目标日志通过 durability barrier。身份匹配的 live Session 经过权威 `ctx.sessions.flush` checkpoint，随后 live 与 cold 路径都会通过 `SessionPersistence.readFrom` 从序列零做物理复读。之后再次校验所得观测的 header 身份与目标。缺少 flush 参与方、身份变化、目标消失或物理读取失败都会阻止伴随记录写入，因此已提交反馈绝不会先于它引用的持久 assistant 消息。 +`put` 提交伴随记录前，会先让目标日志通过 durability barrier。身份匹配的 live Session 经过权威 `ctx.sessions.flush` checkpoint，随后 live 与 cold 路径都会通过新开的持久化读句柄从序列零做物理复读。之后再次校验所得观测的 header 身份与目标。缺少 flush 参与方、身份变化、目标消失或物理读取失败都会阻止伴随记录写入，因此已提交反馈绝不会先于它引用的持久 assistant 消息。 每个消息条目都携带自己的 opaque version，以及 Host 分配的 `createdAt` 和 `updatedAt` 时间戳。`put` 只把调用方的 `ifVersion` 与目标条目比较，因此编辑一条消息不会使另一条消息失效。即使目标值已经相同，比较仍然严格执行，从而防止陈旧请求穿过 ABA 值循环；冲突会返回权威当前条目，调用方无需二次读取即可协调。携带匹配 version 的无变化请求会保留 version 与时间戳；实质更新保留 `createdAt`、替换 version，并保证 `updatedAt` 不倒退。删除已经不存在的条目也同样成功。version 是只能做相等比较的 token，不是调用方可以排序或自行合成的计数器。 diff --git a/.agents/notes/implemented/architecture/2026-08-10-remote-event-delivery.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-10-remote-event-delivery.i18n.yaml index a1f574e789..0e4340ea02 100644 --- a/.agents/notes/implemented/architecture/2026-08-10-remote-event-delivery.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-10-remote-event-delivery.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-10-remote-event-delivery.md -2026-08-10-remote-event-delivery.md: 4b9c2f224e36fb97f798492c999726eb55bad214 -2026-08-10-remote-event-delivery.zh.md: 4ca1f8c244df4e985d1223b6c468b450be943abf +2026-08-10-remote-event-delivery.md: 1706cc525bd7707874353a6658ec2f52597044a9 +2026-08-10-remote-event-delivery.zh.md: ac79642e90791892870c8920e02867faea180b07 diff --git a/.agents/notes/implemented/architecture/2026-08-10-remote-event-delivery.md b/.agents/notes/implemented/architecture/2026-08-10-remote-event-delivery.md index 4b9c2f224e..1706cc525b 100644 --- a/.agents/notes/implemented/architecture/2026-08-10-remote-event-delivery.md +++ b/.agents/notes/implemented/architecture/2026-08-10-remote-event-delivery.md @@ -194,4 +194,4 @@ The few required Client symbols are mirrored on the test side: `scaffold.ts` exp - **Test-side mirrors can drift.** No mechanism compares mirrored Client constants under `apps/web/tests` with their source. Drift instead produces a selector mismatch. `apps/web/tests/README.md` records the review rule; a grep-level gate is deliberately omitted. - **Capabilities deliberately omitted.** Payload projection and redaction are unsupported, scopes other than Agent are unsupported, and ordinary notifications are not replayed. Recoverable state needs a query, cursor, or opening baseline; a waterfall is replayed only while its original Host invocation remains pending. - **Some Client packages remain in the Host graph.** Twelve projects, including `connection`, `runtime`, and `ui-slots`, remain reachable through unsplit `directory-picker-browse`/`-native` and `api/gateway → client/connection`. They compile and no longer pull in API Remotes' Client face, so this change does not split them. Direct `dsh-client-runtime/client` imports in two chat e2e files rely on Runtime's current presence in that graph rather than a general guarantee. -- **The invariant companion intentionally has no runtime check.** A prior revision asserted delivery form on the live event bus, coupling the companion to the allowlist and causing Rolldown to emit a third bundle chunk omitted by the mechanically derived publication list. The Host-face `TypertForwardableEventEntry` assertion already rejects those mismatches at compile time, so the companion is an explained empty installer. +- **The package intentionally publishes no invariant companion.** A prior revision asserted delivery form on the live event bus, coupling diagnostics to the allowlist and causing Rolldown to emit a third bundle chunk omitted by the mechanically derived publication list. The Host-face `TypertForwardableEventEntry` assertion already rejects those mismatches at compile time, and the package README records why no independent runtime relation remains. diff --git a/.agents/notes/implemented/architecture/2026-08-10-remote-event-delivery.zh.md b/.agents/notes/implemented/architecture/2026-08-10-remote-event-delivery.zh.md index 4ca1f8c244..ac79642e90 100644 --- a/.agents/notes/implemented/architecture/2026-08-10-remote-event-delivery.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-10-remote-event-delivery.zh.md @@ -196,4 +196,4 @@ Client 要求首项是带非空 `clientId` 与 `host.home` 的 `ready`；后续 - **测试侧镜像值可能漂移**：没有任何机制核对 `apps/web/tests` 中镜像的 client 常量与其源；安全网只是漂移会让选择器失配。规则写在 `apps/web/tests/README.md`，由 review 守；grep 级门禁经评估后刻意不做。 - **放弃的能力**：不支持投影或脱敏载荷，不支持 Agent 以外的 Scope，也不为普通通知提供重放。需要可靠恢复的状态必须拥有查询、cursor 或 opening baseline；waterfall 只重放仍处于同一次 Host 调用生命周期内的 pending request。 - **仍有 client 包留在 host 图里**：12 个工程（`connection`、`runtime`、`ui-slots` 等）经未拆分的 `directory-picker-browse`/`-native` 与 `api/gateway → client/connection` 仍可达 host 图。它们都能编译且不再牵连 api/remotes 的 client face，因此没有阻塞本次改动；拆分那些包能减少几个，但经评估后不做。两个 chat e2e 直接引 `dsh-client-runtime/client` 依赖 `runtime` 本来就在图里——属偶然而非保证。 -- **invariant companion 不做运行期检查**：早先的修订曾在活事件总线上断言投递形状（`thisArg === null`、`mode === 'emit'`），这让 companion 与名单值耦合，并使 rolldown 把它提成第三个 bundle chunk——而机械推导的发布文件清单并不携带它。host 面的 `TypertForwardableEvent` 断言在编译期已拒绝这两种偏离，因此该 companion 是一个带说明的空 installer。 +- **本包不发布 invariant companion**：早先的修订曾在活事件总线上断言投递形状（`thisArg === null`、`mode === 'emit'`），这让诊断逻辑与名单值耦合，并使 rolldown 把它提成第三个 bundle chunk——而机械推导的发布文件清单并不携带它。Host 面的 `TypertForwardableEventEntry` 断言已在编译期拒绝这些偏离，包 README 也记录了不再存在独立运行时关系的原因。 diff --git a/.agents/notes/implemented/architecture/2026-08-11-trajectory-conversation-context-assembly.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-11-trajectory-conversation-context-assembly.i18n.yaml index 9af53e60e4..321dd79463 100644 --- a/.agents/notes/implemented/architecture/2026-08-11-trajectory-conversation-context-assembly.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-11-trajectory-conversation-context-assembly.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-11-trajectory-conversation-context-assembly.md -2026-08-11-trajectory-conversation-context-assembly.md: 7d0aea2fc09f0f04bd5de923bee15c42a773489a -2026-08-11-trajectory-conversation-context-assembly.zh.md: 1909243da02093de1a5f1d0857a11d3742258492 +2026-08-11-trajectory-conversation-context-assembly.md: fd246922b0ecce146da00e212ecb8053a939017d +2026-08-11-trajectory-conversation-context-assembly.zh.md: 2568de889fc3b6c227d9faa51166186b388de18f diff --git a/.agents/notes/implemented/architecture/2026-08-11-trajectory-conversation-context-assembly.md b/.agents/notes/implemented/architecture/2026-08-11-trajectory-conversation-context-assembly.md index 7d0aea2fc0..fd246922b0 100644 --- a/.agents/notes/implemented/architecture/2026-08-11-trajectory-conversation-context-assembly.md +++ b/.agents/notes/implemented/architecture/2026-08-11-trajectory-conversation-context-assembly.md @@ -18,6 +18,8 @@ Trajectory registers target-owned Conversation Definitions and a `trajectory` Vi Each Definition belongs to one target. Chat and Trajectory may recognize the same durable Event family, but they keep separate State and final Node payloads. They share only the Assembler's exact-ID matching, ordered Matches, Location facts, Reader dependencies, publication scheduling, and replace/prepend/append lifecycle. +Trajectory's target source activates its View Builder on first subscription. Before that subscription, its Definitions maintain current State while the Assembler skips `buildViewNode()` and snapshot assembly. The target remains active after unsubscription, so later visits reuse the incrementally maintained snapshot. + The existing [Trajectory inspection ledger](../feature/2026-07-27-trajectory-inspection-ledger.md) remains the view model. The Trajectory Builder converts materialized target Nodes into its established `eventNodes`, Requests, Tool schemas, running calls, and Location map; layout, table virtualization, selection, Overview, and inspector behavior do not become generic Conversation contracts. ### Business Definitions @@ -40,7 +42,7 @@ Assistant chunks update only their `turn:step` Context. Content-bearing chunks r Trajectory reconstructs steering from durable inbox history, using the same identity rule as the [Chat steering decision](../feature/2026-08-04-web-context-source-and-steer-marks.md) without sharing Chat's final Node. -Each `agent/inbox/spliced` Event targeting `next-step` starts an invisible Context identified by its Event seq. Its `start()` reads the nearest earlier inbox Context, applies the splice, and stores the pending identities plus the cumulative set of claimed message IDs. A later user-origin `user/message` reads the nearest earlier inbox Context: a claimed ID produces a Steering Node, while every other user-origin message produces an ordinary User Node. +Each `agent/inbox/spliced` Event targeting `next-step` starts an invisible Context identified by its Event seq. Its `start()` reads the nearest earlier inbox Context, appends the splice to persistent pending-ID state, and materializes that state only when a claim replaces the current claimed batch. The AgentLoop appends every admitted message from one claim before it can claim another batch; a rejected claim appends no `user/message`. A later user-origin `user/message` reads the nearest earlier inbox Context: an ID in the current claim produces a Steering Node, while every other user-origin message produces an ordinary User Node. A Reader miss while older history remains records a window-gap dependency. When prepend supplies the missing predecessor, the Assembler replays the affected inbox chain and message Contexts in forward Event order. Historical page direction therefore cannot permanently misclassify a message. @@ -52,11 +54,11 @@ Let `E` be the loaded raw Event count, `P` one newly prepended page, `D` the num | Path | Context work | Target snapshot work | Result | |---|---|---|---| -| Initial tail or reconnect replace | Match the loaded window in `O(E × D)` and build State in forward Event order | Build and order `C` contributions | A full replace remains proportional to the loaded window | -| Older-page prepend | Match only fresh Events and replay only Contexts whose Match, Location, or Reader answer changed, in `O(P × D + Mᵣ)` | Rebuild the stage snapshot from `C` contributions | Business folding does not restart over all `E` Events | -| Live append | Match in `O(D)`, locate the keyed Context in `O(1)`, and update only that State | Replace a same-anchor contribution in `O(1)` before snapshot assembly | Business correlation is independent of loaded Event history | +| Initial tail or reconnect replace | Match the loaded window in `O(E × D)` and build State in forward Event order | Inactive: none; active: build and order `C` contributions | A full Context replace remains proportional to the loaded window | +| Older-page prepend | Match only fresh Events and replay only Contexts whose Match, Location, or Reader answer changed, in `O(P × D + Mᵣ)` | Inactive: none; active: rebuild the stage snapshot from `C` contributions | Business folding does not restart over all `E` Events | +| Live append | Match in `O(D)`, locate the keyed Context in `O(1)`, and update only that State | Inactive: none; active: replace a same-anchor contribution in `O(1)` before snapshot assembly | Business correlation is independent of loaded Event history | -The Builder stores contributions by Context key and keeps a key-to-position index. A content update with the same anchor replaces one contribution in place; a new contribution or anchor change rebuilds and sorts contribution order. Snapshot assembly then walks `C` contributions, indexes Request headers and Tool schemas with Maps, and handles Compaction boundaries and Turn errors with linear cursors or indexes. +First activation builds the current `C` target Contexts without rematching Events and calls `replace()` once. Once active, the Builder stores contributions by Context key and keeps a key-to-position index. A content update with the same anchor replaces one contribution in place; a new contribution or anchor change rebuilds and sorts contribution order. Snapshot assembly then walks `C` contributions, indexes Request headers and Tool schemas with Maps, and handles Compaction boundaries and Turn errors with linear cursors or indexes. Final Event and Request ordering keeps a publication's current upper bound at `O(C log C)`. The migration removes repeated reverse lookups and the old raw-history refold, but it does not claim end-to-end `O(1)` publication. Chat retains its existing keyed snapshot behavior and complexity; adding the Trajectory target does not make Chat scan Trajectory Contexts or Nodes. @@ -72,7 +74,7 @@ The Context migration and the following presentation optimizations solve differe | Following Assistant lookup | One reverse pass records the next Assistant for every input position | The former repeated forward lookup falls from worst-case `O(C²)` to `O(C)` | | Group duration | Fixed decimal grouping replaces `toLocaleString('en-US')` for the invariant English numeric shape | Complexity remains linear in Groups, but the Intl formatter leaves the repeated render path | -Display memoization and search indexing stay separate. Search must include off-screen records and may lag live changes by the throttle interval; Table rendering must update the visible changed record immediately and must not inherit the index's commit cadence. +Display memoization and search indexing stay separate. Search includes off-screen records in the current React-visible history window and may lag live changes by the throttle interval; Table rendering must update the visible changed record immediately and must not inherit the index's commit cadence. ## Alternatives considered @@ -86,11 +88,11 @@ Display memoization and search indexing stay separate. Search must include off-s **Replace the Trajectory stages with generic Conversation Nodes.** Rejected: stages organize requests, timing, schemas, and table layout for one view. Making them engine contracts would constrain a future plain Session-log view and return view-specific composition to Client Runtime. -**Share one Markdown cache between display and search.** Rejected: display is immediate and viewport-bound, while search covers the complete loaded record set and intentionally batches updates. A shared cache would couple correctness and scheduling across unrelated consumers. +**Share one Markdown cache between display and search.** Rejected: display is immediate and viewport-bound, while search covers the complete React-visible record set and intentionally batches updates. A shared cache would couple correctness and scheduling across unrelated consumers. ## Verification -Runtime tests pin target registration, exact-ID append, update-before-start replay, prepend identity, Reader window-gap repair, Location replay, and isolation between Chat and Trajectory snapshots. +Runtime tests pin target registration, first-subscription activation, exact-ID append, update-before-start replay, prepend identity, Reader window-gap repair, Location replay, and isolation between Chat and Trajectory snapshots. Trajectory Definition and Builder tests pin Assistant streaming and interruption, nested Tool calls and parallel interruption, Compaction and prompt inheritance, Steering classification and Step placement, Request marker order, stable contribution replacement, and prepend expansion. Table, layout, Timeline, and search tests pin deferred Markdown work, throttled index updates, tooltip-time formatting, and stable search results across append and prepend. @@ -98,7 +100,7 @@ Trajectory Definition and Builder tests pin Assistant streaming and interruption Trajectory business assembly now scales with the changed page or keyed Context instead of restarting from the complete raw Event window. Target-owned Definitions can evolve independently from Chat while retaining one Session window and one set of lifecycle rules. Steering becomes a first-class Trajectory record at its actual Step position without adding steering-specific state to Session. -The retained stage-oriented Builder still performs work proportional to materialized Trajectory contributions and may sort on publication. The search index still performs a light linear signature pass when its input layout changes. These costs are explicit target-view work, not hidden full Event refolding. +After first activation, the retained stage-oriented Builder still performs work proportional to materialized Trajectory contributions and may sort on publication. Before activation, the target retains Context State and one target index but no Builder, materialized Node, or snapshot. Each Trajectory view mount anchors React layout, timeline, and search data to 50 target Nodes at the current tail; live appends extend that window, and the existing earlier-history action extends its prefix before it requests another Session page. If a replacement window no longer contains the previous tail anchor, the same render uses the replacement's latest Node as its bound and adopts that anchor for subsequent appends. Request numbering and cumulative usage remain derived from the complete resident snapshot. The search index still performs a light linear signature pass when its input layout changes. Definition authors must provide stable protocol identities. Old Events without a required ID can disappear from the affected Trajectory business view, which is preferable to joining unrelated records or failing history load; producers that require faithful display must log the identity. diff --git a/.agents/notes/implemented/architecture/2026-08-11-trajectory-conversation-context-assembly.zh.md b/.agents/notes/implemented/architecture/2026-08-11-trajectory-conversation-context-assembly.zh.md index 1909243da0..2568de889f 100644 --- a/.agents/notes/implemented/architecture/2026-08-11-trajectory-conversation-context-assembly.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-11-trajectory-conversation-context-assembly.zh.md @@ -18,6 +18,8 @@ Trajectory 针对共享的 [`ConversationNodeAssembler`](2026-08-09-client-conve 每个 Definition 只属于一个 target。Chat 与 Trajectory 可以识别同一持久 Event 族，但分别维护自己的 State 和最终 Node payload。它们只共享 Assembler 的精确 ID 匹配、有序 Match、Location 事实、Reader 依赖、发布调度，以及 replace/prepend/append 生命周期。 +Trajectory 的 target source 在首次订阅时激活 View Builder。在此次订阅之前，其 Definition 会维护最新 State，而 Assembler 跳过 `buildViewNode()` 和 snapshot assembly。取消订阅后 target 仍保持 active，因此后续访问会复用持续增量维护的 snapshot。 + 既有的 [Trajectory 检查记录表](../feature/2026-07-27-trajectory-inspection-ledger.zh.md)继续作为视图模型。Trajectory Builder 把已物化的 target Node 转换为原有的 `eventNodes`、Requests、Tool schema、运行中调用和 Location map；layout、表格虚拟化、选择、Overview 与检查器行为不会成为通用 Conversation 约定。 ### 业务 Definition @@ -40,7 +42,7 @@ Assistant chunk 只更新对应的 `turn:step` Context。带内容的 chunk 请 Trajectory 从持久 inbox 历史恢复 steering，使用与 [Chat steering 决策](../feature/2026-08-04-web-context-source-and-steer-marks.zh.md)相同的标识规则，但不共享 Chat 的最终 Node。 -每条目标为 `next-step` 的 `agent/inbox/spliced` Event 都会启动一个以 Event seq 标识的不可见 Context。它的 `start()` 读取最近的前序 inbox Context，应用 splice，并存储待处理标识以及累计的已领取 message ID 集合。后续用户来源的 `user/message` 读取最近的前序 inbox Context：已领取的 ID 生成 Steering Node，其余用户来源消息生成普通 User Node。 +每条目标为 `next-step` 的 `agent/inbox/spliced` Event 都会启动一个以 Event seq 标识的不可见 Context。它的 `start()` 读取最近的前序 inbox Context，把 splice 追加到持久的 pending ID state，并只在 claim 时 materialize 该 state、替换当前 claimed batch。AgentLoop 会在领取下一批消息之前追加当前 claim 接纳的全部消息；被拒绝的 claim 不追加 `user/message`。后续用户来源的 `user/message` 读取最近的前序 inbox Context：ID 属于当前 claim 时生成 Steering Node，其余用户来源消息生成普通 User Node。 仍有更早历史时，Reader miss 会记录 window-gap 依赖。prepend 补齐缺失的前驱后，Assembler 按 Event 正序重放受影响的 inbox chain 与 message Context。因此，历史分页方向不会永久错误分类消息。 @@ -52,11 +54,11 @@ Trajectory 从持久 inbox 历史恢复 steering，使用与 [Chat steering 决 | 链路 | Context 工作量 | Target snapshot 工作量 | 结果 | |---|---|---|---| -| 初始尾页或重连 replace | 以 `O(E × D)` 匹配已加载窗口，并按 Event 正序构造 State | 构造并排序 `C` 个 contribution | 完整 replace 仍与已加载窗口成正比 | -| 更早页面 prepend | 只匹配新 Event，并只重放 Match、Location 或 Reader 答案发生变化的 Context，成本为 `O(P × D + Mᵣ)` | 从 `C` 个 contribution 重建 stage snapshot | 业务 fold 不会从头重跑全部 `E` 个 Event | -| 实时 append | 以 `O(D)` 匹配，以 `O(1)` 找到 keyed Context，并只更新对应 State | snapshot 组装前，以 `O(1)` 替换 anchor 未变的 contribution | 业务关联成本与已加载 Event 历史无关 | +| 初始尾页或重连 replace | 以 `O(E × D)` 匹配已加载窗口，并按 Event 正序构造 State | inactive：无；active：构造并排序 `C` 个 contribution | 完整 Context replace 仍与已加载窗口成正比 | +| 更早页面 prepend | 只匹配新 Event，并只重放 Match、Location 或 Reader 答案发生变化的 Context，成本为 `O(P × D + Mᵣ)` | inactive：无；active：从 `C` 个 contribution 重建 stage snapshot | 业务 fold 不会从头重跑全部 `E` 个 Event | +| 实时 append | 以 `O(D)` 匹配，以 `O(1)` 找到 keyed Context，并只更新对应 State | inactive：无；active：在 snapshot 组装前以 `O(1)` 替换 anchor 未变的 contribution | 业务关联成本与已加载 Event 历史无关 | -Builder 按 Context key 保存 contribution，并维护 key-to-position index。anchor 相同的内容更新会原位替换一个 contribution；新增 contribution 或 anchor 变化才会重建并排序 contribution 顺序。随后，snapshot assembly 遍历 `C` 个 contribution，用 Map 索引 Request header 与 Tool schema，并以线性游标或索引处理 Compaction boundary 与 Turn error。 +首次激活会从当前 `C` 个 target Context 构建 Node，而不会重新匹配 Event，并调用一次 `replace()`。激活后，Builder 按 Context key 保存 contribution，并维护 key-to-position index。anchor 相同的内容更新会原位替换一个 contribution；新增 contribution 或 anchor 变化才会重建并排序 contribution 顺序。随后，snapshot assembly 遍历 `C` 个 contribution，用 Map 索引 Request header 与 Tool schema，并以线性游标或索引处理 Compaction boundary 与 Turn error。 最终 Event 和 Request 排序使单次发布的当前上界保持为 `O(C log C)`。本次迁移移除了重复反向查找和旧的原始历史 refold，但不声称端到端发布达到 `O(1)`。Chat 保持既有 keyed snapshot 行为与复杂度；增加 Trajectory target 不会让 Chat 扫描 Trajectory Context 或 Node。 @@ -72,7 +74,7 @@ Context 迁移与下列表现层优化解决的是不同成本。这些优化保 | 后继 Assistant 查找 | 一次反向遍历为每个输入位置记录后续 Assistant | 原先重复向前查找的最坏复杂度从 `O(C²)` 降为 `O(C)` | | Group duration | 以固定十进制分组替代固定英文数字形态下的 `toLocaleString('en-US')` | 复杂度仍与 Group 数线性相关，但重复 render 路径不再调用 Intl formatter | -展示 memo 与搜索索引彼此独立。搜索必须覆盖屏幕外 record，并允许实时变化延迟一个 throttle 周期；Table 必须立即更新发生变化的可见 record，不能继承索引的提交节奏。 +展示 memo 与搜索索引彼此独立。搜索覆盖当前 React 可见历史窗口中的屏幕外 record，并允许实时变化延迟一个 throttle 周期；Table 必须立即更新发生变化的可见 record，不能继承索引的提交节奏。 ## 考虑过的替代方案 @@ -86,11 +88,11 @@ Context 迁移与下列表现层优化解决的是不同成本。这些优化保 **用通用 Conversation Node 替换 Trajectory stage。** 不予采纳：stage 为单一视图组织 Request、计时、schema 和表格 layout。把它变成引擎约定会限制未来的朴素 Session-log 视图，并把视图专属组合重新放回 Client Runtime。 -**在展示与搜索之间共享一套 Markdown cache。** 不予采纳：展示要求立即更新且受 viewport 约束，搜索则覆盖全部已加载 record，并有意批量提交更新。共享 cache 会把两个无关消费方的正确性与调度节奏耦合起来。 +**在展示与搜索之间共享一套 Markdown cache。** 不予采纳：展示要求立即更新且受 viewport 约束，搜索则覆盖全部 React 可见 record，并有意批量提交更新。共享 cache 会把两个无关消费方的正确性与调度节奏耦合起来。 ## 验证 -Runtime 测试固定 target 注册、精确 ID append、先 update 后 start 的 replay、prepend identity、Reader window-gap 修复、Location replay，以及 Chat 与 Trajectory snapshot 隔离。 +Runtime 测试固定 target 注册、首次订阅 activation、精确 ID append、先 update 后 start 的 replay、prepend identity、Reader window-gap 修复、Location replay，以及 Chat 与 Trajectory snapshot 隔离。 Trajectory Definition 与 Builder 测试固定 Assistant streaming 与 interruption、嵌套 Tool call 和并行 interruption、Compaction 与 prompt 继承、Steering 分类和 Step 位置、Request 标记顺序、稳定 contribution 替换与 prepend 扩展。Table、layout、Timeline 与搜索测试固定延迟 Markdown 工作、节流索引更新、Tooltip 展示时格式化，以及 append/prepend 期间稳定的搜索结果。 @@ -98,7 +100,7 @@ Trajectory Definition 与 Builder 测试固定 Assistant streaming 与 interrupt Trajectory 业务组装的成本随变化页面或 keyed Context 增长，不再从完整原始 Event 窗口重新开始。target 自有 Definition 可以独立于 Chat 演进，同时继续共享一份 Session 窗口和一套生命周期规则。steering 会在实际所属 Step 位置成为一等 Trajectory record，不需要向 Session 增加 steering 专属状态。 -保留的 stage-oriented Builder 仍会执行与已物化 Trajectory contribution 数量成正比的工作，并可能在发布时排序。输入 layout 变化时，搜索索引仍会执行一次轻量线性签名检查。这些成本是显式的 target view 工作，不是隐藏的完整 Event refold。 +首次激活后，保留的 stage-oriented Builder 仍会执行与已物化 Trajectory contribution 数量成正比的工作，并可能在发布时排序。激活前，target 保留 Context State 和一个 target 索引，但不保留 Builder、已物化 Node 或 snapshot。每次挂载 Trajectory 视图时，React layout、timeline 与搜索数据都锚定在当前尾部的 50 个 target Node；实时 append 会扩展该窗口，现有更早历史操作则先扩展其前缀，再请求下一个 Session 页面。如果 replacement window 不再包含先前的尾锚，同一次 render 会以替换窗口的最新 Node 为边界，并把该节点采纳为后续 append 的新锚。请求编号与累计用量仍从完整的驻留 snapshot 派生。输入 layout 变化时，搜索索引仍会执行一次轻量线性签名检查。 Definition 作者必须提供稳定的协议标识。缺少必要 ID 的旧 Event 可能不会出现在受影响的 Trajectory 业务视图中；与合并无关记录或让历史加载失败相比，这是更安全的退化方式。要求完整展示的生产方必须记录该标识。 diff --git a/.agents/notes/implemented/architecture/2026-08-18-session-history-and-event-transport.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-18-session-history-and-event-transport.i18n.yaml index f717ba76e2..620239bf08 100644 --- a/.agents/notes/implemented/architecture/2026-08-18-session-history-and-event-transport.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-18-session-history-and-event-transport.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-18-session-history-and-event-transport.md -2026-08-18-session-history-and-event-transport.md: d35ed79dedd5592d15a27b0e1b952e66d80b268f -2026-08-18-session-history-and-event-transport.zh.md: 6e6ccf53e28c9a7ce76bb4aa5d80d94f39e11f10 +2026-08-18-session-history-and-event-transport.md: 10fdd9b256c27aadada97195c8dc5516b4485a43 +2026-08-18-session-history-and-event-transport.zh.md: bf110b5f1a2bea99f9aa086c66a616eddaa0a50e diff --git a/.agents/notes/implemented/architecture/2026-08-18-session-history-and-event-transport.md b/.agents/notes/implemented/architecture/2026-08-18-session-history-and-event-transport.md index d35ed79ded..10fdd9b256 100644 --- a/.agents/notes/implemented/architecture/2026-08-18-session-history-and-event-transport.md +++ b/.agents/notes/implemented/architecture/2026-08-18-session-history-and-event-transport.md @@ -163,7 +163,7 @@ Each method explicitly selects a cold inspection, live-only lookup, or resume-ca Reading titles, lists, and projections does not require an Agent. An observation operation cannot inherit resume authority merely because another Remote endpoint uses Agent lookup. -`SessionQuery.observeSession()` chooses an attached Session or borrows one prepared source from `SessionPersistence.borrowSession()`. The persistence preparation cache shares concurrent cold reads and pins the exact unpublished Session until every observation lease is released. An observation computes either all registered projections or none; callers may expose a subset, but no caller creates a partial projection state. +`SessionQuery.observeSession()` chooses an attached Session or serves a cold one from the reader's own prepared cache, filled through a persistence read handle. The cache shares concurrent cold reads and pins an entry until every observation lease is released. An observation computes either all registered projections or none; callers may expose a subset, but no caller creates a partial projection state. `session.list` never performs an unbounded cold-log scan. It uses cached projection hints when available and may fully observe only an individually stored artifact within the configured small-log byte limit to distinguish an abandoned blank Session. Missing or unreadable hints keep the row visible with unknown metadata. diff --git a/.agents/notes/implemented/architecture/2026-08-18-session-history-and-event-transport.zh.md b/.agents/notes/implemented/architecture/2026-08-18-session-history-and-event-transport.zh.md index 6e6ccf53e2..bf110b5f1a 100644 --- a/.agents/notes/implemented/architecture/2026-08-18-session-history-and-event-transport.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-18-session-history-and-event-transport.zh.md @@ -163,7 +163,7 @@ Session Remote 方法传递 `SessionId` 或 `SessionAddress`，不靠参数类 读取 title、列表和投影不要求 Agent。观察操作不能因为另一个 Remote endpoint 使用了 Agent lookup 而继承其恢复权限。 -`SessionQuery.observeSession()` 选择 attached Session，或从 `SessionPersistence.borrowSession()` 借用 prepared source。Persistence preparation cache 共享并发冷读取，并在所有 observation lease 释放前固定同一个未发布 Session。一次 observation 要么计算所有已注册 projection，要么完全不计算；调用方可以只公开其中一部分，但不会建立只计算部分 projection 的中间状态。 +`SessionQuery.observeSession()` 选择 attached Session，或从读取方自己的 prepared cache——经由持久化读句柄填充——提供冷 Session。该 cache 共享并发冷读取，并在所有 observation lease 释放前固定同一条目。一次 observation 要么计算所有已注册 projection，要么完全不计算；调用方可以只公开其中一部分，但不会建立只计算部分 projection 的中间状态。 `session.list` 不会无界扫描冷日志。它优先使用缓存的 projection hint，仅在独立存储 artifact 不超过配置的小日志字节上限时，才可能完整观察日志以判断不确定的 blank 状态。hint 缺失或不可读时，列表仍保留该行，并把 metadata 视为未知。 diff --git a/.agents/notes/implemented/architecture/2026-08-19-projection-cache-per-session-files.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-19-projection-cache-per-session-files.i18n.yaml index 7409f997a1..94ab898060 100644 --- a/.agents/notes/implemented/architecture/2026-08-19-projection-cache-per-session-files.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-19-projection-cache-per-session-files.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-19-projection-cache-per-session-files.md -2026-08-19-projection-cache-per-session-files.md: 9e102e786a6c06d82d1a0f45cc2f96a50c8abcd8 -2026-08-19-projection-cache-per-session-files.zh.md: d875c3f57800936f66fbf65233637df9bf300e2d +2026-08-19-projection-cache-per-session-files.md: c0ead55557e45905cfd368c70d826bf0c3eaade3 +2026-08-19-projection-cache-per-session-files.zh.md: e5582d14d68c1aed76ef8aab08a8d575e8a1842d diff --git a/.agents/notes/implemented/architecture/2026-08-19-projection-cache-per-session-files.md b/.agents/notes/implemented/architecture/2026-08-19-projection-cache-per-session-files.md index 9e102e786a..c0ead55557 100644 --- a/.agents/notes/implemented/architecture/2026-08-19-projection-cache-per-session-files.md +++ b/.agents/notes/implemented/architecture/2026-08-19-projection-cache-per-session-files.md @@ -20,8 +20,9 @@ Reads and writes share ONE coherent state: every read (`cachedSnapshot`) is a sy - Listing is a synchronous in-memory read; a session without a record document simply lacks the projection column. - ACP, headless, SDK, and Web sessions publish cache rows for later consumers. The log-leading durability barrier may flush a covered prefix at the cache cadence and split otherwise coalesced physical JSONL runs; recorded profile snapshots re-pack the logical event stream so cache timing does not define fixture layout. - The per-record contract scopes failure: a malformed or stale-version document reads as an absent record at open, so one bad file never bricks the cache, and a checkpoint schema bump discards stale sessions per record instead of rejecting the whole domain. -- The json backend bootstraps the per-record tree from the legacy whole-unit cache only when enumeration finds no new-layout document path. Any new document path, including an unreadable or stale file, suppresses the bootstrap for the whole unit; missing session rows refold from the log. The legacy file remains untouched. -- The cache record is bound to the same log lifecycle as before: the stored `{createdAt, cwd}` identity guards against a recreated id. +- The json backend bootstraps the per-record tree from the legacy whole-unit cache only when enumeration finds no new-layout document path, the legacy unit name matches, and its version is current or declared compatible. A version outside that accepted set remains untouched and the new domain opens empty; storage never relabels a version the domain owner did not approve. Any new document path, including an unreadable or stale file, suppresses the bootstrap for the whole unit; missing session rows refold from the log. The [cross-version read-compatibility decision](2026-09-02-projcache-cross-version-read-compat.md) owns the version policy. +- The `session_projcache` domain uses version 6 and declares versions 3, 4, and 5 compatible. Vouched-for records retain their listing projections across upgrades, absent lineage fields normalize to an unseeded identity, and a seeded caller rejects that identity and refolds cold. A record that still fails schema validation is backed up and skipped; every subsequent write stamps version 6. +- The cache record is bound to the same log lifecycle as before: the stored `{createdAt, cwd, isSeeded, inheritedEventCount}` identity guards against a recreated id or a mismatched inherited prefix. ## Alternatives considered @@ -29,3 +30,4 @@ Reads and writes share ONE coherent state: every read (`cachedSnapshot`) is a sy - **Cache-owned per-session files** (`//projection_cache.json`, the first revision of this change). Tried and reverted in review: the cache hand-rolled the medium — paths, per-path write chains, in-flight tracking, owner-only file modes, and a sqlite no-path special case — and its listing read hit the disk directly on every call while writes were throttled, so reads and writes were never consistent. - **Resolve the path through `sessionPersistence.locate(meta)`** (the file beside the session log). Rejected: the cache would have to guess "beside the log" from a log artifact path (`dirname` + fixed filename), coupling the cache to the persistence service and to a backend's layout. - **Make `per-record` a mode of the existing unit instead of a separate unit class.** Rejected: the two layouts have genuinely different state models — `single` is memory-authoritative with whole-file publish, `per-record` is stateless (the directory is the state; `loadAll` re-reads the tree) — so they are separate small classes behind one backend, with record keys validated path-safe instead of encoded. +- **Copy legacy values across undeclared unit versions.** Rejected: the json backend does not know a domain's record schema and cannot derive session-lineage fields. It copies an older record only when the domain explicitly lists that version in `compatibleVersions` and its current schema accepts the value; otherwise the record stays untouched and reads as absent. diff --git a/.agents/notes/implemented/architecture/2026-08-19-projection-cache-per-session-files.zh.md b/.agents/notes/implemented/architecture/2026-08-19-projection-cache-per-session-files.zh.md index d875c3f578..e5582d14d6 100644 --- a/.agents/notes/implemented/architecture/2026-08-19-projection-cache-per-session-files.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-19-projection-cache-per-session-files.zh.md @@ -20,8 +20,9 @@ Status: implemented - 列表读取是同步内存读；没有记录文档的会话只是缺少投影列。 - ACP、headless、SDK 与 Web 会话都会发布缓存行，供后续消费方使用。确保日志领先的持久性屏障可能按缓存节奏 flush 已覆盖的前缀，并拆分原本会合并的物理 JSONL 行；各 profile 的录制快照会重新 pack 逻辑事件流，因此缓存时序不会决定 fixture 布局。 - per-record 契约把故障范围缩小到单记录：畸形或过期版本的文档在打开时读作"无此记录"，单个坏文件不会拖垮整个缓存；检查点 schema 升级按会话丢弃过期行，而不是拒绝整个域。 -- json 后端仅在枚举时没有发现任何新布局文档路径，才从旧整单元缓存引导 per-record 目录树。只要存在任意新文档路径，即使文件不可读或版本陈旧，也会对整个单元禁用引导；缺失的会话行从日志重折叠。旧文件保持不变。 -- 缓存记录仍绑定同一日志生命周期：存储的 `{createdAt, cwd}` 身份防止被重建的 id 误导。 +- json 后端仅在枚举时没有发现任何新布局文档路径、旧单元名称匹配，且其版本为当前版本或已声明兼容版本时，才从旧整单元缓存引导 per-record 目录树。接受集合之外的版本保持不变，新域为空；存储绝不把域 owner 未批准的版本改标为当前版本。只要存在任意新文档路径，即使文件不可读或版本陈旧，也会对整个单元禁用引导；缺失的会话行从日志重折叠。[跨版本读兼容决策](2026-09-02-projcache-cross-version-read-compat.zh.md)是版本策略的权威说明。 +- `session_projcache` 域使用版本 6，并声明版本 3、4、5 兼容。经背书的记录在升级后保留列表投影；缺失的 lineage 字段归一化为 unseeded 身份，seeded 调用方会拒绝该身份并回落冷折叠。仍然通不过 schema 校验的记录会被备份并跳过；后续每次写入都使用版本 6。 +- 缓存记录仍绑定同一日志生命周期：存储的 `{createdAt, cwd, isSeeded, inheritedEventCount}` 身份防止被重建的 id 或不匹配的继承前缀误导。 ## Alternatives considered @@ -29,3 +30,4 @@ Status: implemented - **缓存自持的每会话文件**（`//projection_cache.json`，本改动的第一版）。试过并在评审中回退：缓存手搓了介质——路径、按路径的写链、在途跟踪、仅属主文件权限，以及 sqlite 无路径特判——而且它的列表读每次调用都直读磁盘、写却在节流，读写永不一致。 - **经 `sessionPersistence.locate(meta)` 解析路径**（文件放在会话日志旁）。未采用：缓存得从日志 artifact 路径"猜"日志旁边（`dirname` + 固定文件名），把缓存耦合到持久化服务与后端的布局。 - **把 `per-record` 做成既有单元的一种模式而非独立单元类。** 未采用：两种布局的状态模型本质不同——`single` 内存权威、整文件发布；`per-record` 无状态（目录即状态，`loadAll` 重扫目录树）——所以它们是同一后端下的两个小型独立类，记录键做路径安全校验而非编码。 +- **跨未声明单元版本复制旧值。** 未采用：json 后端不知道域的记录 schema，也无法推导会话 lineage 字段。只有当域在 `compatibleVersions` 中明确列出旧版本，且当前 schema 接受该值时，后端才复制旧记录；否则记录保持不变并读作不存在。 diff --git a/.agents/notes/implemented/architecture/2026-08-20-client-session-conversation-ownership.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-20-client-session-conversation-ownership.i18n.yaml index a1245a8831..6de3fe99ef 100644 --- a/.agents/notes/implemented/architecture/2026-08-20-client-session-conversation-ownership.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-20-client-session-conversation-ownership.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-20-client-session-conversation-ownership.md -2026-08-20-client-session-conversation-ownership.md: 12e137209bb43d21c3437d2ce5e9d2f9180bbc77 -2026-08-20-client-session-conversation-ownership.zh.md: f0d9861eeafc07481c536b85f5ba9667573a66ef +2026-08-20-client-session-conversation-ownership.md: 34848b3d9be4046993f11290a96f963d9cc78bba +2026-08-20-client-session-conversation-ownership.zh.md: 48557a2fc81a71858cc38c96fea630c0aa190c89 diff --git a/.agents/notes/implemented/architecture/2026-08-20-client-session-conversation-ownership.md b/.agents/notes/implemented/architecture/2026-08-20-client-session-conversation-ownership.md index 12e137209b..34848b3d9b 100644 --- a/.agents/notes/implemented/architecture/2026-08-20-client-session-conversation-ownership.md +++ b/.agents/notes/implemented/architecture/2026-08-20-client-session-conversation-ownership.md @@ -298,7 +298,7 @@ Draft and input state belong to Conversation UI and do not enter the Session sna `client/ui-chat` registers target id `chat` and owns the Chat snapshot builder, Conversation Node definitions, keyed node renderers, selection, details, statistics, locale, and Tool-inspection collaboration. -It registers the `chat` target source through `ctx.uiSession.provide()`. `ChatNodeSeat` and internal Chat consumers use `useChat` instead of passing `useConversation(snapshot => snapshot.views.get('chat'))`. +It registers the `chat` target source through `ctx.uiSession.provide()`. `ChatView` uses `useChat` for aggregate order, navigation, and timeline reads; each `ChatNodeSeat` receives identity-stable Node and Turn-process sources from that snapshot and does not subscribe to the aggregate source. Only visible non-command Chat Nodes activate Chat. Ordinary command-only history keeps the Hero visible; the `/goal` `command-input` Node activates a fresh Conversation. diff --git a/.agents/notes/implemented/architecture/2026-08-20-client-session-conversation-ownership.zh.md b/.agents/notes/implemented/architecture/2026-08-20-client-session-conversation-ownership.zh.md index f0d9861eea..48557a2fc8 100644 --- a/.agents/notes/implemented/architecture/2026-08-20-client-session-conversation-ownership.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-20-client-session-conversation-ownership.zh.md @@ -298,7 +298,7 @@ Draft 与输入状态属于 Conversation UI，不进入 Session snapshot。Queue `client/ui-chat` 注册 target id `chat`，并拥有 Chat snapshot builder、Conversation Node definitions、keyed node renderers、selection、details、stats、locale 和 tool inspection 协作。 -它通过 `ctx.uiSession.provide()` 注册 `chat` target source。`ChatNodeSeat` 和 Chat 内部消费者使用 `useChat`，不再传递 `useConversation(snapshot => snapshot.views.get('chat'))`。 +它通过 `ctx.uiSession.provide()` 注册 `chat` target source。`ChatView` 使用 `useChat` 读取聚合 order、navigation 与 timeline；每个 `ChatNodeSeat` 从该 snapshot 接收身份稳定的 Node 与 Turn-process source，不订阅聚合 source。 Chat activity 只由可见且非 command 的 Chat Node 激活。普通 command-only history 保持 Hero，`/goal` 的 `command-input` Node 激活 fresh Conversation。 diff --git a/.agents/notes/implemented/bug-fix/2026-08-04-load-pre-react-loop-sessions.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-21-session-log-read-intent.i18n.yaml similarity index 56% rename from .agents/notes/implemented/bug-fix/2026-08-04-load-pre-react-loop-sessions.i18n.yaml rename to .agents/notes/implemented/architecture/2026-08-21-session-log-read-intent.i18n.yaml index 0a513e049f..c002ef3859 100644 --- a/.agents/notes/implemented/bug-fix/2026-08-04-load-pre-react-loop-sessions.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-21-session-log-read-intent.i18n.yaml @@ -1,6 +1,6 @@ # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: -# pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-08-04-load-pre-react-loop-sessions.md -2026-08-04-load-pre-react-loop-sessions.md: e95817ee60647ca002060a4f90c2263d4fe7ce42 -2026-08-04-load-pre-react-loop-sessions.zh.md: 98fb1f530f5168fc02b312775d1bb8e6d305b8f8 +# pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-21-session-log-read-intent.md +2026-08-21-session-log-read-intent.md: 81cdd770886b24588052d9860e42e4a07650fb03 +2026-08-21-session-log-read-intent.zh.md: e5d0425981e6425323260661bee42281b1e96c98 diff --git a/.agents/notes/implemented/architecture/2026-08-21-session-log-read-intent.md b/.agents/notes/implemented/architecture/2026-08-21-session-log-read-intent.md new file mode 100644 index 0000000000..81cdd77088 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-08-21-session-log-read-intent.md @@ -0,0 +1,32 @@ +# Agent Note: Session log reads state their materialization cost + +Status: implemented + +English | [中文](2026-08-21-session-log-read-intent.zh.md) + +## Problem + +An all-purpose `Session.events` accessor hides an array-sized copy behind every read after an append. The full frozen snapshot can be cached, but streaming invalidates that cache for every new event, so a caller that only needs the log length or one event can repeatedly copy millions of references. Making the return value immutable does not require every read intent to pay that cost; event immutability is owned separately by the [source-owned session immutability decision](2026-06-11-dev-invariants-over-deep-readonly.md). + +## Decision + +`Session` exposes three cost-specific read operations. `seq` reads the current length in constant time, `eventAt(seq)` reads one event in constant time, and `snapshotEvents(fromSeq?, toSeqExclusive?)` explicitly materializes a frozen array for consumers that need array operations or a stable materialized range. Sequence parameters are non-negative log positions, not `Array.prototype.slice` offsets from the end. Recurring domain state such as the selected agent preset is read from a Session projection instead of rescanning live history. + +The complete current snapshot is cached until append because repeated whole-log consumers can share the same immutable array. A range snapshot copies only the selected references and is not cached: arbitrary range caching would retain unbounded arrays and require an eviction policy. Previously returned snapshots remain stable because accepted events are immutable and a snapshot array never grows after append. + +Recurring domain-state reads use [session projections](2026-08-19-session-projection-state-and-client-views.md) when the required value can be maintained incrementally. Raw-log snapshots remain appropriate for persistence, export, replay, and consumers whose output is the event sequence itself. This API makes materialization visible but does not attempt to eliminate every full-log fold in the same change. + +## Alternatives considered + +**Keep a cached `events` array accessor.** This preserves ordinary array syntax but makes scalar and indexed reads appear cheap while an append can turn either into a whole-log copy. + +**Return a custom immutable cut with array-like traversal operations.** A captured length could provide a stable constant-time cut over the growing log, but the abstraction would reimplement selected array semantics and keep expanding as callers request more operations. Explicit indexed reads, explicit materialization, and projections cover the shipped intents with a smaller public API. + +**Cache every materialized range or maintain an incremental chunked snapshot.** Range caching needs retention and eviction rules, while a chunked public representation changes consumers and serialization for a cost that many callers avoid through indexed reads or projections. These representations remain options if measured full-snapshot consumers justify them. + +## Consequences + +- Length and single-event reads do not copy the log. +- Full and ranged snapshots retain an explicit linear cost proportional to the selected event count. +- Consumers choose between raw history and incrementally maintained state at the call site. +- The public API does not emulate an array; callers materialize only when they need array operations. diff --git a/.agents/notes/implemented/architecture/2026-08-21-session-log-read-intent.zh.md b/.agents/notes/implemented/architecture/2026-08-21-session-log-read-intent.zh.md new file mode 100644 index 0000000000..e5d0425981 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-08-21-session-log-read-intent.zh.md @@ -0,0 +1,32 @@ +# Agent Note: 会话日志读取显式表达物化成本 + +Status: implemented + +[English](2026-08-21-session-log-read-intent.md) | 中文 + +## 问题 + +通用的 `Session.events` 访问器会在每次 append 后的首次读取中隐藏一次与数组大小成正比的复制。完整冻结快照可以缓存，但流式输出的每个新事件都会使缓存失效，因此只需要日志长度或单个事件的调用方也可能反复复制数百万个引用。返回值不可变并不要求每种读取意图都承担该成本；事件不可变性由[源端拥有的会话不可变性决策](2026-06-11-dev-invariants-over-deep-readonly.zh.md)另行负责。 + +## 决策 + +`Session` 提供三种成本不同的读取操作。`seq` 以常数时间读取当前长度，`eventAt(seq)` 以常数时间读取一个事件，`snapshotEvents(fromSeq?, toSeqExclusive?)` 则为需要数组操作或稳定物化区间的消费方显式物化冻结数组。序列参数是非负日志位置，不是从末尾计算的 `Array.prototype.slice` 偏移量。所选 agent preset 这类重复读取的领域状态来自会话投影，而不是反复扫描活跃日志。 + +当前完整快照会缓存到下一次 append，使重复读取整个日志的消费方可以共享同一个不可变数组。区间快照只复制所选引用且不缓存：缓存任意区间会保留数量无界的数组，并要求额外的淘汰策略。先前返回的快照保持稳定，因为已接受的事件不可变，且快照数组不会在 append 后增长。 + +当所需值可以增量维护时，重复的领域状态读取使用[会话投影](2026-08-19-session-projection-state-and-client-views.zh.md)。持久化、导出、回放，以及输出本身就是事件序列的消费方仍适合使用原始日志快照。此 API 使物化成本显式可见，但不会在同一项改动中消除所有完整日志折叠。 + +## 曾考虑的替代方案 + +**保留缓存的 `events` 数组访问器。** 这保留了普通数组语法，但会让标量读取和索引读取看似廉价，而一次 append 就可能使其中任何一次读取变成完整日志复制。 + +**返回提供类数组遍历操作的自定义不可变 cut。** 捕获长度可以在持续增长的日志上提供稳定且常数时间的 cut，但随着调用方要求更多操作，该抽象会不断重新实现所选数组语义。显式索引读取、显式物化和投影以更小的公开 API 覆盖了已经交付的读取意图。 + +**缓存每个物化区间或维护增量分片快照。** 区间缓存需要保留与淘汰规则，而公开的分片表示会改变消费方和序列化方式；许多调用方已通过索引读取或投影避开这项成本。若实测的完整快照消费方证明有必要，仍可考虑这些表示。 + +## 后果 + +- 长度读取和单事件读取不会复制日志。 +- 完整快照和区间快照仍有与所选事件数量成正比的显式线性成本。 +- 消费方在调用点选择原始历史或增量维护的状态。 +- 公开 API 不模拟数组；调用方只在需要数组操作时物化。 diff --git a/.agents/notes/implemented/architecture/2026-08-25-session-observations-and-projection-owned-client-state.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-25-session-observations-and-projection-owned-client-state.i18n.yaml index d8e1977615..df7268de5c 100644 --- a/.agents/notes/implemented/architecture/2026-08-25-session-observations-and-projection-owned-client-state.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-25-session-observations-and-projection-owned-client-state.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-25-session-observations-and-projection-owned-client-state.md -2026-08-25-session-observations-and-projection-owned-client-state.md: 492640385215b059761b17a057328cc5c6d24bff -2026-08-25-session-observations-and-projection-owned-client-state.zh.md: 0b892a9cae2c999b4472dd46f19068e2b4139e60 +2026-08-25-session-observations-and-projection-owned-client-state.md: 554d003da1e1767b8955be7a067cbea716f130ef +2026-08-25-session-observations-and-projection-owned-client-state.zh.md: a63b0414f8b0250e4cea95ac39a39aa2c62cd958 diff --git a/.agents/notes/implemented/architecture/2026-08-25-session-observations-and-projection-owned-client-state.md b/.agents/notes/implemented/architecture/2026-08-25-session-observations-and-projection-owned-client-state.md index 4926403852..554d003da1 100644 --- a/.agents/notes/implemented/architecture/2026-08-25-session-observations-and-projection-owned-client-state.md +++ b/.agents/notes/implemented/architecture/2026-08-25-session-observations-and-projection-owned-client-state.md @@ -32,8 +32,8 @@ flowchart LR Cache -->|"small miss"| Observe Observe --> Source{"live or cold"} Source --> Live["attached Session cut"] - Source --> Borrow["borrowSession"] - Borrow --> Prepared["SessionPreparations.borrow"] + Source --> Borrow["persistence read handle"] + Borrow --> Prepared["reader's prepared cache"] Live --> Mode{"all or none"} Prepared --> Mode Mode --> Snapshot["SessionObservation"] @@ -45,7 +45,7 @@ flowchart LR ### Observation is the point-read unit -`SessionQueryEngine.observeSession(sessionId, options)` returns a disposable `SessionObservation` containing one source kind, header, contiguous event prefix, cursor, optional projection snapshot, and the durable revision for a prepared source. An attached Session wins. Otherwise `SessionPersistence.borrowSession()` and `SessionPreparations.borrow()` share and pin one prepared Session, including an in-flight cold load. +`SessionQueryEngine.observeSession(sessionId, options)` returns a disposable `SessionObservation` containing one source kind, header, contiguous event prefix, cursor, optional projection snapshot, and the durable revision for a prepared source. An attached Session wins. Otherwise the reader's own prepared cache — keyed by `stat().revision` and pinned by observation leases — serves the cold Session, sharing one persistence read (`open(id, 'read')` + `read`) across concurrent observations, including an in-flight cold load. Every owner disposes its observation. `retain()` creates another lease over the same cut, which lets `session.follow` publish a snapshot and then transfer that exact prepared source to background Agent promotion without rereading the log. A live Session that appears during cold resolution wins before publication; a disappeared live source is retried as cold. diff --git a/.agents/notes/implemented/architecture/2026-08-25-session-observations-and-projection-owned-client-state.zh.md b/.agents/notes/implemented/architecture/2026-08-25-session-observations-and-projection-owned-client-state.zh.md index 0b892a9cae..a63b0414f8 100644 --- a/.agents/notes/implemented/architecture/2026-08-25-session-observations-and-projection-owned-client-state.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-25-session-observations-and-projection-owned-client-state.zh.md @@ -32,8 +32,8 @@ flowchart LR Cache -->|"small miss"| Observe Observe --> Source{"live or cold"} Source --> Live["attached Session cut"] - Source --> Borrow["borrowSession"] - Borrow --> Prepared["SessionPreparations.borrow"] + Source --> Borrow["persistence read handle"] + Borrow --> Prepared["reader's prepared cache"] Live --> Mode{"all or none"} Prepared --> Mode Mode --> Snapshot["SessionObservation"] @@ -45,7 +45,7 @@ flowchart LR ### Observation 是 point read 单元 -`SessionQueryEngine.observeSession(sessionId, options)` 返回可 dispose（资源释放）的 `SessionObservation`，其中包含同一份 source kind、header、连续事件前缀、cursor、可选 projection snapshot，以及 prepared source 的持久化 revision。已挂载 Session 优先；否则 `SessionPersistence.borrowSession()` 与 `SessionPreparations.borrow()` 共享并固定一份 prepared Session，包括尚未完成的冷加载。 +`SessionQueryEngine.observeSession(sessionId, options)` 返回可 dispose（资源释放）的 `SessionObservation`，其中包含同一份 source kind、header、连续事件前缀、cursor、可选 projection snapshot，以及 prepared source 的持久化 revision。已挂载 Session 优先；否则由读取方自己的 prepared cache——以 `stat().revision` 为键、由 observation lease 固定——提供冷 Session，让并发 observation 共享同一次持久化读取（`open(id, 'read')` + `read`），包括尚未完成的冷加载。 每个 owner 都会 dispose 自己的 observation。`retain()` 为同一切面创建另一份 lease，使 `session.follow` 能够先发布 snapshot，再把完全相同的 prepared source 转交给后台 Agent promotion，而无需重读日志。冷解析期间出现的 live Session 会在发布前胜出；已经消失的 live source 会按 cold source 重试。 diff --git a/.agents/notes/implemented/architecture/2026-08-25-sparse-first-party-prompt-section-orders.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-25-sparse-first-party-prompt-section-orders.i18n.yaml index d91028172d..9ac6e44cd5 100644 --- a/.agents/notes/implemented/architecture/2026-08-25-sparse-first-party-prompt-section-orders.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-25-sparse-first-party-prompt-section-orders.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-25-sparse-first-party-prompt-section-orders.md -2026-08-25-sparse-first-party-prompt-section-orders.md: ffa2e6a4f602178007a6702938dfd71a2f85cbaa -2026-08-25-sparse-first-party-prompt-section-orders.zh.md: d26ac18b075a2f0ebccccdbd072c7c066c2fe1b0 +2026-08-25-sparse-first-party-prompt-section-orders.md: 3cfdb58af3c576b3e46449a763db1c811073517e +2026-08-25-sparse-first-party-prompt-section-orders.zh.md: 31cbb98eab9eb786ac9d854b76df4c18efb88563 diff --git a/.agents/notes/implemented/architecture/2026-08-25-sparse-first-party-prompt-section-orders.md b/.agents/notes/implemented/architecture/2026-08-25-sparse-first-party-prompt-section-orders.md index ffa2e6a4f6..3cfdb58af3 100644 --- a/.agents/notes/implemented/architecture/2026-08-25-sparse-first-party-prompt-section-orders.md +++ b/.agents/notes/implemented/architecture/2026-08-25-sparse-first-party-prompt-section-orders.md @@ -24,7 +24,7 @@ The allocation preserves the established first-party sequence except for two del | Work modes | `plan:policy` 500, `team:policy` 600 | | Invocation prelude | `tools:ptc-only` 800, `context:file-reference` 900 | | Local tools | `tool:bash` 1000, `tool:pwsh` 1010, `tool:read` 1100, `tool:write` 1200, `tool:edit` 1300, `tool:glob` 1400, `tool:grep` 1500, `tool:jobs` 1600, `tool:pty` 1700 | -| Higher-level tools | `tool:web_search` 2000, `tool:web_fetch` 2100, `tool:lsp` 2200, `tool:session-query` 2300, `tool:goal` 2400, `tool:cordis` 2500, `tool:workflow` 2600, `tool:ralph` 2700, continuable-subagent guidance 2800, `tool:report` 2900 | +| Higher-level tools | `tool:web_search` 2000, `tool:web_fetch` 2100, `tool:lsp` 2200, `tool:session-query` 2300, `tool:goal` 2400, `tool:cordis` 2500, `tool:workflow` 2600, `tool:ralph` 2700, continuable-subagent guidance 2800 | | Generated protocol | `tools:sdk` 5000 | | Final-output obligations | deliverable file references 9000, `tool:structured_output` 9900 | diff --git a/.agents/notes/implemented/architecture/2026-08-25-sparse-first-party-prompt-section-orders.zh.md b/.agents/notes/implemented/architecture/2026-08-25-sparse-first-party-prompt-section-orders.zh.md index d26ac18b07..31cbb98eab 100644 --- a/.agents/notes/implemented/architecture/2026-08-25-sparse-first-party-prompt-section-orders.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-25-sparse-first-party-prompt-section-orders.zh.md @@ -24,7 +24,7 @@ Status: implemented | 工作模式 | `plan:policy` 500、`team:policy` 600 | | 调用前置说明 | `tools:ptc-only` 800、`context:file-reference` 900 | | 本地工具 | `tool:bash` 1000、`tool:pwsh` 1010、`tool:read` 1100、`tool:write` 1200、`tool:edit` 1300、`tool:glob` 1400、`tool:grep` 1500、`tool:jobs` 1600、`tool:pty` 1700 | -| 高层工具 | `tool:web_search` 2000、`tool:web_fetch` 2100、`tool:lsp` 2200、`tool:session-query` 2300、`tool:goal` 2400、`tool:cordis` 2500、`tool:workflow` 2600、`tool:ralph` 2700、可继续运行的 subagent 指导 2800、`tool:report` 2900 | +| 高层工具 | `tool:web_search` 2000、`tool:web_fetch` 2100、`tool:lsp` 2200、`tool:session-query` 2300、`tool:goal` 2400、`tool:cordis` 2500、`tool:workflow` 2600、`tool:ralph` 2700、可继续运行的 subagent 指导 2800 | | 生成协议 | `tools:sdk` 5000 | | 最终输出义务 | 可交付文件引用 9000、`tool:structured_output` 9900 | diff --git a/.agents/notes/implemented/architecture/2026-08-26-inspector-execution-realms-and-protocol-planes.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-26-inspector-execution-realms-and-protocol-planes.i18n.yaml index b175565b1d..39748610f6 100644 --- a/.agents/notes/implemented/architecture/2026-08-26-inspector-execution-realms-and-protocol-planes.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-26-inspector-execution-realms-and-protocol-planes.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-26-inspector-execution-realms-and-protocol-planes.md -2026-08-26-inspector-execution-realms-and-protocol-planes.md: e8bff0661d2d0c86c216b0a18e2feb7a2c786709 -2026-08-26-inspector-execution-realms-and-protocol-planes.zh.md: 2db6307d1bfc536ac5e8b0f5d6f03e4cfe334989 +2026-08-26-inspector-execution-realms-and-protocol-planes.md: 1c274bda4cb10458ddc9ecec846b6355fb811c2f +2026-08-26-inspector-execution-realms-and-protocol-planes.zh.md: 49fde96a3b0acc5d9621b1ca4cfd1546ca970931 diff --git a/.agents/notes/implemented/architecture/2026-08-26-inspector-execution-realms-and-protocol-planes.md b/.agents/notes/implemented/architecture/2026-08-26-inspector-execution-realms-and-protocol-planes.md index e8bff0661d..1c274bda4c 100644 --- a/.agents/notes/implemented/architecture/2026-08-26-inspector-execution-realms-and-protocol-planes.md +++ b/.agents/notes/implemented/architecture/2026-08-26-inspector-execution-realms-and-protocol-planes.md @@ -16,7 +16,7 @@ The [cross-realm CDP inspector decision](2026-08-23-cross-realm-cdp-inspector.md Top-level source directories identify execution ownership. `client/` contains only browser Client code, `host/` only Host Node-main-thread code, `worker/` only Worker-thread code, and `shared/` code that is safe in every environment. A module that executes in the Worker on behalf of a Client belongs under `worker/`, not `client/`. -The repository-required `src/index.ts` and `src/invariant.ts` discovery entries are the only root-level source exceptions. They expose the Host package entry and its service type or register the invariant companion, contain no Inspector runtime implementation, and remain at fixed paths for repository tooling. +The repository-required `src/index.ts` discovery entry is the only root-level source exception in this package. It exposes the Host package entry and its service type, contains no Inspector runtime implementation, and remains at a fixed path for repository tooling. The package publishes no invariant companion because its relationships fail in their owning wire, generation, Worker, or CDP operations, as recorded in the README. ```text src/ diff --git a/.agents/notes/implemented/architecture/2026-08-26-inspector-execution-realms-and-protocol-planes.zh.md b/.agents/notes/implemented/architecture/2026-08-26-inspector-execution-realms-and-protocol-planes.zh.md index 2db6307d1b..49fde96a3b 100644 --- a/.agents/notes/implemented/architecture/2026-08-26-inspector-execution-realms-and-protocol-planes.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-26-inspector-execution-realms-and-protocol-planes.zh.md @@ -16,7 +16,7 @@ Inspector 包的代码运行在三个 JavaScript 环境中：浏览器 Client、 顶层源码目录标识执行归属。`client/` 只包含浏览器 Client 代码，`host/` 只包含 Host Node 主线程代码，`worker/` 只包含 Worker thread 代码，`shared/` 只包含在所有环境中都安全的代码。即使某个模块代表 Client，只要它实际在 Worker 中执行，就仍属于 `worker/`，而不是 `client/`。 -仓库要求的 `src/index.ts` 与 `src/invariant.ts` 发现入口是仅有的源码根目录例外。它们暴露 Host package entry 及其 service type，或注册 invariant companion，不包含 Inspector 运行时实现，并为仓库工具保留在固定路径。 +仓库要求的 `src/index.ts` 发现入口是本包唯一的源码根目录例外。它暴露 Host package entry 及其 service type，不包含 Inspector 运行时实现，并为仓库工具保留在固定路径。本包不发布 invariant companion，因为其关系会在所属 wire、generation、Worker 或 CDP 操作中失败，原因记录在 README 中。 ```text src/ diff --git a/.agents/notes/implemented/feature/2026-08-06-continuable-child-report-obligation.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-27-adjacent-agent-steer-messaging.i18n.yaml similarity index 55% rename from .agents/notes/implemented/feature/2026-08-06-continuable-child-report-obligation.i18n.yaml rename to .agents/notes/implemented/architecture/2026-08-27-adjacent-agent-steer-messaging.i18n.yaml index a3fe8365a0..00b52be148 100644 --- a/.agents/notes/implemented/feature/2026-08-06-continuable-child-report-obligation.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-27-adjacent-agent-steer-messaging.i18n.yaml @@ -1,6 +1,6 @@ # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: -# pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-08-06-continuable-child-report-obligation.md -2026-08-06-continuable-child-report-obligation.md: 422d3ab74389e084becb75a145a82139dda38ac7 -2026-08-06-continuable-child-report-obligation.zh.md: ec7e19885204ddc99b2640407c605288f1e9c045 +# pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-27-adjacent-agent-steer-messaging.md +2026-08-27-adjacent-agent-steer-messaging.md: 31c5a4fc8c0e807ff8ab869561d0acd32eccc060 +2026-08-27-adjacent-agent-steer-messaging.zh.md: c49da882a5beb7a7e8379d15a0628fb5c617e040 diff --git a/.agents/notes/implemented/architecture/2026-08-27-adjacent-agent-steer-messaging.md b/.agents/notes/implemented/architecture/2026-08-27-adjacent-agent-steer-messaging.md new file mode 100644 index 0000000000..31c5a4fc8c --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-08-27-adjacent-agent-steer-messaging.md @@ -0,0 +1,82 @@ +# Agent Note: Adjacent Agents share one Steer send_message operation + +Status: implemented + +English | [中文](2026-08-27-adjacent-agent-steer-messaging.zh.md) + +## Problem + +Continuable Agents originally used direction-specific model controls. A parent called `send_message({ subagent_id, message })`, which delegated to a FIFO `followup` service operation. A child instead received a child-scoped `report({ output })` tool, a `tool:report` system-prompt section, and deployment-selected quiet or waking delivery. The tools described one adjacent-Agent operation through different schemas, service paths, provenance, and scheduling. + +A continuable child owns its own Session, so its parent does not automatically receive the child's transcript, tool output, or reasoning. The return path must therefore remain explicit and repeatable: a child may send progress before it finishes, remain available after sending, or fail before it can cooperate. Turning every final assistant message into an implicit result would conflate turn completion with model-selected communication and would not cover abnormal endings. + +The child-only tool and system-prompt section also preceded every inherited fork turn. They made a continuable fork child's request head differ from its parent's before the history that fork exists to reuse, forcing the provider to prefill the entire copied transcript again. + +## Decision + +`SubagentRuntime.sendMessage(sender, targetId, content, { signal })` is the only public model-authored message operation. The continuation manager accepts only the exact live sender and a target on one adjacent edge: + +- parent to direct continuable child, authorized by the child's durable `SessionHeader.parentSession`; +- resident continuable child to its exact live direct parent, authorized by the child's Activation. + +Siblings, self-targets, ancestors beyond one edge, stale Agent objects, unknown targets, and one-shot children are not alternate routes. The operation has no caller-supplied source, delivery mode, offline parent mailbox, or provider dispatch. + +Every accepted message uses `Agent.steer()`. A running target receives it at the nearest step boundary; an idle target starts a turn. An absent direct child is cold-resumed through the existing continuation lifecycle before the same Steer delivery. The manager retains waking-send accounting so a continuation-managed target cannot settle between synchronous inbox insertion and driver admission. + +Every direction uses one durable source. The service derives `senderSessionId` from the authorized Agent and frames the model-visible content as `Agent sent a message:`, so attribution cannot diverge from authority. + +```ts +import type { SessionId } from '@deepseek-ai/dsh-session' + +interface AgentMessageSource { + readonly kind: 'agent-message' + readonly form: 'relay' + readonly senderSessionId: SessionId +} +``` + +### One model tool and one return instruction + +The globally registered model tool is direction-neutral and has one fixed schema: + +```ts +interface SendMessageInput { + readonly agent_id: string + readonly message: string +} +``` + +Parents and children inherit the same definition in the same registry order. The standard definition carries a process-stable internal identity that a scoped same-name tool does not satisfy. A child `toolFilter` may explicitly remove the inherited tool, and a scoped replacement may provide different semantics; neither case receives the standard call instruction. When the standard tool remains visible, the continuation manager appends the JSON-encoded direct parent id and the instruction to send one self-contained result before finishing, plus earlier actionable findings, to the child's initial user task. For a fork child this task follows the inherited completed-turn prefix; no child-only system-prompt section or tool schema precedes that prefix. + +The instruction is guidance, not settlement enforcement. Sending does not end the child's turn, zero or several calls remain mechanically valid, and the runtime never rejects a child for staying silent. The manager-owned `subagent-settled` notice remains unconditional and separately attributed because it records how an Activation ended and preserves terminal output when the child cannot cooperate. + +Human browser prompts are not model-authored Agent messages. The remote prompt path keeps a private Queue delivery so each human prompt remains a distinct turn. Interrupt behavior and settlement delivery remain independent. + +### Complete removal and reintroduction condition + +The standalone `@deepseek-ai/dsh-tool-subagent-report` package, `report` schema, `tool:report` prompt section, `reportDelivery` configuration, report-specific message source, catalog entries, composition rows, and supported-behavior snapshots are absent. The unified tool gives up the recipient-free child shortcut and the old ability for a structural return tool to survive an explicit child allow-list. Those capabilities return only if a concrete use case requires semantics that an adjacent `agent_id` and fixed Steer cannot express; reintroducing them requires a distinct model operation and prefix-cost evidence, not an alias over `sendMessage()`. + +## Alternatives considered + +**Keep `followup` and add child-to-parent routing.** The name promises a later turn and inherits `Agent.followup()` semantics. It would obscure the chosen nearest-step behavior and preserve a parent-centric name for a direction-neutral capability. + +**Keep a recipient-free `report` wrapper over `sendMessage()`.** This preserves a convenient child shortcut and lets a scope-local registration survive global tool filtering. It loses because the separate schema and prompt duplicate one operation, make parent and child request heads differ, and let equivalent directions drift again. + +**Make `report` global.** Roots, one-shot children, remote children, and agentless callers cannot derive a report recipient. Advertising it globally would make schema visibility disagree with authority, while `send_message` already makes the recipient explicit. + +**Turn every child final message into an implicit send.** A long-lived child may have nothing useful to send in one turn and several findings in another. Automatic delivery would merge model-authored communication with the runtime's settlement account and could not replace the unconditional notice on errors, cancellation, or token exhaustion. + +**Rely only on the tool description.** A tool description helps after the model considers that tool; the failure mode is a child that believes it is finished without considering any return call. Initial-task guidance reaches that decision without changing the inherited system or tool prefix. + +**Keep quiet delivery as deployment policy.** A quiet model-authored message can be accepted while an idle target never reads it. Fixed Steer gives both directions one delivery meaning and preserves accepted order with later settlement notices. + +## Consequences + +- Model consumers expose one `send_message({ agent_id, message })` definition to parents and children, with no model-selected Queue versus Steer parameter. +- The continuation manager remains the sole owner of adjacency authorization, residency, cold resume, waking admission, and teardown races. +- Accepted messages may extend a running target's current turn; messages waiting together share next-step FIFO ordering. +- Caller cancellation owns work only until inbox acceptance and does not retract an accepted message or dispose the target. +- The initial task carries JSON-encoded dynamic parent addressing after a fork prefix, while the request-head system prompt and tool ordering remain reusable. +- Human prompts, settlement notices, QueueDock, and the base bundle's one-shot fork policy remain separate decisions. + +This decision consolidates and removes the fully superseded report-tool and child-report-obligation records. It supersedes the `followup` naming choice in [Intent-named subagent continuation operations](../simplification/2026-07-27-intent-named-subagent-continuation-operations.md) and retains the accepted-order guarantee in [Child Agent messages precede their settlement notices](../bug-fix/2026-08-17-subagent-message-settlement-ordering.md). diff --git a/.agents/notes/implemented/architecture/2026-08-27-adjacent-agent-steer-messaging.zh.md b/.agents/notes/implemented/architecture/2026-08-27-adjacent-agent-steer-messaging.zh.md new file mode 100644 index 0000000000..c49da882a5 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-08-27-adjacent-agent-steer-messaging.zh.md @@ -0,0 +1,82 @@ +# Agent Note: 相邻 Agent 共享一个 Steer send_message 操作 + +Status: implemented + +[English](2026-08-27-adjacent-agent-steer-messaging.md) | 中文 + +## 问题 + +可继续 Agent 最初使用方向专属的模型控制。parent 调用 `send_message({ subagent_id, message })`，委托给 FIFO `followup` 服务操作。child 则获得 child 作用域的 `report({ output })` 工具、`tool:report` 系统提示词 section，以及由部署选择的静默或唤醒投递。两个工具用不同 schema、服务路径、来源与调度描述同一个相邻 Agent 操作。 + +可继续 child 拥有自己的 Session，因此 parent 不会自动收到 child 的 transcript（文本记录）、工具输出或推理。返回路径必须保持显式且可重复：child 可以在结束前发送进度、发送后仍保持可用，也可能在来得及配合前失败。把每条最终 assistant 消息变成隐式结果会混淆轮次完成与模型选择的通信，而且无法覆盖异常结束。 + +child 专属工具与系统提示词 section 还位于每个继承 fork 轮次之前。它们让可继续 fork child 的请求头在 fork 旨在复用的历史之前就与 parent 不同，迫使提供方重新预填充整份复制 transcript。 + +## 决策 + +`SubagentRuntime.sendMessage(sender, targetId, content, { signal })` 是唯一公开的模型编写消息操作。继续执行管理器只接受确切在线 sender 与一条相邻边上的目标： + +- parent 到直接可继续 child，由 child 的持久化 `SessionHeader.parentSession` 授权； +- 驻留的可继续 child 到其确切在线直接 parent，由 child 的 Activation 授权。 + +sibling、自身目标、超过一条边的 ancestor、陈旧 Agent 对象、未知目标与一次性 child 都不是替代路由。该操作没有调用方提供的 source、投递模式、离线 parent mailbox 或提供方分发。 + +每条被接受的消息都使用 `Agent.steer()`。运行中目标在最近 step 边界接收消息；空闲目标启动轮次。缺失的直接 child 会先通过现有继续执行生命周期冷恢复，再接受同一 Steer 投递。管理器保留唤醒发送记账，因此受继续执行管理的目标不会在同步 inbox 插入与 driver 准入之间结算。 + +两个方向使用同一种持久来源。服务从已授权 Agent 推导 `senderSessionId`，并把模型可见内容组装为 `Agent sent a message:`，因此来源信息不会偏离权限。 + +```ts +import type { SessionId } from '@deepseek-ai/dsh-session' + +interface AgentMessageSource { + readonly kind: 'agent-message' + readonly form: 'relay' + readonly senderSessionId: SessionId +} +``` + +### 一个模型工具与一条返回指令 + +全局注册的模型工具与方向无关，并使用一个固定 schema： + +```ts +interface SendMessageInput { + readonly agent_id: string + readonly message: string +} +``` + +parent 与 child 以相同注册表顺序继承相同定义。标准定义携带进程稳定的内部身份，同名的作用域工具不满足该身份。child `toolFilter` 可以显式移除继承的工具，作用域替代工具也可以提供不同语义；两种情况都不会收到标准调用指令。当标准工具仍可见时，继续执行管理器会把经过 JSON 编码的直接 parent id、结束前发送一份自包含结果的指令，以及更早发送可操作发现的指令追加到 child 初始用户任务。对 fork child 而言，该任务位于继承的已完成轮次前缀之后；没有 child 专属系统提示词 section 或工具 schema 位于此前缀之前。 + +该指令是指导，不是结算强制。发送不会结束 child 轮次，机制仍允许零次或多次调用，runtime 绝不会因 child 保持沉默而拒绝它。由管理器负责的 `subagent-settled` 通知仍无条件发送并采用独立来源，因为它记录 Activation 如何结束，并在 child 无法配合时保留终态输出。 + +浏览器中的人类提示不是模型编写的 Agent 消息。远程提示路径保留私有 Queue 投递，使每条人类提示保持为独立轮次。中断行为与结算投递保持独立。 + +### 完整移除与重新引入条件 + +独立的 `@deepseek-ai/dsh-tool-subagent-report` 包、`report` schema、`tool:report` 提示词 section、`reportDelivery` 配置、report 专属消息来源、目录项、组合行和受支持行为快照均已不存在。统一工具放弃了无需接收方的 child 快捷方式，也放弃了让结构性返回工具绕过显式 child allow-list 的旧能力。只有具体用例需要相邻 `agent_id` 与固定 Steer 无法表达的语义时，这些能力才会重新出现；重新引入需要独立的模型操作与前缀成本证据，而非 `sendMessage()` 之上的别名。 + +## 考虑过的替代方案 + +**保留 `followup` 并添加 child 到 parent 路由。** 该名称承诺后续轮次并继承 `Agent.followup()` 语义。它会掩盖选定的最近 step 行为，并为方向无关能力保留以 parent 为中心的名称。 + +**保留 `sendMessage()` 之上无需接收方的 `report` 包装层。** 这会保留便利的 child 快捷方式，并让作用域局部注册绕过全局工具过滤。它落选是因为独立 schema 与提示词重复一个操作、使 parent 与 child 请求头不同，并允许等价方向再次漂移。 + +**让 `report` 全局可见。** 根 Agent、一次性 child、远程 child 与无 Agent 调用方无法推导 report 接收方。全局宣传它会让 schema 可见性与权限不一致，而 `send_message` 已显式给出接收方。 + +**把每条 child 最终消息变成隐式发送。** 长期运行的 child 可能在某个轮次没有值得发送的内容，在另一个轮次却有多条发现。自动投递会混合模型编写通信与 runtime 结算说明，而且无法替代错误、取消或 token 耗尽时的无条件通知。 + +**只依赖工具描述。** 工具描述会在模型考虑该工具后提供帮助；失败模式是 child 认为自己已经完成而根本没有考虑返回调用。初始任务指导能触及该决策，又不会改变继承的系统或工具前缀。 + +**保留静默投递作为部署策略。** 静默的模型编写消息可能被接受，但空闲目标永远不会读取它。固定 Steer 为两个方向提供一种投递含义，并保持与后续结算通知的接受顺序。 + +## 后果 + +- 模型 Consumer 向 parent 与 child 公开一个 `send_message({ agent_id, message })` 定义，不提供模型选择的 Queue 与 Steer 参数。 +- 继续执行管理器仍是相邻关系授权、驻留、冷恢复、唤醒准入与拆卸竞态的唯一所有者。 +- 被接受的消息可以延长运行中目标的当前轮次；一起等待的消息共享 next-step FIFO 顺序。 +- 调用方取消只在 inbox 接受前掌管工作，不会撤回已接受消息或 dispose（资源释放）目标。 +- 初始任务在 fork 前缀之后携带经过 JSON 编码的动态 parent 地址，而请求头系统提示词与工具顺序保持可复用。 +- 人类提示、结算通知、QueueDock 与 base bundle 的一次性 fork 策略仍是独立决策。 + +本决策合并并删除了已完全被取代的 report 工具与 child report 义务记录。它取代[按意图命名的 subagent 继续执行操作](../simplification/2026-07-27-intent-named-subagent-continuation-operations.zh.md)中的 `followup` 命名选择，并保留[Child Agent 消息先于其结算通知](../bug-fix/2026-08-17-subagent-message-settlement-ordering.zh.md)中的接受顺序保证。 diff --git a/.agents/notes/implemented/bug-fix/2026-07-28-load-pre-identity-session-messages.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-27-handle-based-session-persistence.i18n.yaml similarity index 55% rename from .agents/notes/implemented/bug-fix/2026-07-28-load-pre-identity-session-messages.i18n.yaml rename to .agents/notes/implemented/architecture/2026-08-27-handle-based-session-persistence.i18n.yaml index a353fc98ef..2c0b6abd41 100644 --- a/.agents/notes/implemented/bug-fix/2026-07-28-load-pre-identity-session-messages.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-27-handle-based-session-persistence.i18n.yaml @@ -1,6 +1,6 @@ # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: -# pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-07-28-load-pre-identity-session-messages.md -2026-07-28-load-pre-identity-session-messages.md: 6d022cb4b37345cd61cc9a89c6fc55c19ad402a5 -2026-07-28-load-pre-identity-session-messages.zh.md: 86439337b3c646a72b7584fbf4640799226fc9ca +# pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-27-handle-based-session-persistence.md +2026-08-27-handle-based-session-persistence.md: ced7a78178d3036fd5fa9a09ca49f8c751c3a169 +2026-08-27-handle-based-session-persistence.zh.md: e13b5b0e9e7286411b6d38f3b9dc0e86fa742a3d diff --git a/.agents/notes/implemented/architecture/2026-08-27-handle-based-session-persistence.md b/.agents/notes/implemented/architecture/2026-08-27-handle-based-session-persistence.md new file mode 100644 index 0000000000..ced7a78178 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-08-27-handle-based-session-persistence.md @@ -0,0 +1,42 @@ +# Agent Note: Handle-based session persistence + +Status: implemented + +English | [中文](2026-08-27-handle-based-session-persistence.zh.md) + +## Problem + +The previous persistence seam owned far more than storage. A shared coordinator subscribed to `session/created`/`session/event`/`session/flush`/`session/disposed` and adopted any published session (ownerless claims, HMR re-seeding, stored-prefix adoption); a bounded prepared-Session LRU with exclusive reservations served resume and read-only observation from one cache; committing crash repair lived inside `load`/`prepare`; and optimistic revision read/check/read loops stood in for ownership, so a continuous external writer could livelock a read and nothing excluded a second writer. The service surface (twelve methods) mixed storage with Session construction and lifecycle. Cross-process write ownership — the next step — has no honest home in that shape: ownership belongs to an explicit per-session channel with an owner, not to a global listener. + +## Decision + +**The seam is five service methods returning or serving per-session handles.** `create(header)` stores a new session and returns its owned write handle; `open(id, 'read' | 'write')` opens an existing one; `stat(id)`/`list()` observe snapshots (`header`, opaque `revision`, optional `eventCount`/`sizeBytes` hints — the JSONL backend supplies `sizeBytes`) without reading logs; service-level `flush()` is one backend-wide durability barrier that drains and flushes every active write handle, aggregating per-session failures without abandoning the sweep. The seam carries no raw-artifact export: the WebUI ZIP download serializes the logical log (header line + events) from a read handle in `dsh-session-log-export`, so every backend exports identically and the JSONL-only 501 path is gone. A `SessionHandle` carries `read(offset?, length?)` (validated contiguous prefix slices, never a torn tail, monotonic per handle), `append` (contiguous; persistence is best-effort on resolution, and the shipped JSONL backend happens to persist each batch immediately), `flush` (the durability barrier, which also materializes an empty session), and idempotent uncancellable `close`. One handle type serves both accesses — a mutation on a read handle is a runtime `SessionReadOnlyError`, the deliberate convention of this codebase's other seams rather than a typed split. Single-writer ownership is enforced in-process by a registry (`SessionAlreadyOwnedError`); the durable cross-process lease is the planned next layer on the same shape. + +**The agent lifecycle owns handle acquisition; the backend owns the event-driven flow.** agent-loop — the sole production publication point for sessions — acquires the handle before publication (`create` for fresh sessions, appending any constructor seed through it; `open(id, 'write')` for resume) and closes it in the same memoized teardown that drains the loop. Because persistence already enforces one active write handle per session id, the backend installs the session listeners once and routes by id: `session/event` into the owning handle's bounded write-behind window (an internal scheduling policy, not configuration), `session/flush` as the durability and error-observation barrier, `session/disposed` as final drain and close. Nothing about the write path crosses the package boundary — no writer component, no batching configuration, no drain registry. Root-fiber disposal runs every fiber's disposers concurrently, so `close()` itself drains the routed buffer through the still-open storage; backend teardown's close sweep keeps application shutdown lossless regardless of which fiber unwinds first. The drain guarantees only buffered already-emitted events; a root dispose mid-turn still loses the turn's unemitted remainder by design — the next resume's `interruptedTurnClosers` repairs that tail durably. Sessions published outside the lifecycle no longer persist implicitly; nothing in production does that. + +**Semantic crash repair moved out of persistence.** Resume reads the physically valid log through its write handle, computes `interruptedTurnClosers`, and appends them (plus the constructor's `session/end-seed` marker) through the same handle as ordinary batches — repair is not a special storage entry point. Read-only observers (session-query) balance an interrupted cold log in memory only, and own their cold-Session cache keyed by `stat().revision`; the persistence-side prepared cache and revision convergence loops are deleted. + +**Visibility and freshness are explicit.** A created session is observable in-process from `create`; physical materialization may be deferred (a pure optimization) until the first append or flush, other processes see only materialized sessions, and a crash before materialization means the session never existed. Once an append or flush resolves, reads started afterwards on the same backend instance observe at least that prefix — the guarantee `message-feedback`'s durable-target check rides on. + +**Revision simplifies to a per-instance change token.** Equal tokens may be treated as an unchanged log; ownership churn never changes one. JSONL derives a best-effort token and `sizeBytes` from one `fs.stat`; a backend whose medium can count events cheaply may supply the `eventCount` hint instead. The session-list cold blank probe returns on this metadata (`coldBlankProbeMaxEvents`/`coldBlankProbeMaxBytes`), restoring the capability removed with the path query. + +## Alternatives considered + +**Typed read/write handle classes (or overloads).** Rejected as the seam style: this codebase's seams prefer one access-tagged type with runtime refusal, and the split would double every consumer-facing type for one compile-time check. + +**Keeping the coordinator's adoption/HMR write path beside handles.** Rejected: adoption exists to guess ownership after the fact; with the lifecycle handing the handle over explicitly, a reloaded backend that cannot serve old handles fails the writer loudly instead of silently re-claiming logs, and a session with no handle is a composition bug surfaced by absent persistence rather than masked by adoption. + +**A service-level `append(id, events)` beside handles.** Rejected: an id-addressed write path bypasses ownership; every write flows through the owning handle so the future lease check has exactly one door. + +**Persistence-owned batching configuration.** Rejected: the batching window is internal write-path scheduling, not a deployment-varying choice, so it is a provider constant and no configuration knob exists anywhere. + +## Consequences + +Resume, fork, subagent, ACP, webhook, and SDK sessions all persist through one explicit acquisition point, and dispose provably releases write ownership (reopening for write succeeds after teardown). The costs: a backend plugin reload under live sessions invalidates their handles — writes fail loudly until the sessions restart, where adoption previously re-attached silently; `ctx.sessions.create` + `flush` in a test persists nothing without a handle (tests seed through `create`/`append`/`close`); resume re-reads a cold log only when no immediately preceding observation parsed the same artifact — a bounded provider-local memo (session id + stat revision, invalidated by every local mutation) serves the observe-then-promote and authorize-then-resume handoffs without restoring the deleted borrow/reservation lifecycle, and the session-query reader's own prepared cache remains the pin-capable layer above it (a later consolidation may fold one into the other); and an empty created session is invisible to other processes until an explicit flush (ACP forces one for its resumable-empty-session promise). `SESSION_FORMAT_VERSION` stays 0. + +## Related + +- [Session persistence as an abstract service](2026-06-14-session-persistence.md) — the seam this reshapes; its interface list reflects the handle API. +- [Persistence export() and pre-release trims](../simplification/2026-08-27-persistence-export-and-pre-release-trims.md) — the preparatory removals, including the blank probe this note's metadata restores. +- [Retain ignorable external session events](2026-08-30-retain-ignorable-external-session-events.md) — the read-side refusal contract, now shared through `storage-contract` helpers. +- [Bounded session-persistence write batching](2026-08-08-bounded-session-persistence-write-batching.md) — the batching semantics the routed write path preserves as internal scheduling policy. diff --git a/.agents/notes/implemented/architecture/2026-08-27-handle-based-session-persistence.zh.md b/.agents/notes/implemented/architecture/2026-08-27-handle-based-session-persistence.zh.md new file mode 100644 index 0000000000..e13b5b0e9e --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-08-27-handle-based-session-persistence.zh.md @@ -0,0 +1,42 @@ +# Agent Note: 基于句柄的会话持久化 + +Status: implemented + +[English](2026-08-27-handle-based-session-persistence.md) | 中文 + +## 问题 + +先前的持久化 seam 承担的远不止存储。一个共享协调器订阅 `session/created`/`session/event`/`session/flush`/`session/disposed`，并接管任何已发布的会话（无主认领、HMR 重新播种、已存储前缀接管）；一个带独占预留的有界已准备 Session LRU 用同一个缓存服务恢复与只读观察；提交式崩溃修复内嵌在 `load`/`prepare` 中；乐观的 revision 读取/复核/再读取循环充当所有权的替身，因此持续的外部写入方可能使读取活锁，而且没有任何机制排除第二个写入方。服务表面（十二个方法）把存储与 Session 构造和生命周期混在一起。跨进程写所有权——下一步——在那种形态里没有诚实的归宿：所有权属于一条有明确持有者的逐会话通道，而不属于一个全局监听器。 + +## 决策 + +**该 seam 是五个返回或供给逐会话句柄的服务方法。**`create(header)` 存储一个新会话并返回其持有的写句柄；`open(id, 'read' | 'write')` 打开一个已有会话；`stat(id)`/`list()` 观察快照（`header`、不透明 `revision`、可选的 `eventCount`/`sizeBytes` 提示——JSONL 后端提供 `sizeBytes`），而不读取日志；服务级 `flush()` 是一道后端范围的持久性屏障，排空并 flush 每一个活跃写句柄，逐会话聚合失败而不中途放弃清扫。该 seam 不承载原始工件导出：WebUI 的 ZIP 下载在 dsh-session-log-export 中从读句柄序列化逻辑日志（header 行 + 事件），因此每个后端的导出完全一致，仅 JSONL 可用的 501 路径也随之消失。`SessionHandle` 承载 `read(offset?, length?)`（经过验证的连续前缀切片，绝不返回撕裂尾部，逐句柄单调）、`append`（连续；完成时的持久化是尽力而为的，交付的 JSONL 后端恰好会立即持久化每个批次）、`flush`（持久性屏障，同时把空会话实体化）以及幂等且不可取消的 `close`。一种句柄类型同时服务两种访问——在读句柄上执行修改是运行时的 `SessionReadOnlyError`，这是本代码库其他 seam 的既定惯例，而非类型层面的拆分。单写者所有权由注册表在进程内强制（`SessionAlreadyOwnedError`）；持久的跨进程租约是计划在同一形态上叠加的下一层。 + +**agent 生命周期负责获取句柄；后端负责事件驱动的流程。**agent-loop——会话在生产环境中唯一的发布点——在发布之前获取句柄（新建会话用 `create`，并通过它追加构造 seed；恢复用 `open(id, 'write')`），并在与排空循环相同的记忆化 teardown 中关闭它。由于持久化已保证每个会话 id 只有一个活跃写句柄，后端一次性安装会话监听器并按 id 路由：`session/event` 进入持有句柄的有界 write-behind 窗口（内部调度策略，而非配置），`session/flush` 作为持久性与错误观察屏障，`session/disposed` 作为最终排空并关闭。写路径没有任何部分跨越包边界——没有写入器组件，没有批处理配置，没有排空注册表。根 fiber 的 dispose 会并发运行每个 fiber 的 disposer，因此 `close()` 本身会经由仍然打开的存储排空已路由的缓冲；后端 teardown 的关闭清扫使应用关闭无论哪个 fiber 先解退都不丢数据。该排空只保证已发出并缓冲的事件；turn 中途的根 dispose 仍会按设计丢失该 turn 尚未发出的剩余部分——下一次恢复的 `interruptedTurnClosers` 会持久地修复这段尾部。在生命周期之外发布的会话不再隐式持久化；生产环境中没有任何地方那样做。 + +**语义崩溃修复移出了持久化。**恢复通过其写句柄读取物理上有效的日志，计算 `interruptedTurnClosers`，并把它们（连同构造器的 `session/end-seed` 标记）作为普通批次通过同一句柄追加——修复不是特殊的存储入口。只读观察方（session-query）仅在内存中配平被中断的冷日志，并拥有以 `stat().revision` 为键的冷 Session 缓存；持久化侧的已准备缓存与 revision 收敛循环被删除。 + +**可见性与新鲜度是显式的。**已创建的会话自 `create` 起即可在进程内被观察到；物理实体化（纯粹的优化）可以推迟到第一次 append 或 flush，其他进程只能看到已实体化的会话，实体化之前崩溃意味着该会话从未存在。一旦某次 append 或 flush 完成，其后在同一后端实例上开始的读取至少能观察到该前缀——这正是 `message-feedback` 持久目标检查所依赖的保证。 + +**revision 简化为逐实例变更令牌。**令牌相等可视为日志未变；所有权变动绝不会改变令牌。JSONL 通过一次 `fs.stat` 派生尽力而为的令牌与 `sizeBytes`；存储介质能够廉价统计事件数的后端可以改为提供 `eventCount` 提示。会话列表的冷空白探测回归到这些元数据之上（`coldBlankProbeMaxEvents`/`coldBlankProbeMaxBytes`），恢复了随路径查询一起移除的能力。 + +## 考虑过的替代方案 + +**类型化的读/写句柄类（或重载）。**不作为该 seam 的风格采纳：本代码库的 seam 偏好带访问标记的单一类型加运行时拒绝，而拆分会为一个编译期检查让每个面向消费方的类型翻倍。 + +**在句柄旁保留协调器的接管/HMR 写路径。**不采纳：接管的存在是为了事后猜测所有权；当生命周期显式移交句柄后，无法服务旧句柄的重载后端会向写入器响亮地失败，而不是静默地重新认领日志，而没有句柄的会话是一个由持久化缺席暴露、而非被接管掩盖的组合缺陷。 + +**在句柄旁提供服务级 `append(id, events)`。**不采纳：按 id 寻址的写路径绕过所有权；每次写入都流经持有句柄，使未来的租约检查恰好只有一扇门。 + +**由持久化持有批处理配置。**不采纳：批处理窗口是写路径内部的调度策略，而非随部署变化的选择，因此它是 provider 常量，任何地方都不存在配置旋钮。 + +## 后果 + +恢复、fork、subagent、ACP、webhook 与 SDK 会话全部经由一个显式获取点持久化，且 dispose 可证明地释放写所有权（teardown 之后重新以写模式打开可以成功）。代价：在有活跃会话时重载后端插件会使它们的句柄失效——写入会响亮地失败，直到会话重启，而以前接管会静默重连；测试中 `ctx.sessions.create` + `flush` 在没有句柄时什么也不持久化（测试通过 `create`/`append`/`close` 播种）；只有当紧邻其前没有观察读解析过同一产物时，恢复才重新读取冷日志——一个有界的 provider 内部 memo（按会话 id + stat 修订号，任何本地修改都使其失效）服务观察后提升与授权后恢复这两类交接，而不恢复已删除的 borrow/reservation 生命周期；session-query reader 自己的已准备缓存仍是其上方具备 pin 能力的一层（后续可考虑二者收敛）；空的已创建会话在显式 flush 之前对其他进程不可见（ACP 为其可恢复空会话承诺强制执行一次 flush）。`SESSION_FORMAT_VERSION` 保持为 0。 + +## 相关 + +- [作为抽象服务的会话持久化](2026-06-14-session-persistence.zh.md)——本 Note 重塑的 seam；其接口列表已反映句柄 API。 +- [持久化 export() 与预发布读取路径精简](../simplification/2026-08-27-persistence-export-and-pre-release-trims.zh.md)——预备性的移除，包括本 Note 的元数据所恢复的空白探测。 +- [保留可忽略的外部会话事件](2026-08-30-retain-ignorable-external-session-events.zh.md)——读取侧的拒绝约定，现经由 `storage-contract` 辅助函数共享。 +- [为会话持久化写入批处理设定上界](2026-08-08-bounded-session-persistence-write-batching.zh.md)——被路由写路径作为内部调度策略保留的批处理语义。 diff --git a/.agents/notes/implemented/architecture/2026-08-30-retain-ignorable-external-session-events.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-30-retain-ignorable-external-session-events.i18n.yaml index 8c92e95dac..a1a3a0b8ec 100644 --- a/.agents/notes/implemented/architecture/2026-08-30-retain-ignorable-external-session-events.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-08-30-retain-ignorable-external-session-events.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-30-retain-ignorable-external-session-events.md -2026-08-30-retain-ignorable-external-session-events.md: c796a8dab1a9d127473a341fc98bc9934429fbdf -2026-08-30-retain-ignorable-external-session-events.zh.md: 0c635b1082a31a0a35d01669ff9f933a1f218bee +2026-08-30-retain-ignorable-external-session-events.md: e79713cf16c627706939f33f6748eaa4821de870 +2026-08-30-retain-ignorable-external-session-events.zh.md: 8fddbfdd7e60716b0a78ef6b33de09935e9bd4d7 diff --git a/.agents/notes/implemented/architecture/2026-08-30-retain-ignorable-external-session-events.md b/.agents/notes/implemented/architecture/2026-08-30-retain-ignorable-external-session-events.md index c796a8dab1..e79713cf16 100644 --- a/.agents/notes/implemented/architecture/2026-08-30-retain-ignorable-external-session-events.md +++ b/.agents/notes/implemented/architecture/2026-08-30-retain-ignorable-external-session-events.md @@ -12,7 +12,7 @@ That producer inventory did not cover a third-party plugin that currently depend ## Decision -The canonical `SessionEvent` envelope retains `ignorable?: true`, and every representation preserves it: seed validation, JSONL, API transport, generated catalogs, and test fixtures. `PersistenceCoordinator` continues to refuse an unknown event unless its stored envelope explicitly carries `ignorable: true`; absent remains required-on-read. +The canonical `SessionEvent` envelope retains `ignorable?: true`, and every representation preserves it: seed validation, JSONL, API transport, generated catalogs, and test fixtures. The persistence seam's stored-event validation (`validateStoredEvents`) continues to refuse an unknown event unless its stored envelope explicitly carries `ignorable: true`; absent remains required-on-read. The field is removable only after a replacement supports the current third-party plugin across event production, persistence, reload, and transport, with an explicit cutover for sessions already containing the marker. The [session log versioning decision](2026-08-10-session-log-version-mechanism.md) continues to own the default-required safety rule and format-version policy. diff --git a/.agents/notes/implemented/architecture/2026-08-30-retain-ignorable-external-session-events.zh.md b/.agents/notes/implemented/architecture/2026-08-30-retain-ignorable-external-session-events.zh.md index 0c635b1082..8fddbfdd7e 100644 --- a/.agents/notes/implemented/architecture/2026-08-30-retain-ignorable-external-session-events.zh.md +++ b/.agents/notes/implemented/architecture/2026-08-30-retain-ignorable-external-session-events.zh.md @@ -12,7 +12,7 @@ Status: implemented ## 决定 -标准 `SessionEvent` 信封保留 `ignorable?: true`，每种表示都保留它：seed 校验、JSONL、API 传输、生成目录与测试 fixture。`PersistenceCoordinator` 继续拒绝未知事件，除非已存信封显式带有 `ignorable: true`；字段不存在时仍表示读取必需。 +标准 `SessionEvent` 信封保留 `ignorable?: true`，每种表示都保留它：seed 校验、JSONL、API 传输、生成目录与测试 fixture。持久化 seam 的已存事件校验（`validateStoredEvents`）继续拒绝未知事件，除非已存信封显式带有 `ignorable: true`；字段不存在时仍表示读取必需。 只有替代机制在事件生产、持久化、重新加载与传输中都支持当前第三方插件，并为已包含该标记的会话提供显式切换方案后，才能删除此字段。[Session log 版本决策](2026-08-10-session-log-version-mechanism.zh.md)继续定义默认读取必需的安全规则与格式版本策略。 diff --git a/.agents/notes/implemented/architecture/2026-08-31-session-sequence-and-log-offset-brands.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-31-session-sequence-and-log-offset-brands.i18n.yaml new file mode 100644 index 0000000000..06883b0ef8 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-08-31-session-sequence-and-log-offset-brands.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-31-session-sequence-and-log-offset-brands.md +2026-08-31-session-sequence-and-log-offset-brands.md: 0057b2b6c95391abbb4ec2464ee58b54e2fe24f2 +2026-08-31-session-sequence-and-log-offset-brands.zh.md: f51ad5953cab34294a53db6e3c91884292497faf diff --git a/.agents/notes/implemented/architecture/2026-08-31-session-sequence-and-log-offset-brands.md b/.agents/notes/implemented/architecture/2026-08-31-session-sequence-and-log-offset-brands.md new file mode 100644 index 0000000000..0057b2b6c9 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-08-31-session-sequence-and-log-offset-brands.md @@ -0,0 +1,45 @@ +# Agent Note: Distinguish Session event identities from log offsets + +Status: implemented + +English | [中文](2026-08-31-session-sequence-and-log-offset-brands.zh.md) + +## Problem + +Session positions used one structural `number` type for two incompatible meanings. An event reference names an existing row, while a prefix length, next append position, or read cut names a gap and may equal the event count. The compiler therefore accepted an offset where an event identity was required and could not expose a missed sequence-field migration. + +`SessionHeader.seedLength` also mixed a v0 storage coordinate into metadata used by body-free readers. Listing needs to know whether a Session has fork lineage, but only a reader that holds the event body can interpret the exact inherited prefix length. + +## Decision + +`@deepseek-ai/dsh-brand` exports the erased numeric primitive `BrandedNumber` and the runtime-identity helper `brandNumber()`. `@deepseek-ai/dsh-session` owns two validated brands: `SessionSeq` names one existing event and `SessionLogOffset` names a log gap, prefix length, or read offset. `SessionSeqCursor = SessionSeq | -1` represents an inclusive watermark before or after the first event, and `OptionalSessionSeq = SessionSeq | null` represents an event identity whose absence is data. + +`SessionEvent.seq`, surface replacement endpoints, provenance, and owner payload fields that identify Session events use `SessionSeq`. `Session.seq`, `Session.firstLiveSeq`, `Session.inheritedEventCount`, body-read offsets, and inherited prefix cuts use `SessionLogOffset`. Arithmetic returns an ordinary number and re-enters either domain through its validating constructor. + +The logical `SessionHeader` carries `isSeeded: boolean` and no numeric seed cut. Body-bearing storage values and observations carry `inheritedEventCount` beside the header; `Session.ownEvents()` and `Session.isOwnSeq()` hide the comparison from ordinary consumers. A seeded constructor requires an explicit seed and exact cut, including an empty seed with cut zero, because constructor input may contain child-owned setup events after the inherited prefix. + +The v0 JSONL header remains byte-compatible: absent `seedLength` decodes to `isSeeded: false` with cut zero, while present zero or nonzero values decode to `isSeeded: true` with the exact cut. Header-only listing translates only the presence bit. API, SDK, DeepSeek, telemetry, query-row, and JSON representations continue to carry ordinary numbers; their owning adapters validate and brand values when they enter same-process domain code. + +## Admission and ownership + +Domain constructors reject negative, fractional, non-finite, and unsafe integer values. Parsers validate a raw number once and retain the parsed object where the brand does not require a runtime wrapper. A compile-time brand does not discover unknown numeric fields in an external event; a format migration still needs an exhaustive owner disposition and must refuse schemas it cannot safely rewrite. + +`session/end-seed` remains a lifecycle marker, not the source of the inherited cut. Every constructor restore appends or retains that marker, including unseeded replay, so projections and cold readers receive `inheritedEventCount` explicitly instead of scanning the log. + +## Alternatives considered + +**Keep every position as `number`.** Rejected because event identities, counts, and cursors cross package and persistence seams frequently enough that accidental interchange is a migration risk, not a local arithmetic convenience. + +**Use one branded Session position for identities and offsets.** Rejected because it would again permit `eventCount` or `fromSeq` where an existing event is required and would force the `-1` and `null` sentinels into unrelated operations. + +**Derive the inherited cut from `session/end-seed`.** Rejected because the marker records constructor lifecycle, not only fork lineage, and a constructor seed may contain child-owned events after the inherited prefix. + +## Consequences + +Sequence-bearing code now states whether a number identifies an event or a gap. Header-only readers receive stable lineage metadata without opening the body, while persistence, projection, query, and authorization paths retain the exact cut they need. The on-disk v0 format and public numeric wires do not change. + +The cost is explicit conversion at durable and wire parsers and a separate exact-cut field on body-bearing observations. Projection-cache identity includes the lineage bit and exact cut, so its disposable storage domain advances and older rows rebuild on demand; body-free readers skip seeded cache hints when they do not hold the cut. Turn numbers, step numbers, message-list indexes, workflow member ordinals, token counts, and unrelated numeric domains remain plain numbers because they do not identify Session events. + +## Testing + +Type assertions pin that `SessionSeq` and `SessionLogOffset` are not interchangeable. Runtime suites cover constructor validation, mixed inherited and child-owned seeds, empty seeds, `ownEvents()` and `isOwnSeq()`, v0 JSONL absent/zero/nonzero headers in plain and Zstandard encodings, header-only listing, cold prepare and reopen, query and projection cuts, and unchanged numeric wire values. diff --git a/.agents/notes/implemented/architecture/2026-08-31-session-sequence-and-log-offset-brands.zh.md b/.agents/notes/implemented/architecture/2026-08-31-session-sequence-and-log-offset-brands.zh.md new file mode 100644 index 0000000000..f51ad5953c --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-08-31-session-sequence-and-log-offset-brands.zh.md @@ -0,0 +1,45 @@ +# Agent Note: 区分 Session 事件身份与日志偏移 + +Status: implemented + +[English](2026-08-31-session-sequence-and-log-offset-brands.md) | 中文 + +## Problem + +Session 位置曾用同一个结构化 `number` 类型表达两种不兼容的含义。事件引用指向一条已存在的记录，而前缀长度、下一追加位置或读取切点指向记录间隙，并且可以等于事件总数。因此，编译器会在需要事件身份的位置接受偏移，也无法暴露迁移时漏改的序号字段。 + +`SessionHeader.seedLength` 还把 v0 存储坐标混入了无须读取正文的 metadata consumer。列表只需要知道 Session 是否有 fork lineage，只有同时持有事件正文的读取方才能解释精确的继承前缀长度。 + +## Decision + +`@deepseek-ai/dsh-brand` 导出编译后消失的数值原语 `BrandedNumber` 与运行时保持原值的 helper `brandNumber()`。`@deepseek-ai/dsh-session` 拥有两个经验证的 brand：`SessionSeq` 指明一条已存在事件，`SessionLogOffset` 指明日志间隙、前缀长度或读取偏移。`SessionSeqCursor = SessionSeq | -1` 表达首条事件之前或之后的闭区间 watermark，`OptionalSessionSeq = SessionSeq | null` 表达允许以缺失为数据的事件身份。 + +`SessionEvent.seq`、surface 替换端点、provenance 以及 owner payload 中指向 Session 事件的字段使用 `SessionSeq`。`Session.seq`、`Session.firstLiveSeq`、`Session.inheritedEventCount`、带正文读取的偏移与继承前缀切点使用 `SessionLogOffset`。算术结果恢复为普通 number，并通过对应的验证构造函数重新进入任一领域。 + +逻辑 `SessionHeader` 携带 `isSeeded: boolean`，不携带数值 seed cut。包含正文的存储值和 observation 在 header 旁携带 `inheritedEventCount`；`Session.ownEvents()` 与 `Session.isOwnSeq()` 向普通 consumer 隐藏比较。seeded constructor 必须显式提供 seed 与精确 cut，包括 cut 为零的空 seed，因为 constructor 输入可能在继承前缀之后还包含 child-owned setup event。 + +v0 JSONL header 保持字节兼容：缺少 `seedLength` 时解码为 `isSeeded: false` 和零 cut，存在零或非零值时解码为 `isSeeded: true` 和对应精确 cut。仅 header 的 listing 只转换字段是否存在。API、SDK、DeepSeek、telemetry、query row 与 JSON 表示继续携带普通 number；由它们各自的 adapter 在值进入同进程 domain code 时完成验证与 brand。 + +## Admission and ownership + +Domain constructor 拒绝负数、小数、非有限值与非安全整数。parser 验证原始 number 一次；brand 不需要运行时 wrapper 时，保留原解析对象。编译期 brand 无法发现外部事件里的未知数值字段；格式迁移仍须获得穷尽的 owner disposition，并拒绝无法安全改写的 schema。 + +`session/end-seed` 仍是 lifecycle marker，不是继承 cut 的来源。每次 constructor restore 都会追加或保留该 marker，unseeded replay 也一样，因此 projection 与 cold reader 会显式接收 `inheritedEventCount`，而不是扫描日志。 + +## Alternatives considered + +**继续让所有位置都使用 `number`。** 拒绝，因为事件身份、计数与 cursor 已频繁跨越 package 与 persistence seam，意外混用是迁移风险，而不是局部算术便利。 + +**用同一个 branded Session position 表达身份和偏移。** 拒绝，因为这样仍会在需要已存在事件的位置接受 `eventCount` 或 `fromSeq`，还会迫使 `-1` 与 `null` sentinel 进入互不相关的操作。 + +**从 `session/end-seed` 推导继承 cut。** 拒绝，因为该 marker 记录 constructor lifecycle，并不只记录 fork lineage，而且 constructor seed 可以在继承前缀之后包含 child-owned event。 + +## Consequences + +携带序号的代码会明确说明一个 number 指向事件还是间隙。仅 header 的 reader 无须打开正文即可取得稳定 lineage metadata，persistence、projection、query 与 authorization path 则保留所需的精确 cut。磁盘 v0 格式与公共数值 wire 不变。 + +代价是在 durable 与 wire parser 处显式转换，并让含正文 observation 携带独立的精确 cut 字段。Projection cache identity 包含 lineage bit 与精确 cut，因此其可丢弃的 storage domain 会推进，旧 row 按需重建；不持有 cut 的仅 header reader 会跳过 seeded cache hint。turn number、step number、message-list index、workflow member ordinal、token count 与无关数值领域保持普通 number，因为它们不指向 Session 事件。 + +## Testing + +类型断言钉住 `SessionSeq` 与 `SessionLogOffset` 不可互换。运行时 suite 覆盖 constructor 验证、混合继承与 child-owned seed、空 seed、`ownEvents()` 与 `isOwnSeq()`、plain 与 Zstandard 编码中的 v0 JSONL 缺失/零/非零 header、仅 header 的 listing、cold prepare 与 reopen、query 和 projection cut，以及不变的数值 wire 值。 diff --git a/.agents/notes/implemented/architecture/2026-08-31-viewport-activated-syntax-highlighting.i18n.yaml b/.agents/notes/implemented/architecture/2026-08-31-viewport-activated-syntax-highlighting.i18n.yaml new file mode 100644 index 0000000000..d8166a7a02 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-08-31-viewport-activated-syntax-highlighting.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-08-31-viewport-activated-syntax-highlighting.md +2026-08-31-viewport-activated-syntax-highlighting.md: 050c4e50240f9e1d1eb4f2708df6bc93c3a1faf1 +2026-08-31-viewport-activated-syntax-highlighting.zh.md: abd02161ae00d8bdd31863826565175e1d8fddf9 diff --git a/.agents/notes/implemented/architecture/2026-08-31-viewport-activated-syntax-highlighting.md b/.agents/notes/implemented/architecture/2026-08-31-viewport-activated-syntax-highlighting.md new file mode 100644 index 0000000000..050c4e5024 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-08-31-viewport-activated-syntax-highlighting.md @@ -0,0 +1,41 @@ +# Agent Note: Viewport-activated syntax highlighting + +Status: implemented + +English | [中文](2026-08-31-viewport-activated-syntax-highlighting.zh.md) + +## Problem + +A long conversation mounts code fences and read cards far outside the visible viewport. Eager highlighting tokenizes every supported block and creates every token span during that mount, so main-thread work and DOM size scale with the whole rendered history rather than the code the reader can see. The [Shiki selection](../process/2026-07-26-web-syntax-highlighting-shiki.md) and [incremental streaming highlighter](../feature/2026-08-20-web-streaming-fence-highlight.md) bound initialization and repeated prefix work, but neither avoids the first per-block tokenization for unseen history. + +## Decision + +`useViewportHighlighting` owns one shared `IntersectionObserver` for syntax-highlightable `CodeBlock` and `ReadBlock` instances. A supported block renders its existing plain-text arm until its root first intersects the viewport. An absent or unsupported language never registers with the observer. A browser without `IntersectionObserver` activates highlighting after mount so the capability still works. + +The first intersecting entry removes its target from the observer and activates that component for the rest of its lifetime. Leaving the viewport never returns it to plain text. This one-way transition avoids repeated tokenization, token-DOM construction, and visual churn while scrolling. The shared observer disconnects when no inactive registered blocks remain. + +`CodeBlock` gates both settled `highlightToHtml` calls and streaming `StreamingHighlightSession` creation. It starts an activated stream from the current accumulated source, then retains the existing incremental tokenizer and React line caches. `ReadBlock` gates `highlightLines` while retaining its line rows and gutter. The plain and highlighted arms keep the same source text, code font, padding, wrapping, and line height; Shiki's color, bold, italic, and underline token styles remain unchanged. + +The module-level Shiki singleton warm-up remains eager. Viewport activation defers code-block content tokenization and token-span construction, not the fixed boot-grammar warm-up or the plain content DOM. + +## Testing + +The focused jsdom test replaces the process-global `IntersectionObserver`, mounts several code surfaces, and proves that non-intersecting and unsupported blocks remain plain, intersecting blocks share one observer, leaving the viewport does not remove highlighting, and an activated block continues to highlight changed source. It also covers read-card activation and observer disposal. Each test restores the global and unmounts every component, so the module-level registry cannot leak registrations into another case. + +Existing component tests run without `IntersectionObserver` and therefore cover the immediate fallback together with the established Shiki output, font styles, streaming caches, and plain-language behavior. Browser geometry is not measured by this unit suite; geometry stability relies on the unchanged shared typography and box styles of the plain and highlighted arms. + +## Alternatives considered + +**Deactivate highlighting when a block leaves the viewport.** This can reclaim token DOM from blocks already viewed, but scrolling repeatedly rebuilds the same token tree, discards streaming caches, and changes visible presentation at both viewport edges. One-way activation pays the cost at most once per mounted block. + +**Drop bold and italic token styles to make every token use identical font metrics.** This weakens syntax presentation, especially for highlighted Markdown, and is unnecessary for the chosen lifecycle: both render arms already use the same code font and fixed line height. Shiki's existing token styles remain intact. + +**Pin a measured pixel height during activation.** A fixed measurement becomes stale when a streaming fence grows or responsive wrapping changes, and can clip content or introduce an inner vertical scrollbar. The plain arm stays in normal flow instead of adding measurement state. + +**Virtualize complete code blocks or retain only a token window.** This can also bound DOM after a reader has visited every block, but it changes selection, copy, scroll anchoring, and streaming-cache ownership. Viewport activation removes unseen work without changing those behaviors. + +## Consequences + +Supported code that is never viewed incurs no content tokenization and creates no token spans. The first viewport intersection pays the normal synchronous highlight cost; a lazily imported grammar may keep the block plain until its existing load notification arrives. Activated blocks retain their highlighted DOM when scrolled away, so memory use grows with blocks the reader has visited rather than shrinking with the current viewport. + +The optimization is local to presentation. Markdown parsing, Shiki grammar selection and styling, stream-tail tokenization, copy text, and settled output remain unchanged. diff --git a/.agents/notes/implemented/architecture/2026-08-31-viewport-activated-syntax-highlighting.zh.md b/.agents/notes/implemented/architecture/2026-08-31-viewport-activated-syntax-highlighting.zh.md new file mode 100644 index 0000000000..abd02161ae --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-08-31-viewport-activated-syntax-highlighting.zh.md @@ -0,0 +1,41 @@ +# Agent Note: 视口激活的语法高亮 + +Status: implemented + +[English](2026-08-31-viewport-activated-syntax-highlighting.md) | 中文 + +## 问题 + +长对话会挂载远在可见视口之外的代码围栏和读取卡片。预先高亮会在挂载时 tokenize 每个受支持的块并创建全部 token span，使主线程工作量和 DOM 大小随完整渲染历史增长，而不是随读者能看到的代码增长。[Shiki 选型](../process/2026-07-26-web-syntax-highlighting-shiki.zh.md)与[流式增量高亮](../feature/2026-08-20-web-streaming-fence-highlight.zh.md)分别约束初始化成本和重复处理前缀的成本，但都无法避免不可见历史首次发生的逐块 tokenize。 + +## 决策 + +`useViewportHighlighting` 为可进行语法高亮的 `CodeBlock` 和 `ReadBlock` 实例持有一个共享的 `IntersectionObserver`。受支持的块在根元素首次与视口相交之前渲染既有的纯文本臂。语言缺失或不受支持时不会向 observer 注册。浏览器不提供 `IntersectionObserver` 时会在挂载后激活高亮，使该能力仍然可用。 + +首个相交条目会从 observer 中移除自己的目标，并在该组件余下的生命周期里保持激活。离开视口不会恢复为纯文本。这种单向转换避免滚动时反复 tokenize、创建 token DOM 和发生视觉切换。不存在尚未激活的注册块时，共享 observer 会断开。 + +`CodeBlock` 同时控制定稿态的 `highlightToHtml` 调用和流式 `StreamingHighlightSession` 的创建。流式块激活时从当前累积源码开始，随后保留既有的增量 tokenizer 与 React 行缓存。`ReadBlock` 控制 `highlightLines`，同时保留其行和行号槽。纯文本臂与高亮臂使用相同的源码文本、代码字体、内边距、换行规则和行高；Shiki 现有的 token 颜色、粗体、斜体和下划线样式保持不变。 + +模块级 Shiki 单例仍然预先预热。视口激活延迟的是代码块内容的 tokenize 与 token span 创建，不是固定的启动语法预热或纯文本内容 DOM。 + +## Testing + +聚焦的 jsdom 测试替换进程全局 `IntersectionObserver`，挂载多个代码表面，并证明未相交和不受支持的块保持纯文本、相交块共用一个 observer、离开视口不会移除高亮，且已激活的块会继续高亮变化后的源码。测试还覆盖读取卡片激活和 observer 释放。每个测试都会恢复全局值并卸载全部组件，因此模块级注册表不会把注册项泄漏到其他用例。 + +既有组件测试在没有 `IntersectionObserver` 的环境中运行，因此同时覆盖立即回退路径，以及既有 Shiki 输出、字体样式、流式缓存和纯文本语言行为。该单元测试套件不测量浏览器几何尺寸；几何稳定性依赖纯文本臂与高亮臂不变的共享字体排印和盒模型样式。 + +## 曾考虑的替代方案 + +**块离开视口时停用高亮。** 这可以回收已经看过的块所占用的 token DOM，但滚动会反复重建同一棵 token 树、丢弃流式缓存，并在视口两端改变可见呈现。单向激活使每个已挂载块最多支付一次成本。 + +**移除粗体和斜体 token 样式，使所有 token 使用完全相同的字体度量。** 这会削弱语法呈现，尤其影响高亮后的 Markdown；所选生命周期也不需要这一取舍，因为两个渲染臂已经使用相同的代码字体和固定行高。Shiki 现有 token 样式保持不变。 + +**在激活时锁定测量得到的像素高度。** 当流式围栏增长或响应式换行变化时，固定测量会陈旧，并可能裁切内容或引入内部纵向滚动条。纯文本臂继续处于正常文档流中，不增加测量状态。 + +**虚拟化完整代码块，或仅保留一个 token 窗口。** 这也能在读者访问所有块之后限制 DOM，但会改变文本选择、复制、滚动锚定与流式缓存的归属。视口激活消除不可见工作，同时不改变这些行为。 + +## 后果 + +从未进入视口的受支持代码不会发生内容 tokenize，也不会创建 token span。首次与视口相交时支付普通的同步高亮成本；若语法采用懒加载，代码块可能继续保持纯文本，直到既有的加载通知到达。已激活的块滚出视口后会保留高亮 DOM，因此内存占用随读者访问过的块增长，而不会随当前视口缩减。 + +该优化仅作用于呈现层。Markdown 解析、Shiki 语法选择与样式、流式尾行 tokenize、复制文本和定稿输出均保持不变。 diff --git a/.agents/notes/implemented/feature/2026-07-30-continuable-subagent-report-tool.i18n.yaml b/.agents/notes/implemented/architecture/2026-09-01-streamed-tool-call-identity.i18n.yaml similarity index 55% rename from .agents/notes/implemented/feature/2026-07-30-continuable-subagent-report-tool.i18n.yaml rename to .agents/notes/implemented/architecture/2026-09-01-streamed-tool-call-identity.i18n.yaml index a0028ab5c1..aa070add66 100644 --- a/.agents/notes/implemented/feature/2026-07-30-continuable-subagent-report-tool.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-09-01-streamed-tool-call-identity.i18n.yaml @@ -1,6 +1,6 @@ # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: -# pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-30-continuable-subagent-report-tool.md -2026-07-30-continuable-subagent-report-tool.md: 07d17f18f318a86070d9b8612512fa3c2a3815e2 -2026-07-30-continuable-subagent-report-tool.zh.md: 9cbf76a9f3f50d9b1a84b573621cd6fc2a1611c4 +# pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-09-01-streamed-tool-call-identity.md +2026-09-01-streamed-tool-call-identity.md: c52f39b735199270d65ed30a388333a217003237 +2026-09-01-streamed-tool-call-identity.zh.md: 9a7ffb6343870a08e06507408e007ad94fbad178 diff --git a/.agents/notes/implemented/architecture/2026-09-01-streamed-tool-call-identity.md b/.agents/notes/implemented/architecture/2026-09-01-streamed-tool-call-identity.md new file mode 100644 index 0000000000..c52f39b735 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-09-01-streamed-tool-call-identity.md @@ -0,0 +1,33 @@ +# Agent Note: Streamed tool-call identity survives empty continuation deltas + +Status: implemented + +English | [中文](2026-09-01-streamed-tool-call-identity.zh.md) + +## Problem + +The DeepSeek SSE translator assigned `id` and `name` on every tool-call delta that carried the field, so a continuation delta repeating either as an empty string erased the identity established by the call's first delta. The assembled block reached the loop with an empty name, which the tool registry refuses as `unknown tool ""`, leaving the affected models unable to run any tool. Gateways that fill those fields with `null` erased the identity the same way, and `WireToolCallDelta` declared both as `string | undefined`, keeping the observed `null` out of the compiler's reach. + +The empty identity outlived the turn. `appendToolCall` and `appendToolResult` write the block's id verbatim and no write path validates it, while `adoptSessionEvent` refuses a `tool/result` whose `callId` is empty, so the persistence coordinator wrapped that refusal in `SessionPersistenceCorruptionError`. A session that recorded one such call was writable and no longer loadable. + +## Decision + +`acceptIdentity` accepts only a non-empty string for a tool call's `id` and `name`; `undefined`, `null`, `''`, and any non-string leave the established value in place. The assignment set only narrows, so no input reaches a worse outcome than before. `WireToolCallDelta` widens `id`, `function.name`, and `function.arguments` to admit `null`, putting the values gateways actually send into the type system and making the runtime guard load-bearing rather than speculative. + +## Alternatives considered + +**Concatenate `id` and `name` across deltas.** Rejected: they are identity, not accumulation. Concatenation produces `Globnull` against a gateway that sends `null`, and a doubled name against one that repeats a non-empty value. + +**Refuse a conflicting non-empty identity mid-stream.** Deferred: a gateway that fragments a long tool name would be refused for it, and no observed provider re-sends a different non-empty identity within one call index. + +**Refuse a response whose tool call never receives an identity.** Deferred. It requires a new failure code, a change to the default retryable set, and a `[DONE]` gate that must not override the finish reason a provider already sent — cost and risk that the reported defect does not carry. The lenient wire it guards against is hypothetical: no report describes a stream that omits identity entirely. + +**Relax the session reader's empty-`callId` refusal.** Rejected: an empty `callId` cannot be paired back to the provider on the next request, so accepting it moves the failure into the model request. That refusal is the durable-boundary gate; the producer was the defect. + +## Consequences + +A continuation delta repeating identity empty or null is inert, so a call keeps the identity its first delta established, and the reported path to `unknown tool ""` and an unreadable session is closed. A stream that never carries identity at all still assembles an empty one, exactly as before; that path and the recovery of sessions already holding an empty `callId` are outside this change. + +## Testing + +`translate.spec.ts` covers empty and null continuation deltas, a repeated identical identity, and parallel calls holding separate identities under empty continuations. The existing cases for a wire that omits identity entirely keep their recorded empty-identity output. diff --git a/.agents/notes/implemented/architecture/2026-09-01-streamed-tool-call-identity.zh.md b/.agents/notes/implemented/architecture/2026-09-01-streamed-tool-call-identity.zh.md new file mode 100644 index 0000000000..9a7ffb6343 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-09-01-streamed-tool-call-identity.zh.md @@ -0,0 +1,33 @@ +# Agent Note：流式工具调用身份不被空续传分片抹除 + +Status: implemented + +[English](2026-09-01-streamed-tool-call-identity.md) | 中文 + +## 问题 + +DeepSeek SSE 翻译器对每个携带该字段的工具调用分片都直接赋值 `id` 与 `name`，因此续传分片把其中任一字段重复发送为空串时，会抹掉该调用首个分片已建立的身份。组装出的块带着空名字进入循环，工具注册表以 `unknown tool ""` 拒绝它，受影响的模型上任何工具都跑不起来。把这些字段填成 `null` 的网关会造成同样的抹除，而 `WireToolCallDelta` 把两者都声明为 `string | undefined`，让实际观察到的 `null` 落在编译器视野之外。 + +空身份还会活过本轮。`appendToolCall` 与 `appendToolResult` 原样写入块的 id 且没有任何写入路径校验它，而 `adoptSessionEvent` 拒绝 `callId` 为空的 `tool/result`，持久化协调器于是把该拒绝包装成 `SessionPersistenceCorruptionError`。记录过一次这种调用的会话可写但不再可读。 + +## 决定 + +`acceptIdentity` 对工具调用的 `id` 与 `name` 只接受非空字符串；`undefined`、`null`、`''` 以及任何非字符串都保留已建立的值。会触发赋值的输入集合只减不增，因此没有任何输入会比改动前更差。`WireToolCallDelta` 把 `id`、`function.name` 与 `function.arguments` 放宽到允许 `null`，使网关实际发送的值进入类型系统，运行时守卫因此是承重的而非臆测的。 + +## 考虑过的替代方案 + +**跨分片拼接 `id` 与 `name`。** 否决：它们是身份而非累积。面对发送 `null` 的网关，拼接产生 `Globnull`；面对重复发送非空值的网关，产生重复的名字。 + +**流中途拒绝冲突的非空身份。** 推迟：分片发送长工具名的网关会因此被拒，且没有观察到任何提供方在同一个调用 index 内改发不同的非空身份。 + +**拒绝始终未获得身份的响应。** 推迟。它需要新增失败 code、改动默认可重试集，还需要一个不得覆盖提供方已给出终止原因的 `[DONE]` 闸门——这些代价与风险，已报告的缺陷并不需要承担。它所防的宽松线上格式是假想的：没有任何报告描述过完全不发送身份的流。 + +**放宽会话读取端对空 `callId` 的拒绝。** 否决：空 `callId` 无法在下一次请求中与提供方配对，接受它只是把失败推进模型请求。该拒绝是持久化边界的闸门；缺陷在生产方。 + +## 后果 + +重复发送空或 null 身份的续传分片不产生作用，调用因此保有其首个分片建立的身份，通往 `unknown tool ""` 与不可读会话的已报告路径就此切断。完全不携带身份的流仍会组装出空身份，与改动前一致；该路径以及已经写入空 `callId` 的会话恢复都不在本次改动范围内。 + +## 测试 + +`translate.spec.ts` 覆盖空与 null 续传分片、重复的相同身份，以及空续传下并行调用各自保有身份。原有那些描述完全不发送身份的线上格式的用例，保留其记录的空身份输出。 diff --git a/.agents/notes/implemented/architecture/2026-09-02-projcache-cross-version-read-compat.i18n.yaml b/.agents/notes/implemented/architecture/2026-09-02-projcache-cross-version-read-compat.i18n.yaml new file mode 100644 index 0000000000..f41f4b3e1b --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-09-02-projcache-cross-version-read-compat.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-09-02-projcache-cross-version-read-compat.md +2026-09-02-projcache-cross-version-read-compat.md: 1d157d90b04fe92e86d858a8e5424a5e517721f6 +2026-09-02-projcache-cross-version-read-compat.zh.md: 3948126271a84810abf4fe1c1e6e874b279a4f5e diff --git a/.agents/notes/implemented/architecture/2026-09-02-projcache-cross-version-read-compat.md b/.agents/notes/implemented/architecture/2026-09-02-projcache-cross-version-read-compat.md new file mode 100644 index 0000000000..1d157d90b0 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-09-02-projcache-cross-version-read-compat.md @@ -0,0 +1,74 @@ +# Agent Note: Projection-cache cross-version read compatibility (session_projcache v3/v4/v5 → v6) + +Status: implemented + +English | [中文](2026-09-02-projcache-cross-version-read-compat.zh.md) + +## Problem + +The `session_projcache` storage domain evolved through three on-disk generations across published releases. An upgraded DSH_HOME failed in two ways: + +- **A v3 single-file home bricked startup after the upgrade**: the per-record layout's legacy bootstrap migrated the old whole-unit file without checking its `unit.version`, stamping the old records with the current version into the new tree; the domain layer's per-record zod validation at open then hit the missing now-required fields → `invalid-record` → the whole domain refused to open → the plugin tree failed to load. And because the bootstrap writes before validation runs, **the first boot permanently wrote the bad documents into the new tree** ("poisoning") — every later boot saw a non-empty tree, never took the legacy path again, and the home stayed unusable. +- **A v4 per-record home lost its listing titles after the upgrade**: v4 documents were silently discarded by the version-stamp check (the per-record contract), and SessionList is a zero-I/O cache-only read, so a miss served the row without projections; titles only returned as each session was individually reopened. + +The cache domain's own contract is "a stale or unreadable cache costs a longer tail replay, never a wrong value, never a refused load" — the hard failure and the wholesale discard each broke the first half of that contract or the product expectation. + +## The three on-disk generations + +| domain version | shipped in | layout | on-disk form | identity fields | row fields | +|---|---|---|---|---|---| +| 3 | 0.1.1-rc.2 | single | one file `storages/session_projcache.json` (`{unit:{name,version}, global, tables}`) | `createdAt`, `cwd?` | `ver`, `seq`, `val` | +| 4 | 0.1.2-alpha.3 | per-record | one file per session `storages/session_projcache/sessions/.json` (`{version, record}`) | `createdAt`, `cwd?` | same | +| 5 | 0.1.2-alpha.4 | per-record | same as v4 | + `isSeeded` (shipped required; now optional), `inheritedEventCount` (same) | same (`seq` numbers mean the same as v4; only type brands were added) | + +The only substantive v4→v5 difference is the two new lineage identity fields; the `ver/seq/val` row shape is identical across all three generations, and `seq` numbering did not change ([the 2026-08-31 seq/offset brands note](2026-08-31-session-sequence-and-log-offset-brands.md) pins the on-disk numbers as unchanged). v3→v4 was a layout migration with identical record content. + +One derived shape also exists: a v3 home that ran the v5 build once (the poisoned state) — its new tree holds documents **stamped 5 whose content is a v3 record** (no lineage fields). + +## Decision + +Declared read compatibility — reads tolerate vouched-for older versions, writes always stamp the current one: + +1. **`DomainSpec.compatibleVersions` (new, optional)**: the domain owner declares "records stored under these older versions are also readable under the current record schemas" (typically by declaring the fields old records lack as optional). `defineDomain` validates each entry as a non-negative integer below the current version; `descriptorOf` projects the set onto the backend `KvUnitDescriptor`. +2. **json backend per-record reads** accept version stamps in "current ∪ compatibleVersions"; anything outside the set is still discarded as foreign. **The write path always stamps the current version** (the first checkpoint after reading an old record naturally advances it). The `single` layout stays exact-version. +3. **Legacy-bootstrap version gate (the actual bug fix)**: the old whole-unit file's `unit.version` must fall inside the accepted set to be migrated; otherwise the file is left alone and the unit reads empty — stamping records the owner never vouched for turns a discardable stale cache into hard schema failures at the domain layer. +4. **The projcache domain declares `version: 6, compatibleVersions: [3, 4, 5]`**, and the two lineage fields become `.optional()`. The single reader of stored identities, `identityMatches`, normalizes absence to the unseeded lineage (`?? false` / `?? 0`): exact for unforked sessions, while a forked session's expectation is seeded → natural mismatch → discard and cold rebuild, so the lineage binding loses none of its protection. +5. **The poisoned state self-heals**: documents stamped 5 without lineage fields are declared compatible and parse under the optional schema (their content is the real pre-upgrade cache data), so the home boots again and titles serve immediately. +6. **Schema-validation backstop: `invalidRecords: 'backup-and-skip'` (declared by this domain only)**. A stored record that still fails to parse beyond read compatibility no longer refuses the whole domain: the domain layer calls the backend's `KvUnit.backupRecord` (json per-record implementation = rename the document to `.json.bak.`, bytes kept, never read again), prints the concrete failure with `logger.error` (domain, table, key, destination, zod cause), and continues the open with the record absent; the next cold read rebuilds and rewrites that session's cache. **The policy is an explicit per-domain declaration and the default stays fail-loud** — other domains still refuse the whole load on invalid stored data, and a backend without `backupRecord` (single layout, row stores) also falls back to fail-loud. Naming history: quarantine → backup-and-skip (user ruling: the word must carry both "back up" and "skip", sharing its root with the `.bak` suffix; skip-backup was rejected because the CLI `--skip-X` convention reads it as "do not back up"). For this domain it supersedes the reset/destroy recovery path of the [2026-07-28 storage recovery proposal](../../proposed/architecture/2026-07-28-storage-root-and-derived-medium-recovery.md), which stays live for authoritative and whole-medium damage. + +### v5 → v6 compatibility + +Version 6 changes only the current write stamp and keeps the v5 record schema. `compatibleVersions: [3, 4, 5]` therefore admits both healthy v5 records and v5-stamped lineage-less records produced by the faulty bootstrap. The current schema accepts absent lineage; `identityMatches` interprets it as unseeded and rejects the record for a seeded session. The next successful checkpoint rewrites an accepted v5 record with a v6 stamp and complete lineage. No separate v5→v6 rewrite runs at startup: an unaccepted version reads as absent, while a schema-invalid accepted record follows `backup-and-skip`. + +### Upgrade matrix + +| home shape | behavior after the fix | +|---|---| +| v3 single-file (not poisoned) | bootstrap migrates (3 ∈ accepted set) → titles serve immediately | +| v3 + poisoned new tree | new-tree documents read directly (optional tolerance) → boot restored, titles serve immediately | +| v4 per-record | documents read directly (4 ∈ accepted set) → titles serve immediately | +| v5 healthy | documents read directly (5 ∈ accepted set) → titles serve immediately | +| v6 current | unaffected | +| old records of forked (seeded) sessions | identity mismatch → discarded, cold rebuild when the session opens (safe side) | + +## Alternatives considered + +- **Discard-and-rebuild only** (bootstrap gate without compatible versions): fixes the boot, but every SessionList title is lost after the upgrade until each session is reopened — fails the upgrade-and-go product requirement. +- **Schema `.default()` fills**: behaviorally equivalent to optional + reader normalization, but bakes the "absent = unseeded" interpretation into the durable schema's output type; ruled for optional — the schema honestly describes every accepted on-disk shape and the interpretation lives at the consumer (user ruling, 2026-09-02). +- **Roll the domain version back to 4**: a small diff, but breaks version monotonicity, depends on the "bootstrap skips no versions" bug itself, and drops every poisoned and healthy v5 home's cache. + +## Consequences + +- A deployment routing this domain to the sqlite backend gets none of the tolerance: sqlite implements neither `compatibleVersions` nor `backupRecord`, so behavior degrades to the old strict-version semantics (a whole-unit version mismatch still refuses with `version-mismatch`; nothing loosens, nothing serves wrong values). Shipped compositions route this domain to json, so this stays a deployment-configuration risk only. +- The optional lineage fields let accepted records omit lineage: a lineage-less record decodes as unseeded. The identity match still refuses it for seeded callers, and the per-row `ver` guard still screens every value, so the residual exposure is an unseeded caller reading an unseeded-shaped record — the same trust extended to genuine pre-lineage records. +- `backupRecord` overwrites a same-minute backup of the same key (the newer bytes win); distinct minutes and distinct keys never collide. + +## Testing + +- `storage-json` unit tests: compat-stamped reads / out-of-set discards / writes stamping current; legacy bootstrap migrating only accepted versions (including the migrated-documents-stamp-current assertion); `backupRecord` move / absent read / rewrite / closed guard. +- `storage-domain` unit tests: `compatibleVersions` / `invalidRecords` declaration validation; backup-and-skip falling back to fail-loud when the backend has no `backupRecord`. +- `session-projection-cache` unit tests: records without lineage fields serve unseeded sessions verbatim and are discarded for seeded ones. +- **Archived-fixture recovery tests** (`tests/fixtures.spec.ts` + `tests/fixtures/`): four media archives produced by the real released builds — `v3-single-unit.json` (the 0.1.1-rc.2 whole-unit file), `v4-session-doc.json` (0.1.2-alpha.3), `v5-session-doc.json` (0.1.2-alpha.4), `v5-lineageless-doc.json` (the unguarded bootstrap's poisoned shape, synthesized from the v3 record) — each opened through the real storage stack, asserting the listing serves the archived title and that a live write rewrites the document to the current version (v6 stamp + lineage fields + fresh value); plus the backup-and-skip behavior for a schema-failing record (boot survives, `.bak` lands, log is concrete, neighbor records unharmed). +- End-to-end acceptance, executed against the real release artifacts: the published 0.1.1-rc.2 and 0.1.2-alpha.3 npm builds seeded homes through their own web apps (model turns plus a rename RPC), the published 0.1.2-alpha.4 build reproduced both failures (including the poisoned tree), and the fixed build served every home shape — pristine v3, poisoned v3, v4, and fresh — with the SessionList RPC returning the recorded titles verbatim. + +Future bump procedure: when a new version's shape can tolerate old records through "optional fields + reader normalization", add the old version to `compatibleVersions`; otherwise bump normally (discard and rebuild) and remove the no-longer-compatible versions from the set. Either way, the package README requires the bump to land with archived fixtures and tests proving the chosen disposition. diff --git a/.agents/notes/implemented/architecture/2026-09-02-projcache-cross-version-read-compat.zh.md b/.agents/notes/implemented/architecture/2026-09-02-projcache-cross-version-read-compat.zh.md new file mode 100644 index 0000000000..3948126271 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-09-02-projcache-cross-version-read-compat.zh.md @@ -0,0 +1,74 @@ +# Agent Note: 投影缓存跨版本读兼容（session_projcache v3/v4/v5 → v6） + +Status: implemented + +[English](2026-09-02-projcache-cross-version-read-compat.md) | 中文 + +## 问题 + +`session_projcache` 存储域在已发布版本间演进了三代磁盘结构。升级后的 DSH_HOME 出现两类故障： + +- **v3 单文件 home 升级后启动硬失败**：per-record 布局的 legacy bootstrap 迁移旧单文件时不检查其 `unit.version`，把旧记录原样打上当前版本戳写入新树；domain 层开域时逐条 zod 校验，旧记录缺新增必填字段 → `invalid-record` → 整个域拒开 → 插件树加载失败。且 bootstrap 先写盘后校验，**首次启动即把坏文档永久写入新树**（"投毒"）——此后每次启动新树非空、连 legacy 路径都不再走，home 持续不可用。 +- **v4 per-record home 升级后列表丢标题**：v4 文档被版本戳检查静默丢弃（per-record 契约），SessionList 是零 I/O 纯缓存读，miss 后整行不带投影；标题要等每个会话被逐个重新打开后才恢复。 + +缓存域自身的契约是"过期或不可读的缓存只付出更长的尾部重放，绝不给出错值、绝不拒载"——硬失败与整体丢弃都违背该契约的前半句或产品预期。 + +## 三代磁盘结构差异 + +| domain version | 携带发布 | 布局 | 磁盘形态 | identity 字段 | 行字段 | +|---|---|---|---|---|---| +| 3 | 0.1.1-rc.2 | single | 单文件 `storages/session_projcache.json`（`{unit:{name,version}, global, tables}`） | `createdAt`, `cwd?` | `ver`, `seq`, `val` | +| 4 | 0.1.2-alpha.3 | per-record | 每会话一份 `storages/session_projcache/sessions/.json`（`{version, record}`） | `createdAt`, `cwd?` | 同上 | +| 5 | 0.1.2-alpha.4 | per-record | 同 v4 | + `isSeeded`（v5 首发必填；现为 optional）、`inheritedEventCount`（同前） | 同上（`seq` 数值语义与 v4 相同，仅类型加 brand） | + +v4→v5 的唯一实质差异是 identity 新增两个 lineage 字段；行内 `ver/seq/val` 三代一致，`seq` 的数值含义未变（[2026-08-31 seq/offset brands note](2026-08-31-session-sequence-and-log-offset-brands.zh.md) 明确 on-disk 数值不变）。v3→v4 是布局迁移，记录内容结构一致。 + +另有一种衍生形态：跑过一次 v5 版本的 v3 home（投毒态）——新树里存在**版本戳为 5 但内容是 v3 记录**（缺 lineage 字段）的文档。 + +## 决策 + +声明式读兼容——读容忍 owner 背书过的旧版本，写恒戳当前版本： + +1. **`DomainSpec.compatibleVersions`（新增，可选）**：域 owner 声明"这些旧版本的存量记录在当前记录 schema 下也可读"（典型手段：新增字段标 optional）。`defineDomain` 校验各项为小于当前 version 的非负整数；`descriptorOf` 透传到后端 `KvUnitDescriptor`。 +2. **json 后端 per-record 读**：接受"当前版本 ∪ compatibleVersions"内的版本戳，集合外照旧视为 foreign 丢弃；**写路径永远戳当前版本**（读到旧记录后的下一次 checkpoint 自然把它推进到当前版本）。single 布局维持 exact-version 不变。 +3. **legacy bootstrap 版本把关（bug 修复本体）**：旧单文件的 `unit.version` 必须落在接受集合内才迁移，否则视为空 unit 留在原地——为 owner 未背书的记录打当前版本戳，会把"可丢弃的过期缓存"变成 domain 层的 schema 硬失败。 +4. **projcache 域声明 `version: 6, compatibleVersions: [3, 4, 5]`**；两个 lineage 字段改为 `.optional()`。唯一消费 stored identity 的读点 `identityMatches` 把缺失归一化为 unseeded lineage（`?? false` / `?? 0`）：对非 fork 会话这是精确值；fork 会话的 expected 是 seeded → 天然 mismatch → 丢弃冷读重建，lineage 绑定的防护不放松。 +5. **投毒态自愈**：v5 戳缺 lineage 字段的文档被声明为兼容，并由 optional schema 接受（内容本就是升级前的真实缓存数据），home 恢复可启动且标题立即可服务。 +6. **schema 校验兜底：`invalidRecords: 'backup-and-skip'`（仅本域声明）**。读兼容之外仍然解析失败的存量记录不再让整个域拒开：domain 层调用后端的 `KvUnit.backupRecord`（json per-record 实现＝把文档改名为 `.json.bak.`，字节留档、不再被读取），用 `logger.error` 打印具体失败信息（域名、表、键、移动去向、zod 失败原因），随后当该记录不存在继续启动；下一次冷读会重建并重写该会话的缓存。**该策略是域级显式声明，缺省仍为 fail-loud**——其他业务域的存量数据校验失败照旧整域拒载；后端没有 `backupRecord` 能力（single 布局、行存储）时也回退 fail-loud。命名沿革：quarantine → backup-and-skip（用户裁决：词要同时含"备份"与"跳过"两义，且与 `.bak` 后缀同源；skip-backup 因 CLI `--skip-X` 惯例存在"不备份"反读而弃用）。对本域而言，该策略取代了 [2026-07-28 存储恢复提案](../../proposed/architecture/2026-07-28-storage-root-and-derived-medium-recovery.zh.md)中 reset/destroy 的恢复途径；该提案对权威介质与整介质损坏仍然有效。 + +### v5 → v6 兼容方式 + +版本 6 只改变当前写入的版本戳，沿用 v5 记录 schema。`compatibleVersions: [3, 4, 5]` 因此同时接受健康的 v5 记录，以及错误 bootstrap 生成的 v5 戳、缺 lineage 记录。当前 schema 允许 lineage 缺失；`identityMatches` 将其解释为 unseeded，并对 seeded 会话拒绝该记录。下一次成功的 checkpoint 会用 v6 戳和完整 lineage 重写已接受的 v5 记录。启动时不单独运行 v5→v6 重写：未接受的版本读作不存在，schema 校验失败的已接受记录则执行 `backup-and-skip`。 + +### 升级矩阵 + +| home 形态 | 修复后行为 | +|---|---| +| v3 单文件（未投毒） | bootstrap 迁移（3 ∈ 接受集）→ 标题立即可服务 | +| v3 + 投毒新树 | 新树文档直接读入（optional 容忍）→ 启动恢复、标题立即可服务 | +| v4 per-record | 文档直接读入（4 ∈ 接受集）→ 标题立即可服务 | +| v5 正常 | 文档直接读入（5 ∈ 接受集）→ 标题立即可服务 | +| v6 当前版本 | 不受影响 | +| fork（seeded）会话的旧记录 | identity mismatch → 丢弃，打开会话时冷读重建（安全侧） | + +## 备选方案 + +- **只丢弃重建**（bootstrap 把关但不声明兼容版本）：启动可修，但升级后 SessionList 标题全丢、要逐会话打开才恢复——不满足升级即用的产品要求。 +- **schema `.default()` 填缺省**：行为与 optional+读点归一化等价，但把"缺失=unseeded"的解释固化进 durable schema 的输出类型；拍板为 optional——schema 如实描述介质上所有被接受的形态，解释权在消费点（2026-09-02 用户裁决）。 +- **域版本回退到 4**：改动很小，但破坏版本单调性、依赖"bootstrap 不查版本"这个 bug 本身、且投毒态与正常 v5 home 的缓存全被丢弃。 + +## 影响 + +- 部署方若把本域路由到 sqlite 后端，得不到任何容忍能力：sqlite 既未实现 `compatibleVersions` 也没有 `backupRecord`，行为退化为原有的严格版本语义（整 unit 版本不匹配仍 `version-mismatch` 拒开；不放松、不出错值）。shipped 组合固定路由 json，此风险仅存在于部署配置层面。 +- optional lineage 字段允许被接受的记录缺少 lineage：无 lineage 的记录会解码为 unseeded。身份比对仍会对 seeded 调用方拒收，逐行 `ver` 守卫仍筛查每个值，残余暴露面只是 unseeded 调用方读到 unseeded 形态的记录——与真实 pre-lineage 记录享有的信任完全相同。 +- `backupRecord` 对同一键的同一分钟内重复备份会覆盖前一份（新字节胜出）；不同分钟、不同键永不冲突。 + +## 测试 + +- `storage-json` 单测：compat 版本戳读入/集合外丢弃/写恒当前版本；legacy bootstrap 仅在版本被接受时迁移（含迁移后文档戳当前版本断言）；`backupRecord` 移档/读缺席/重写/封闭守卫。 +- `storage-domain` 单测：`compatibleVersions`/`invalidRecords` 声明校验；后端无 `backupRecord` 时 backup-and-skip 回退 fail-loud。 +- `session-projection-cache` 单测：缺 lineage 字段的记录对 unseeded 会话按原值服务、对 seeded 会话丢弃。 +- **归档 fixtures 独立恢复测试**（`tests/fixtures.spec.ts` + `tests/fixtures/`）：真实发布物产出的四份介质存档——`v3-single-unit.json`（0.1.1-rc.2 整域单文件）、`v4-session-doc.json`（0.1.2-alpha.3）、`v5-session-doc.json`（0.1.2-alpha.4）、`v5-lineageless-doc.json`（无守卫 bootstrap 的投毒形态，由 v3 记录合成）——逐一走真实存储栈开域，断言列表读出归档标题、且 live 写把文档重写为当前版本（v6 戳 + lineage 字段 + 新值）；外加 schema 失败记录的 backup-and-skip 行为（启动不失败、`.bak` 落盘、日志具体、邻居记录不受累）。 +- 端到端验收，以真实发布物执行：已发布的 0.1.1-rc.2 与 0.1.2-alpha.3 npm 包经各自 web app 造数（真实模型对话 + rename RPC），已发布的 0.1.2-alpha.4 包复现两类故障（含投毒树），修复后构建对纯净 v3、投毒 v3、v4、全新四种 home 形态经 SessionList RPC 原样返回记录在案的标题。 + +未来 bump 流程：新版本结构若可用"optional 字段 + 读点归一化"容忍旧记录，就把旧版本加入 `compatibleVersions`；否则正常 bump（丢弃重建），并把不再兼容的版本从集合中移除。无论哪条路，包 README 都要求 bump 随附归档 fixture 和论证所选处置方式的测试。 diff --git a/.agents/notes/implemented/bug-fix/2026-07-28-load-pre-identity-session-messages.md b/.agents/notes/implemented/bug-fix/2026-07-28-load-pre-identity-session-messages.md deleted file mode 100644 index 6d022cb4b3..0000000000 --- a/.agents/notes/implemented/bug-fix/2026-07-28-load-pre-identity-session-messages.md +++ /dev/null @@ -1,38 +0,0 @@ -# Agent Note: Load sessions persisted before message identity - -Status: implemented - -English | [中文](2026-07-28-load-pre-identity-session-messages.zh.md) - -## Problem - -The identified immutable message change replaced four durable event payloads with complete message values. Existing v0 JSONL Sessions still held the immediately preceding forms: direct `content`/`source` on user and steering events, `content`/`provenance` on assistant events, and `callId`/`content`/`isError` on tool results. Their headers still matched `SESSION_FORMAT_VERSION`, but current-form validation rejected them before resume could construct a live `Session`. - -Changing the message representation without a version bump made those logs indistinguishable at the header level from current v0 logs. The runtime needs a narrow import rule that restores data created by the supported first-party provider without weakening validation for unrelated obsolete or malformed events. - -## Decision - -`PersistenceCoordinator` normalizes the four exact pre-identity message payloads after backend decoding and before current message validation. It wraps their existing semantic fields in the current role-specific message shape and assigns `legacy-message::` as the deterministic imported `MessageId`. A legacy `tool/result` content replacement inherits the imported id of its replacement target, preserving the current content-only rewrite invariant. - -The same normalization runs for `load`, `inspect`, an ownerless loaded state claiming its live session, and HMR prefix adoption. Prefix comparisons therefore compare the live current-shape seed with the same normalized stored view. Current-looking wrappers with missing or invalid fields are not repaired, and unsupported event vocabulary, request headers, versions, and surface relations retain their existing rejection paths. - -The upgrade is read-only. Stored legacy records remain unchanged; a resumed session appends only current-shape events after them. Deterministic identities make repeated loads and a mixed legacy/current log reproduce the same message ids without a backend-specific rewrite transaction. - -## Alternatives considered - -**Reject the logs under the pre-release compatibility stance.** This is the default for unrelated v0 churn, but it strands real first-party sessions even though every old field maps unambiguously to the current message representation. - -**Rewrite the complete stored log in place.** This would canonicalize the artifact but violate the append-only storage contract, require an atomic replacement mechanism, and expand a read compatibility fix into a migration system. - -**Mint random ids on each load.** The messages would satisfy the type shape but lose stable identity across inspect, resume, restart, and mixed legacy/current appends. - -## Consequences - -Pre-identity JSONL Sessions resume with their original message content, sources, assistant provider/model fields, tool correlation, errors, metadata, and surface replacements. The returned events are otherwise indistinguishable from current imported message snapshots and remain deeply frozen. - -This is one explicit same-version import exception, not a general v0 compatibility layer. Adding another exception requires another complete, unambiguous mapping at the persistence boundary; malformed current data continues to fail rather than being guessed into validity. The shared coordinator contract exercises the upgrade against the in-memory reference and JSONL provider, including deterministic reload and tool-result replacement identity. - -## Related - -- [Create every message as an identified immutable value](../architecture/2026-07-28-identified-immutable-message-values.md) — owns the current message identity and immutability contract. -- [Session persistence as an abstract service](../architecture/2026-06-14-session-persistence.md) — owns the append-only backend and resume boundary. diff --git a/.agents/notes/implemented/bug-fix/2026-07-28-load-pre-identity-session-messages.zh.md b/.agents/notes/implemented/bug-fix/2026-07-28-load-pre-identity-session-messages.zh.md deleted file mode 100644 index 86439337b3..0000000000 --- a/.agents/notes/implemented/bug-fix/2026-07-28-load-pre-identity-session-messages.zh.md +++ /dev/null @@ -1,38 +0,0 @@ -# Agent Note: 加载消息标识机制引入前持久化的会话 - -Status: implemented - -[English](2026-07-28-load-pre-identity-session-messages.md) | 中文 - -## 问题 - -带标识的不可变消息变更将四种持久化事件载荷替换为完整消息值。现有 v0 JSONL Session 仍保留紧邻该变更之前的表示：用户事件和 steering（中途引导）事件直接携带 `content`/`source`，assistant 事件携带 `content`/`provenance`，工具结果则携带 `callId`/`content`/`isError`。这些 Session 的 header 仍与 `SESSION_FORMAT_VERSION` 匹配，但当前表示验证会拒绝它们，导致恢复流程无法构造 live `Session`。 - -消息表示改变时没有提升版本，导致这些日志无法仅凭 header 与当前 v0 日志区分。运行时需要一条范围受限的导入规则，既能恢复受支持的 first-party provider 所创建的数据，又不削弱对无关过时事件或格式错误事件的验证。 - -## 决策 - -`PersistenceCoordinator` 会在后端解码之后、当前消息验证之前，规范化消息标识机制引入前的四种特定消息载荷。它将载荷现有的语义字段包装进当前按角色区分的消息形状，并为其分配确定性的导入用 `MessageId`：`legacy-message::`。旧版 `tool/result` 的内容替换会继承替换目标导入后的 id，从而保持当前仅改写内容的不变量。 - -同一项规范化也用于 `load`、`inspect`、无 owner 的已加载状态认领其活跃会话，以及 HMR（热模块替换）前缀接管。因此，前缀比较会将活跃会话的当前形状 seed 与同一份规范化存储视图进行比较。看似当前形状、但字段缺失或无效的包装层不会被修复；不受支持的事件词汇、请求 header、版本和 surface 关系仍沿用现有拒绝路径。 - -这项升级只发生在读取时。存储中的旧版记录保持不变；会话恢复后，只会在其后追加当前形状的事件。确定性标识使重复加载以及新旧形状混合的日志无需执行后端专用的重写事务，也能复现相同的消息 id。 - -## 考虑过的替代方案 - -**按照预发布兼容性立场拒绝这些日志。** 这是处理其他 v0 形状变动的默认方式，但即使每个旧字段都能明确映射到当前消息表示，它仍会导致真实的第一方会话无法恢复。 - -**就地重写完整的存储日志。** 这会使产物规范化，但违反仅追加存储约定，还需要原子替换机制，并将一次读取兼容性修复扩大为迁移系统。 - -**每次加载时随机生成 id。** 这些消息会满足类型形状，却无法在检查、恢复、重启以及新旧形状混合追加之间保持稳定标识。 - -## 后果 - -消息标识机制引入前的 JSONL Session 可以恢复，并保留原始消息内容、来源、assistant 的 provider／model 字段、工具调用关联、错误、元数据和 surface 替换。除此之外，返回事件与当前导入的消息快照无法区分，并且仍然经过深度冻结。 - -这是一个显式的同版本导入例外，而非通用的 v0 兼容层。若要增加另一个例外，必须在持久化边界提供另一套完整且无歧义的映射；当前数据若格式错误，系统仍会拒绝，而不会猜测如何将其变成有效数据。共享协调器约定会通过内存参考实现与 JSONL provider 验证这项升级，包括重新加载时的确定性，以及工具结果替换时的标识继承。 - -## 相关 - -- [将每条消息创建为带标识的不可变值](../architecture/2026-07-28-identified-immutable-message-values.zh.md)：该记录负责当前的消息标识与不可变性约定。 -- [会话持久化作为抽象服务](../architecture/2026-06-14-session-persistence.zh.md)：该记录负责仅追加后端与恢复边界。 diff --git a/.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.i18n.yaml index 7593352560..d8a34eb622 100644 --- a/.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.i18n.yaml +++ b/.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.md -2026-07-29-human-transcript-append-origin.md: 72cafb5a1a149bcdb73d885d4e4fc4ac01e48cd2 -2026-07-29-human-transcript-append-origin.zh.md: 79fcc56b7b60082ebd2946d70419a659ff5687b7 +2026-07-29-human-transcript-append-origin.md: 00f4cf09139bdba4c0fffcdc476b698d098154bc +2026-07-29-human-transcript-append-origin.zh.md: 631ba71d841033c19a9daa21e646468a0e366966 diff --git a/.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.md b/.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.md index 72cafb5a1a..00f4cf0913 100644 --- a/.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.md +++ b/.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.md @@ -8,7 +8,7 @@ English | [中文](2026-07-29-human-transcript-append-origin.zh.md) The terminal and the host history gateway both treated the model-visible surface as the human transcript. A successful compaction replaces a surface range with one checkpoint node, so the moment that replacement landed the terminal dropped every message it shadowed — conversation the user had already read — and re-ran that destructive rebuild on any later replacement. The same confusion reached pagination: `maxMessages` counted every `user/message` and `assistant/message` in the window, so a model-only replacement copy consumed a page slot the human never filled, and the cut could land between a compaction's log-only `compaction/summary` event and the replacement that cites it. -Nothing was lost from the log. `Session.events` still held every original message and full tool result; the surface only decides what the model is sent next. The defect was entirely in the projection. +Nothing was lost from the log. `Session.snapshotEvents()` still returned every original message and full tool result; the surface only decides what the model is sent next. The defect was entirely in the projection. ## Decision diff --git a/.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.zh.md b/.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.zh.md index 79fcc56b7b..631ba71d84 100644 --- a/.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.zh.md +++ b/.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.zh.md @@ -8,7 +8,7 @@ Status: implemented 终端与宿主历史网关都把模型可见的 surface 当作 transcript（文本记录）。一次成功的压缩（compaction）会用一个检查点节点替换一段 surface 范围，因此该替换一落地，终端就丢弃了它所遮蔽的每条消息——那些是用户已经读过的对话——并在此后任何替换到来时重新执行这次破坏性重建。同样的混淆也波及分页：`maxMessages` 统计窗口内的每个 `user/message` 和 `assistant/message`，于是仅供模型使用的替换副本占用了一个人类从未填充的页面额度，而切分点还可能落在压缩的仅日志 `compaction/summary` 事件与引用它的替换之间。 -日志本身没有丢失任何内容。`Session.events` 仍保存着每条原始消息和完整的工具结果；surface 只决定接下来发送给模型的内容。缺陷完全在投影层。 +日志本身没有丢失任何内容。`Session.snapshotEvents()` 仍返回每条原始消息和完整的工具结果；surface 只决定接下来发送给模型的内容。缺陷完全在投影层。 ## 决策 diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-code-runtime-python-settlement-fixes.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-07-31-code-runtime-python-settlement-fixes.i18n.yaml new file mode 100644 index 0000000000..34a16685a4 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-07-31-code-runtime-python-settlement-fixes.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-07-31-code-runtime-python-settlement-fixes.md +2026-07-31-code-runtime-python-settlement-fixes.md: 26a5947b6602d56dc291f2f2e692743522580b12 +2026-07-31-code-runtime-python-settlement-fixes.zh.md: f2e7955a40b00f5b08017f44d2cf0529b2b31bfb diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-code-runtime-python-settlement-fixes.md b/.agents/notes/implemented/bug-fix/2026-07-31-code-runtime-python-settlement-fixes.md new file mode 100644 index 0000000000..26a5947b66 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-07-31-code-runtime-python-settlement-fixes.md @@ -0,0 +1,151 @@ +# Agent Note: Settlement, framing, and lifecycle fixes in the CPython backend + +Status: implemented + +English | [中文](2026-07-31-code-runtime-python-settlement-fixes.zh.md) + +## Problem + +The CPython subprocess backend for PTC mode, built on the [fd-3 frame protocol](../architecture/2026-07-31-code-runtime-python-fd3-protocol.md), resolves every program outcome as a `CodeRunResult`, rejects `run()` only for seam misuse, and disposes to quiescence so no subprocess that stays in the child's process group outlives the fiber (a descendant that escapes the group with `setsid()` is the documented exception — see the package README's Known Limitations). A sequence of review passes surfaced defects that broke those contracts in ways unit coverage did not catch — each hid behind a `/* v8 ignore */`, a captured-callable that read as a fix but was not, a memory effect invisible through the seam, a load-time bound that double-counted, a process-group escalation that a survivor could outlast, a cross-event-loop completion that silently deadlocked, a synchronous throw outside the settlement path, or a transport boundary rendered as a log boundary. Most behavioral fixes ship with a test that fails without them; eleven do not, and are called out as such — the chunked frame read (a syscall-count improvement with no cross-platform-deterministic failure), the confirmed-empty finalize (its only seam-observable effect, a frozen heartbeat, freezes the instant SIGKILL is delivered, which the pre-fix finalize-on-delivery code also produced, and the discriminating probe is the signal-0 check the Alternatives reject as cross-environment-unreliable), the shared stdout/stderr budget (its only seam-observable difference is which entry boundary a mid-stream flush lands on, and that depends on the relative arrival timing of two independent OS pipes, which `os.sched_yield` does not make deterministic; the per-pipe memory bounds it strengthens ARE covered by the single-pipe flood tests), the `flush_line` join-clear-push reorder (it lowers the settlement-flush peak from three copies to two, but the 12x load gate already covers the three-copy newline path, so every gate-admitted config stays within the address space under both orders and no seam-observable difference exists — the memory effect is inside the Python child, unmeasurable through the seam like the shared-budget case), pacing binding replies (its in-tree case only asserts the framed replies still round-trip; the peak it removes lives inside the host's fd-3 writable buffer, unseen through the seam, so the 32.0 MiB → 0.0 MiB reduction is measurable only out-of-tree), dropping a late binding resolution before snapshot (its three assertions all also hold pre-fix, because `sendReply` already dropped after-settlement values — just later than the snapshot), the done-value TOCTOU pre-encoding (a concurrent mutation racing the encode cannot be deterministically constructed through the seam — its daemon-mutation regression only asserts the result is never a `worker-exit`, which is probabilistic and non-discriminating, so under the existing no-fail-before-with-a-reason precedent it is registered as no-fail-before), the stray-UTF-8 budget-flush retention (a budget flush landing exactly on a multibyte boundary is not schedulable through the seam; it is cross-referenced as v8-ignored), and the late-rejection settled guard (a rejection arriving after the run has already settled cannot be deterministically constructed from the seam), and the log-fragment seal (a 25 M single-character drip that would OOM is not deterministically constructible in CI; the in-tree case only asserts it completes and truncates), and the unknown-binding preview cap (the whole-target `JSON.stringify` peak is a transient allocation inside the reply path — its only seam-observable trace is peak memory under a forged near-ceiling `global`/`name`, not measurable through the seam; the in-tree case only asserts the run completes). + +## Decision + +Independent corrections, each in the package that owns the defect. + +### The unknown-binding preview is escaped from a 1 KiB prefix + +The unknown-binding reply built its message with `JSON.stringify` over the WHOLE capped target (`global` + `.` + `name`, each up to `maxValueBytes` code units) — the escaped form could reach ~6x the input under control-heavy fields, a multi-hundred-MB spike near the `maxValueBytes` ceiling that no hostile-peer bound would have admitted. The preview is now escaped from a 1 KiB prefix of the target (enough to identify the binding); `capMessage` still enforces the reply budget. + +### A merged open-log entry is billed once, split across its fragments + +An explicit `flush()` of an unterminated line emits a `log` frame with `open: true`, and the host appends the next frame to the SAME entry (`print('a', end='', flush=True); print('b')` reads back as one `'ab'` entry, not a fake newline). The split-billing arithmetic — first fragment pays quotes+content+separator, continuations and the closing frame pay content only, host caps `logBudget - 1`/`logBudget + 2`, the sub-2-byte walk guard, the child's `_open_started` keying — is stated once, in the [fd-3 protocol note's wire-contract section](../architecture/2026-07-31-code-runtime-python-fd3-protocol.md). + +### Boot-write failure no longer rejects run() + +In [`src/index.ts`](../../../../packages/experimental/code-runtime-python/src/index.ts) the fd-3 boot-frame write is the last statement of `run()`'s synchronous setup. Its `catch` calls `finish()`, and `finish()` reads `wallTimer` and `onAbort` and — through `settle()` — `live`. Those bindings are `const` and were declared AFTER the boot-write, so on a synchronous write failure `finish()` touched them in their temporal dead zone and threw a `ReferenceError`. That escaped the Promise executor and REJECTED `run()`, violating the seam's "outcomes resolve" contract: the caller saw a thrown error instead of the `worker-exit` the catch constructs. The boot-write block is now emitted after `wallTimer`, `onAbort`, and `live` are initialized, and the `/* v8 ignore */` that had hidden the branch from coverage is removed so the catch is measured. + +### Log capture is serialized against settlement + +In [`py/bootstrap.py`](../../../../packages/experimental/code-runtime-python/py/bootstrap.py) the settlement `flush_out()`/`flush_err()` on the main coroutine read and clear each stream's `_pending` list and mutate the shared `LogBuffer` ledger. Model code may start daemon threads whose `print`/`write` mutate the same state concurrently. Capturing the bound method (`out_stream.flush_line`) fixed only WHICH callable settlement invokes, not what it reads mid-flight: an interleaved flush could join a `_pending` list being mutated under it, corrupting the ledger and costing the `done` frame — stranding the run to the wall clock. `LogBuffer` now owns one re-entrant lock shared by both streams; `_LogStream.write` and `flush_line`, and `LogBuffer.push`, take it, so the whole read-modify-write is atomic across threads. + +### Fd-3 residual is copied, not viewed + +Also in `src/index.ts`, after the newline loop over a `Buffer.concat` of the pending fd-3 chunks, the leftover partial line was carried forward as the `subarray` VIEW it was sliced to. A view keeps the entire concat backing allocation alive, so a large frame followed by a tiny trailing fragment pinned a whole frame's worth of memory while `pendingBytes` — set to the fragment's length — reported far less than was retained. The residual is now detached into a fresh right-sized `Buffer` via the exported `detachResidual` helper, letting the concat allocation be collected and keeping `pendingBytes` an honest measure. + +### Output-cap load bound is parse-cap minus envelope, not divided by six + +The load-time check that rejects a `maxLogBytes`/`maxValueBytes` larger than one fd-3 frame can carry divided the frame ceiling by six for worst-case escape expansion. But both budgets are metered in ALREADY-ESCAPED serialized bytes — the host log ledger charges the serialized cost via `jsonStringCostUpTo` (which walks to the cap without allocating the escaped copy), `checkDoneValue` measures the escaped form, and the producing-side `_cap_message` also caps by serialized cost — so a payload admitted under the cap occupies at most `cap + envelope` on the wire; escaping is inside the charge and must not be multiplied in again. The bound is now `FRAME_PARSE_CAP_BYTES - FRAME_ENVELOPE_BYTES` (the receive path rejects raw frames past the 64 MiB parse cap before decoding — the run settles as a worker-exit — so a budget must not exceed what an honest child's frame can carry through that parser), and the unused `MAX_JSON_ESCAPE_EXPANSION` constant is gone. The old bound was not unsafe — it under-admitted — but it silently forbade legitimate large caps. The same load check also rejects a NON-INTEGER `maxLogBytes`/`maxValueBytes`: the child reads each budget through `int(...)`, which floors a float, so `maxLogBytes: 3.5` would truncate at 3 bytes child-side while the host meters the fraction — the two sides enforcing different public config. Rejecting the float at load keeps them in step, matching the worker backend. + +### Same-group survivors are reaped before the fiber goes quiescent + +A model program can leave a descendant in the child's OWN process group (no `setsid`, so `kill(-pid)` reaches it) that ignores SIGTERM but releases the inherited stdout/stderr/fd-3 pipes. The leader then exits, its `close` fires because the pipes drained, and settlement runs while that descendant is still alive. `kill()` arms an `unref`'d SIGKILL timer after SIGTERM; the fix is that `settle()` no longer resolves the run's `finished` promise — nor drops the run from `live` — immediately when an escalation is in flight. Instead, when `killing` is set and the process group is not yet empty (`process.kill(-pid, 0)` does not throw ESRCH), it polls the group on a REF'd timer, bounded by `graceMs + CLOSE_REAP_MARGIN_MS`, and both drops the run from `live` and resolves `finished` only once the group has emptied. The ref'd poll is the load-bearing part: it keeps the host event loop alive until the SIGKILL has actually reaped the group, so even a short-lived host — a one-shot headless run, a config subprocess — cannot exit and reparent the survivor to init. Deferring the `live` removal is what makes a `dispose()` racing a just-resolved `run()` still await the survivor: dropping the run from `live` at settlement (before the reap) would let teardown snapshot an empty set and return while the descendant lived. In the normal case (the leader was the only member) the first probe returns ESRCH and settlement finalizes with zero added latency. `teardown()` awaits each run's `finished`, so disposal is genuinely quiescent, matching its JSDoc — including for a run that already resolved. + +Settlement also CANCELS the SIGKILL timer the moment the group is confirmed empty (the normal path, and when the poll sees the survivor gone). Leaving it armed would expose a PID-reuse hazard: a `kill(-pid)` left pending for up to `graceMs` after the leader was reaped could hit a RECYCLED pgid once the kernel reused the leader's pid, SIGKILLing an unrelated group (`killGroup` swallowing ESRCH does not help — the danger is precisely the kill that SUCCEEDS against a reused group). Clearing it on the empty probe bounds the reuse window to only the genuine-survivor case, where the group cannot be empty to reuse. + +The window the cleared timer cannot cover is closed by an IDENTITY check inside `killGroup`. Every signal it sends is a raw `process.kill(-child.pid, sig)`, which — unlike `child.kill()` — has no handle guard, so it would reach a recycled pgid during the interval between the leader being reaped and `close` firing (measured at 3039 ms with a pipe-holding descendant). The leader's start time is therefore read once at spawn (`/proc//stat` field 22) and re-read before each signal, with two rulings: a reading that is PRESENT AND DIFFERENT means the number now belongs to another process, so the signal is withheld; an ABSENT reading means the leader was already reaped, which is the ordinary case for every escalation — its `/proc` entry is gone while the group it led can still hold the survivor this teardown exists to reap — so the signal proceeds. Absent is also the constant reading on a platform with no `/proc`, where the guard is inert and the pre-existing behavior stands. Reading absent as a mismatch is not hypothetical: the first version did, which withheld the grace SIGKILL and the poll deadline's SIGKILL, and the three same-group heartbeat cases went red on the Linux coverage lane while passing on Darwin, where the reader always returns undefined. + +The reap poll also handles a host event loop BLOCKED past both timers. If a synchronous computation holds the loop from before the poll was scheduled until after its deadline, both the poll timer and the grace-window SIGKILL timer are overdue when the loop resumes, and Node runs the earlier-scheduled poll first — so the grace SIGKILL may never have fired. The deadline branch therefore sends SIGKILL ITSELF (idempotent if the timer already ran) rather than cancelling the unfired escalation, then grants ONE more `CLOSE_REAP_MARGIN_MS` and keeps polling until the group is confirmed empty, because finalizing on mere signal delivery would declare quiescence while the group is still dying. The outer bound on the wait is therefore `graceMs + 2 * CLOSE_REAP_MARGIN_MS`. A final hard bound finalizes if that extra margin elapses with the group still non-empty; that branch carries a `/* v8 ignore */` because it is reachable only where a SIGKILL'd survivor lingers as a zombie and is never `wait()`'d — a container whose PID 1 does not reap orphans — which cannot be built deterministically across CI platforms. The ignore's reason states that environment dependence rather than claiming the branch cannot run, cross-referencing the Alternatives entry that rejected the signal-0 reap assertion for the same reason. + +### RLIMIT clamps against the inherited soft limit, not only the hard + +In [`py/bootstrap.py`](../../../../packages/experimental/code-runtime-python/py/bootstrap.py) `_clamped` bounded a requested `(soft, hard)` rlimit pair by the inherited HARD limit alone. A deployment that inherited a soft limit below the requested one — say inherited `(100, 200)`, requested `(150, 160)` — got back `(150, 160)`, RAISING the effective soft from 100 to 150: for `RLIMIT_AS` that loosens the memory ceiling, for `RLIMIT_CPU` it defers SIGXCPU, both violating "strictest of configured and inherited". `_clamped` now clamps each side against its own inherited counterpart (`RLIM_INFINITY` imposing no ceiling), then pins soft under hard so `setrlimit` never sees an inverted pair. The settlement-time CPU recheck (`die_if_cpu_exhausted`) follows the same rule: it compares spent CPU against the EFFECTIVE clamped `cpu_soft`, not the configured `cpuSeconds`, so a program that traps SIGXCPU and burns past a stricter inherited soft before returning is reported as a timeout rather than a false success. The recheck restores SIG_DFL BEFORE unblocking a program-masked SIGXCPU (`pthread_sigmask(SIG_UNBLOCK, ...)`, captured at import): a program that installed a custom handler AND masked the signal would otherwise have that pending handler run at the unblock — in model code, able to re-mask or raise — so the disposition must already be SIG_DFL when the signal is released; with SIG_DFL first the pending signal kills inside the kernel with no bytecode window, and the `kill` re-raise is the fallback for the never-pending case. The SIGXCPU diagnostic no longer names the configured `cpuSeconds` as the effective budget — under a stricter inherited soft that number is wrong — and instead reports that CPU time was exhausted at "at most the configured N seconds", which holds whichever limit fired. + +### Concurrent binding replies are paced against fd 3 + +`sendReply` ignored `proto.write`'s `false` return, so a program resolving several large values in one `asyncio.gather` round encoded every reply in the same turn and queued all of them in fd 3's writable buffer. Binding resolution carries no seam-level byte cap to bound that, and the failure kills the HOST process rather than failing the run: measured on a 64 KiB-highWaterMark pipe, eight 4 MiB replies buffered 32.0 MiB at once. Replies now go through a queue that encodes and writes one frame at a time, awaiting `drain` when the pipe is full, which measured a 0.0 MiB peak for the same shape. The encode happens inside the loop so a queued reply the run no longer needs is dropped by the `settled` check without ever being serialized. The same `settled` predicate also guards the reply callback AFTER `await fn(...)` but BEFORE `snapshotJsonValue`, so a wide value that resolves after settlement is dropped before its width is walked — the host does not expand a late value for a run whose outcome is already fixed. + +Pacing changes nothing the model can observe. The child matches each reply to its `call` by id from a pump that reads fd 3 continuously, so arrival order was never observable, and the bindings themselves still run concurrently -- only the host's peak memory and the flush timing change. That is also why serializing is not a narrowing of the seam's concurrency contract, which was the reason this was first deferred; that reasoning was wrong. + +### Binding replies complete on the calling loop's thread + +Also in `py/bootstrap.py`, a binding reply Future is created on the loop that ran `dispatch`. When the model calls a binding from a worker THREAD via `asyncio.run(tools.x(...))`, that Future belongs to the thread's loop, not the main loop where `_pump_replies` reads the reply. `asyncio.Future` is not thread-safe: completing it from another thread does not wake its own loop, so the direct `set_result`/`set_exception` left the awaiting thread stranded and the run degraded to a wall-clock timeout. Each pending entry now records its Future's loop alongside the Future, and `_pump_replies` completes it via that loop's `call_soon_threadsafe`. The shared `pending`/`next_id` state is guarded by a `threading.Lock` held across the id claim, the fd-3 write, and the counter advance, so concurrent callers cannot interleave frames out of the id order the host requires. `call_soon_threadsafe` onto a loop that has already CLOSED (the worker thread finished and abandoned its call before the reply arrived) raises `RuntimeError`; that schedule is wrapped so the moot reply is dropped rather than letting the exception end the pump task and strand every later reply. + +### The blocking frame reader reads in chunks, not byte by byte + +`ProtocolChannel.read_frame` — used for the `boot` and `run` handshake frames — read through `FileIO.readline()` on the unbuffered (`buffering=0`) fd, which issues one `os.read(1)` per byte. The `run` frame arrives AFTER `RLIMIT_CPU` is in force, so a legitimate multi-megabyte program burned seconds of CPU in millions of single-byte syscalls before `ast.parse` ran — potentially exhausting the budget on the read alone. It now reads in `_READ_CHUNK_BYTES` chunks into the same `_pending` residual buffer the async reader already uses (the wrapping `os.fdopen` object is gone; both readers call `os.read(self._fd, ...)` directly), so the read cost is trivial and read-ahead past a newline is preserved for the next frame. Both readers track a running scan offset (`find(b"\n", scanned)`) so a large frame accumulated across many chunks is scanned once, not re-scanned from index 0 per chunk — a chunked rescan would have replaced the byte-at-a-time cost with an O(N²) memchr cost on the same large-frame path. + +### Synchronous spawn failure resolves worker-exit, not reject + +Also in `src/index.ts`, `spawn` is called before the settlement Promise executor exists. Node defers only a fixed set of spawn errnos (EACCES, EAGAIN, EMFILE, ENFILE, ENOENT) to an asynchronous `error` event, which the settlement path already turns into a `worker-exit`; every other errno throws SYNCHRONOUSLY from `spawn`. A `pythonBin` longer than the platform PATH_MAX passes the load-time validation (non-empty, no NUL) but makes `spawn` throw `ENAMETOOLONG` here — outside the executor — so `run()` REJECTED instead of resolving, violating resolve-don't-reject, and left this run's just-materialized staging directory on disk since only `settle()` removes it. The `spawn` call and the fd-3 narrowing are now wrapped: a synchronous throw removes the staging directory and resolves the same `worker-exit` class (`python spawn error: …`) the async `error` event produces. + +### Interpreter selection and the child environment settle at load + +`pythonBin` resolves once at plugin load to an executable absolute path and is version-probed under the same scrubbed environment used for runs. The provider requires CPython 3.10 or newer and retains that exact path, so a later `PATH` or working-directory change cannot switch interpreters; an explicit path that is not an executable regular file, an unresolved basename, or an unsupported interpreter fails before `ctx.codeRuntime` registers. The synchronous probe has a fixed five-second deadline and sends `SIGKILL` at that deadline, so a wrapper that ignores `SIGTERM` cannot block plugin load. Each probe and run receives only `TMPDIR`: macOS system Python needs it to avoid emitting a startup warning into captured stderr, while credentials, `PATH`, `HOME`, and every other ambient host value remain unavailable to model code. If the validated executable disappears after activation, the ordinary spawn settlement still resolves `worker-exit`. + +### Stray pipe output is aggregated by line, not by transport chunk + +Also in `src/index.ts`, native stdout/stderr bytes (C-extension writes, `os.write` past the pipe buffer) were pushed to `logs` one entry per Node `data` chunk. `logs` entries are joined with `\n` downstream (PTC mode), so a single newline-free write larger than one pipe read — arriving as several `data` chunks — read back with model-visible newlines inserted at arbitrary transport boundaries. Capture now accumulates raw `Buffer` chunks (the same shape as the fd-3 reader, and for the same reasons: a string `+=` accumulator re-copies the whole residual per chunk and scanning it from index 0 each chunk is a second quadratic — both O(N²) on a large newline-free write), splits on the raw `0x0a` byte, and admits one entry per complete line. A newline never appears inside a UTF-8 multibyte sequence, so decoding each split line is safe without a streaming decoder. Three separate bounds keep the residual from exhausting host memory, each mirroring the fd-3 reader: the fragment list SEALS into finished blocks past `MAX_PENDING_CHUNKS` so a program pacing single-byte `os.write`s cannot accumulate millions of live Buffer objects (whose per-object overhead no byte count sees); the residual is flushed when the COMBINED running SERIALIZED cost of both pipes — tracked through `accrueStrayCost`, which decodes UTF-8 structurally across chunks so a byte that renders as U+FFFD is charged the three bytes that replacement character serializes to — would cross the budget, so a control-char or illegal-UTF-8 flood flushes at a fraction of the raw bytes rather than accumulating a full budget's worth of raw bytes first, and stdout and stderr are metered together rather than each against the full budget (which would let both retain nearly a budget's worth at once, doubling the peak); and once the ledger has truncated, buffering stops so nothing accumulates for output that can never be admitted. `accrueStrayCost` charging illegal bytes their U+FFFD width is the fix for a byte that never begins a valid sequence (0x80–0xC1, 0xF5–0xFF), a multibyte sequence that breaks before completing, or a structurally-complete but ILLEGAL sequence: `toString('utf8')` renders each of those bytes as its own U+FFFD (3 bytes), so it validates each lead's first-continuation range (WHATWG: `E0`→A0-BF, `ED`→80-9F, `F0`→90-BF, `F4`→80-8F, others 80-BF) and charges 3 per byte of any sequence outside it. Charging the raw 1 undercounted a `b"\xff"` flood threefold, and charging only the structural width undercounted a CESU-8 surrogate (`ED A0 80`) or overlong (`E0 80 80`) threefold just as cheaply, letting the residual grow to a full budget's worth of raw bytes before flushing and, near a large `maxLogBytes`, expand toward a ~1 GiB peak in the flush's concat plus `toString`. The per-entry charge on the admitted string is metered by SERIALIZED cost through `jsonStringCostUpTo`, which walks the string to the cap and stops — the previous `Buffer.byteLength(JSON.stringify(text))` allocated the whole escaped form first, so a near-budget control-char-dense line under a large `maxLogBytes` could momentarily allocate over a gigabyte just to measure it. `jsonStringCostUpTo` (the string-walking function, reached by a forged `log` frame whose text `JSON.parse` produced) charges a LONE surrogate the full six escaped bytes (`\uXXXX` under ES2019 well-formed `JSON.stringify`), not the three bytes `Buffer.byteLength` reports for its U+FFFD rendering, so a `\ud800` flood is not undercharged by half; `accrueStrayCost` walks raw bytes and never sees a surrogate as such — a CESU-8-encoded surrogate reaches it as three bytes its per-lead range check rejects, each charged 3 (total 9), matching what `toString('utf8')` renders. The residual is flushed on the pipe's `end` and also explicitly in the `closeDeadline` handler before it destroys the streams: a `setsid` escapee holding the pipes open forces settlement through that path without an `end`, so a final newline-free `os.write(1, …)` the leader emitted before exiting would otherwise be dropped from `logs`. + +### An incompatible output-budget/addressSpaceMb pair is rejected at load + +The child ([`py/bootstrap.py`](../../../../packages/experimental/code-runtime-python/py/bootstrap.py)) builds, charges, and frames a `maxLogBytes` log entry or a `maxValueBytes` completion value under `RLIMIT_AS`, and both ledgers trigger on CHARACTER count against a serialized-BYTE budget. An astral character is one character but four bytes of CPython `str` storage and four UTF-8 bytes, and the heaviest path holds THREE such copies at once: a single `sys.stdout.write(line + "\n")` keeps the caller's `text` argument (alive for the whole `write` call, ~4×), the line slice handed to `LogBuffer.push` (~4×), and the `text.encode("utf-8")` copy `_push_locked` takes to charge and ship it (~4×) — a peak of ~12× the budget. The settlement `flush_line` path holds only two (its `"".join(...)` and that encode copy — it drops the pending chunks before pushing), so the newline path is the binding worst case. When a budget approaches `addressSpaceMb`, a LEGITIMATE near-budget output breaches the address space during that build-and-encode and dies as `worker-exit` instead of truncating (log) or failing as `output-limit` (value). Metering every child write against the address space at runtime is the wrong fix: an exact serialized-cost check on the hot path is either a full `encode` (the allocation being avoided) or a per-character Python loop (which burns the CPU budget — a 10 MB legitimate write hits SIGXCPU under `cpuSeconds: 1`). Both trade one resource bound for another. Instead [`src/index.ts`](../../../../packages/experimental/code-runtime-python/src/index.ts) rejects the incompatible pair at LOAD: each budget times `OUTPUT_BUDGET_WORST_CASE_ADDRESS_SPACE_MULTIPLE` (twelve — the three simultaneous ~4× copies of the newline path) must fit the address space LEFT after a fixed `INTERPRETER_BASELINE_BYTES` reservation for the interpreter's own footprint, with a `>=` so a budget whose worst-case peak exactly equals that room is rejected (that peak plus the reserved baseline is the whole address space, the `RLIMIT_AS` edge). `flush_line` was also made to drop the pending chunks BEFORE its push, matching the newline path's join-clear-push order, so it holds at most the join and its encode rather than three copies. The baseline is reserved SEPARATELY from the multiple because it is a fixed cost, not one that scales with the budget: folding it into the multiple would leave a budget sized right at `addressSpaceMb / 12` admitted while its peak plus the interpreter still overran. Both `maxLogBytes` and `maxValueBytes` are gated symmetrically; the value path's build-and-encode is the same shape. The check runs on every platform, not just where `RLIMIT_AS` is enforced: the incompatibility is a property of the config values, so a Linux deployment OOMs regardless of the host that assembled the config, and a uniform load-time rejection is the fail-loud contract (Darwin skips only the runtime `setrlimit`). This eliminates the class at the config seam rather than patching the write path, so `_LogStream` keeps its original character-count buffering (a valid lower bound on serialized cost, memory-safe once the budget fits the address space). The value path enforces the same discipline in a second place: `_check_done_value` (the byte meter) and `_encode_json_plain` (the frame encoder) walk in O(DEPTH), not O(width). Each container pushes ONE cursor frame that pulls its children one at a time rather than one traversal tuple or stack entry per child — a flat `[0] * 6_000_000` serializes to ~12 MB but a per-element walk allocates ~400 MB of bookkeeping (~28× the serialized size, far past the 12× the gate reserves), so a value the meter admits could OOM on the walk's own frames. With the cursor, the only width-proportional allocation is the output string the meter already bounded. + +The host gate validates against the CONFIGURED `addressSpaceMb`, but a launch environment can inherit a STRICTER `RLIMIT_AS` (a `ulimit -v` wrapper below `addressSpaceMb`), which the bootstrap's `_clamped` correctly lowers the EFFECTIVE limit to — leaving the budgets sized for a ceiling the child never gets. So `bootstrap.py` re-checks both budgets against the effective clamped soft limit after applying it, mirroring the host gate's multiple and baseline, and raises at boot (caught by the setrlimit-phase handler and reported as `exception`, the same class as any other resource-limit-application failure) rather than letting a near-budget output OOM mid-run. The two child constants are kept in step with the host's by the shared reasoning, not a wire field. + +One residual write-path copy is fixed alongside, independent of the config gate: `_LogStream.write`'s newline branch buffered the whole unterminated tail after the last newline (`text[pos:]`) into `_pending` before the flush trigger could bound it, so an early newline followed by a huge tail (`"\n" + "A" * 30 MiB`) made a second full copy of the model's own string — the `RLIMIT_AS` death the path exists to avoid, and one the config gate does not cover because the tail can far exceed `maxLogBytes`. The tail is now sliced to a `remaining + 4`-character prefix (anything past `remaining` characters cannot be admitted, the char count being a lower bound on the serialized cost), which the flush trigger then rejects with the marker. + +### The completion value and error are pre-encoded at their validation point + +In [`py/bootstrap.py`](../../../../packages/experimental/code-runtime-python/py/bootstrap.py), `_done_with_value` now returns the whole terminal frame as a PRE-ENCODED JSON string on the success path: the admitted value is serialized once here, at the validation point inside `_run`'s `try`, as `'{"type": "done", "value": ' + _encode_json_plain(value) + "}"`. The program can keep mutating a returned list/dict from a daemon thread or signal handler after it returns, so a second traversal held at a later point would be a TOCTOU — a concurrent mutation into a non-JSON type would let that later encode throw outside the settlement handler and downgrade a settled run host-side to `worker-exit`. Serializing once, inside the `try` that wraps this call, closes the window: if a concurrent mutation makes the encode throw, the exception handler classifies it as an `exception`, and once the string is produced the frame is written verbatim with no further touching of the live value. `_run` binds the `_done_with_value` ENTRY NAME into a local (`done_with_value_bound`) before the program runs, and `_done_with_value` itself binds `_check_done_value` and `_encode_json_plain` as DEF-TIME default arguments — so a `__main__` rebind of the entry name or those two names after model execution cannot rewrite a legitimate success into an `exception`. The log ledgers (host `logBudget` and child `_remaining`) start ONE byte below the budget, reserving the serialized outer-array envelope (two brackets and n-1 commas over n entries' separators), so a result that exactly exhausts the ledger still serializes within the configured cap; the truncation marker remains envelope, not payload. Once a ledger has truncated, the host clears both stray pipes' buffered output wholesale (every later byte would be no-op'd by `admit`, so retaining it would spend host memory on output that can never be admitted); the child runs `-u` so `sys.__stdout__`/`sys.__stderr__` writes are visible to stray capture immediately, and the settlement flush still drains the original std streams before the done frame (a guard against a buffered wrapper surviving a `sys.__stdout__ = boom` rebind). The constructor rejects a `maxLogBytes` below 64 (the smallest budget with one byte of room for the truncation marker's own serialized form); `maxValueBytes` keeps only the positive-integer requirement, since a completion can be a single byte and the done-frame envelope is seam protocol cost. The marker remains envelope, so a truncated run with admitted entries serializes to at most `maxLogBytes + marker + envelope` (recorded in the package README). A rebind of a transitive dep the encoder reaches (e.g. `_dump_scalar`/`_dump_string`/`json`/`io` — a non-exhaustive set) can still make the encode throw and downgrade a success to an `exception`, which is registered as an accepted residual in the package README. + +`send_done` (a local function inside `_run`) writes the pre-encoded string through a BOUND `channel.write_encoded`, and encodes a dict error frame through a bound `_encode_json_plain` before writing it — it never calls `channel.send_sync`, whose body re-resolves `self.write_encoded` and the module-level `_encode_json_plain` at call time. `_encode_json_plain` and `channel.write_encoded` are bound into locals before the program runs, for the same reason `flush_out`/`flush_err`/`safe_model_traceback` are: the program runs as `__main__`, so `import __main__; __main__.ProtocolChannel.send_sync = boom` or `__main__._encode_json_plain = boom` would otherwise re-resolve the send/encode to a rebranded callable at call time and, when that replacement raises, skip the `done` frame and downgrade a settled verdict to a host-side `worker-exit`. + +### A budget flush retains an unfinished trailing multibyte sequence + +Also in [`src/index.ts`](../../../../packages/experimental/code-runtime-python/src/index.ts), `flushStray(stray, retainPartialTail)` withholds an unfinished multibyte tail from the decode on the BUDGET-triggered flush (the combined-cost threshold in `captureStray`): when the residual ends on a partial UTF-8 lead sequence (`stray.utf8.expected > 0`), the leading byte plus the continuations consumed so far (≤3 bytes) are detached from the frame as the new residual, and only the complete prefix is admitted and decoded. Nothing is admitted when the whole residual is a single unfinished sequence, so a legal, un-finished character is never rendered as U+FFFD in a released, un-truncated entry, and no bogus empty entry is pushed. The withheld tail is re-accrued from a FRESH `stray.utf8` state — metering it against the post-flush `expected > 0` state would charge the carried lead byte as an illegal continuation — so the next chunk continues the walk correctly and the pipe's cost/UTF-8 state is rebuilt over the retained tail. The `end`/`closeDeadline` paths pass `false` and decode the FULL residual unchanged, because there a trailing incomplete sequence is real truncated input and the U+FFFD is the honest render. + +### A late binding rejection returns before formatting the error + +Also in `src/index.ts`, the binding-rejection catch branch now checks `settled` and returns BEFORE formatting `messageOf(error)`. A rejection that arrives after `maxWallMs`, an abort, or dispose has already settled the run would otherwise have `messageOf(error)` run hostile `toString`/`message` getters — spending host heap and time on a run whose outcome is already fixed — before `sendReply` peeks at `settled`. Dropping the framed reply early spares that waste. The running loop's otherwise-mostly-linear reply drain also reads by a head cursor into the queue array instead of `shift()`ing each entry, so a large `asyncio.gather` of wide bindings awaiting fd 3's `drain` drains in linear time rather than O(n²) from repeated re-slicing. + +### A newline-free drip seals its fragments; the CPU soft limit is kept below the hard; the done frame falls back to a fixed literal; the reply queue clears consumed slots + +In [`py/bootstrap.py`](../../../../packages/experimental/code-runtime-python/py/bootstrap.py), `_LogStream` now seals the pending-fragment list past a cap: a newline-free drip of one character per `write` would otherwise accumulate one list slot (and one str object) per call, and under a large `maxLogBytes` a 25 M single-character flood OOMs on its own accounting (plus the same-size list `_push_bounded_prefix` then builds) before the byte budget is reached. Past `_PENDING_MAX_CHUNKS` the current fragments are joined into ONE block moved to a `_pending_blocks` list (the character count is unchanged), bounding the live fragment count exactly as the host-side `captureStray` seal does; the join is only the ≤cap current fragments, never the whole accumulated buffer, so a large drip stays O(B) rather than re-copying the growing block O(B²/cap) times. The HOST-side open hold mirrors the same seal: a budget-sized single-character open flood (`print('x', end='', flush=True)` in a loop is an honest-child path) would otherwise accumulate one fragment array slot plus string object header per frame — ~30× overhead the byte cap cannot see, up to ~2 GB of host auxiliary heap near the `maxLogBytes` load ceiling. Past `MAX_PENDING_CHUNKS` the held fragments coalesce into `openSealed`; the closing-frame merge, `truncateLogs`, and the `finish` residual all read sealed + current fragments and clear the seal. + +`_clamped` also lowers a clamped RLIMIT_CPU soft limit that EQUALS the hard by one unit (when the hard is at least 2). A `ulimit -t N` sets both, and with soft == hard the kernel checks the hard limit in the same tick and SIGKILLs a busy loop directly, so SIGXCPU is never delivered — and the host classifies a CPU overrun ONLY on `signal === 'SIGXCPU'`, so a definite budget exhaustion would be misreported as a `worker-exit`. Lowering the soft one unit gives SIGXCPU a window to fire, so the overrun is reported as a timeout. This is scoped to RLIMIT_CPU (a one-byte soft differential on RLIMIT_AS would only misalign the child's applied limit with the host budget gate, with no signal to preserve). The `hard >= 2` guard leaves a `hard == 1` blind spot — a 1-second dual limit cannot lower the soft to 0, so a definite overrun there is still reported as `worker-exit`. + +`send_done` wraps its encode+write in a try and, on any throw from a rebound transitive name (`_dump_scalar`/`os`), writes a fixed pre-encoded done frame via the `_run`-local bound `_os_write`/`_memoryview`/`_FALLBACK_DONE_FRAME` — so a settled `exception` verdict is never downgraded to a `worker-exit`, and the host still gets a verdict. The reply queue's head-cursor drain clears each consumed slot so a wide written payload is released immediately, bounding host memory to the current backlog under sustained fd-3 backpressure. The exception classes the settlement-path `except` clauses catch are likewise bound before any model code runs: `_BaseException` is a `_run` LOCAL and a closure cell in `_make_failure_reporter`; `_RuntimeError`, `_BindingRejection`, `str`, and `bool` are DEF-TIME default arguments of `_pump_replies` (a body-local `X = X` binding is too late — the model's top-level statements run before the pump's first step). A rebind of `__main__.BaseException` cannot make a program exception escape the handler and lose the `done` frame; a rebind of `__main__.RuntimeError` (or `_BindingRejection`/`str`/`bool`) cannot make a closed-loop scheduling failure escape the pump catch and strand every later reply to the wall clock. + +## Testing + +- `tests/runtime.spec.ts` rejects absent, non-executable, non-CPython, pre-3.10, and unresponsive interpreter configurations at load; changes `PATH` after activation to prove the resolved executable is frozen; removes that executable after activation to preserve the late `worker-exit` path; and asserts a running program sees `TMPDIR` but not `PATH`, `HOME`, or `DEEPSEEK_API_KEY`. The native-output case pins each source stream's order without requiring a total order across independent channels, and the Darwin resource-limit cases state or skip the platform-specific `RLIMIT_AS` behavior. +- `snapshots/session/ptc-python-turn` replaces the headless PTC worker provider with the private Python provider through the real Loader, replays a Python `run_code` program over real bash bindings, and pins the Python SDK prompt, tool schema, dispatch events, captured log, and completion value. +- `tests/boot-write-failure.spec.ts` mocks `spawn` so the fd-3 pipe throws on the boot write — the one path a real subprocess cannot be coerced into — and asserts `run()` resolves a `worker-exit` rather than rejecting. A sibling case makes the mocked `spawn` throw SYNCHRONOUSLY and asserts `run()` still resolves a `worker-exit` and removes its staging directory, keyed off the exact bootstrap path the mocked `spawn` received in its argv so a sibling worker's concurrent staging cannot flake it. Both are isolated in this spec so the real-subprocess suite is untouched. +- `tests/residual-detach.spec.ts` unit-tests `detachResidual`: the carried copy equals the residual, owns a backing store sized to its own length (fixture kept above Node's Buffer pool threshold), and does not share the source frame's `ArrayBuffer`. +- `tests/runtime.spec.ts` — the output-cap case asserts the `parse-cap - envelope` bound (67108800) and its message. A daemon-thread case drives four threads emitting unterminated writes through settlement's flush. A native-write case writes 200 KiB with no newline via `os.write` under a raised `maxLogBytes` and asserts it reads back as EXACTLY one log entry (proving stray output is aggregated by line, not split at pipe-chunk boundaries); a companion writes `b"one\ntwo\nthree"` and asserts three entries (proving real newlines still delimit). A newline-free-flood case writes 2 MiB under a 4 KiB `maxLogBytes` and asserts the capture ends at the truncation marker and stays under budget (proving the residual is bounded by the ledger, not buffered whole); a NUL-flood companion writes 4000 newline-free NULs under the same budget and asserts truncation (proving the residual is charged by SERIALIZED cost, ~6× raw, measured without allocating the escaped copy); an illegal-UTF-8 case paces single-byte `\xff` writes under a 3072-byte budget with `Buffer.concat` wrapped to measure the peak merged buffer, asserting it stays under 2048 (charged at the U+FFFD width 3 the residual flushes near 1024 raw bytes; a raw-byte undercount would let it reach ~3072, so the bound discriminates); a CESU-8/overlong case paces the structurally-well-formed but illegal `ED A0 80` one byte at a time and asserts the same peak bound (charged at the true 9 per sequence it flushes early; charging the structural width 3 triples the peak, so reverting the per-lead range check turns it red); a broken-multibyte case writes a 3-byte lead then a fresh ASCII byte in separate chunks and asserts both a captured `A` and a U+FFFD (exercising `accrueStrayCost`'s cross-chunk broken-sequence branch); a post-truncation case writes a 108-byte payload (under the smallest PIPE_BUF, so one atomic write) whose first line exhausts a 64-byte budget and asserts the second line is dropped (exercising the post-truncation admit no-op in one `data` callback, no v8-ignore); a short-escape case writes a line mixing a tab, quote, backslash, a `\uXXXX` control, a multibyte character, and ASCII, asserting it round-trips verbatim (exercising every branch of `jsonStringCostUpTo`); a reassembly case writes a payload spanning every valid multibyte lead class (E0-range, plain 3-byte, F0, and F4) past the pipe buffer and asserts it round-trips with no U+FFFD (exercising `accrueStrayCost`'s per-lead ranges and cross-chunk reassembly); a lone-surrogate case forges an fd-3 `log` frame flooding 1000 `\ud800` escapes under a 4 KiB budget and asserts truncation (the count sits in the window where charging 3 bytes would admit and 6 bytes truncates, proving the surrogate is charged its full escaped width); a stray-sealing case paces 60000 single-byte newline-free `os.write(1, …)` calls under a raised budget with `Buffer.concat` wrapped to measure copy volume, asserting the trickle coalesces to one entry and the cumulative copy stays under a measured 256 KiB threshold (the sealed shape copies ~120 KB, the re-merge shape ~538 KB, so reverting the seal to a re-merge turns the assertion red — proving the fragment list seals into blocks past `MAX_PENDING_CHUNKS`). A closeDeadline-flush case has the leader write a newline-free diagnostic then spawn a `setsid` orphan holding the pipes open, and asserts the diagnostic survives in `logs` (proving the residual is flushed before the deadline destroys the streams). The same-group reap case spawns a SIGTERM-ignoring same-group descendant that releases the pipes and bumps a heartbeat file; the test asserts the heartbeat STOPS after the grace-window SIGKILL — an assertion robust whether the killed descendant is reaped or lingers as a zombie, so it holds where PID 1 does not wait() orphans. A dispose-after-resolve case asserts `dispose()` of a completed run with a same-group survivor returns only after the survivor stops executing (proving the run stays in `live` until its group is reaped), with an `expect(afterDispose).toBeGreaterThan(0)` guard so the frozen-heartbeat assertion cannot pass vacuously when the file was never written. A deadline case busy-blocks the event loop past both timers and asserts the survivor's heartbeat freezes (proving the poll's deadline arm sends SIGKILL itself rather than cancelling the unfired escalation). The cross-loop case runs a binding from a worker thread's own `asyncio.run` loop while the main coroutine yields with `await asyncio.sleep`, asserting the reply round-trips instead of timing out; a companion case abandons a thread's call so its loop closes, then answers it before a later binding — asserting the pump survives the closed-loop `call_soon_threadsafe` (host-gated ordering makes it deterministic, fail-before hangs the later binding to the wall clock). The inherited-soft-limit case runs the interpreter through a `ulimit -S -t` wrapper that sets a CPU soft limit below `cpuSeconds` and asserts the applied `RLIMIT_CPU` soft is the inherited value, not the configured one (CPU rather than address space, since macOS ignores `ulimit -v`); a companion inherits a 1 s CPU soft, has the program trap SIGXCPU and busy-loop past it, and asserts the settlement recheck reports a timeout — proving the recheck uses the effective soft, not the configured `cpuSeconds`. A control-heavy-diagnostic case raises a NUL-flood exception under a small `maxValueBytes` and asserts the serialized frame fits (proving the diagnostic is metered by serialized cost). A tail-copy case (`maxLogBytes: 256`, `addressSpaceMb: 384`) has the program build a tail in a variable and write `"\n" + tail` where `tail` is 150 MiB — construction peaks at ~2× (~300 MiB, within the address space, so the model's own allocation succeeds and any OOM belongs to the defect path), and the pre-fix whole-tail re-buffer added a third ~150 MiB copy past 384 MiB; the sliced prefix lets the run truncate and complete (Linux-only RLIMIT_AS repro, macOS happy path — the fixture's own construction must fit the address space, a general rule for these RLIMIT_AS cases). An output-budget/address-space case asserts a `maxLogBytes` of 50 MB AND a `maxValueBytes` of 50 MB each reject at load against a 256 MiB `addressSpaceMb` (past the room left after the interpreter baseline when multiplied by the worst-case 12) while the default caps against 512 MiB load, gating both budgets symmetrically; a discriminating case asserts a 48 MiB `maxLogBytes` against a 512 MiB `addressSpaceMb` rejects — 48×8 = 384 MiB fits the 448 MiB budgetable (the old 8× multiple wrongly admitted it) but 48×12 = 576 MiB does not. The ~12× peak the multiple covers is the NEWLINE path's single near-budget write — the caller's own string, the line slice, and the encode copy live at once; the settlement flush is no longer the binding case, because `flush_line` drops the pending chunks before its push and so holds two copies rather than three. An inherited-RLIMIT_AS case runs the interpreter through a `ulimit -v 131072` wrapper with a 32 MiB `maxLogBytes` the configured 512 MiB `addressSpaceMb` admits, and asserts the boot re-check rejects it as an `exception` whose message names the inherited RLIMIT_AS (the 128 MiB inherited limit leaves too little after the baseline; Linux-only, macOS ignores `ulimit -v` and the run proceeds). A non-integer-budget case asserts a fractional `maxLogBytes`/`maxValueBytes` rejects at load. A combined-peak case (`maxLogBytes: 32 MiB`, `maxValueBytes: 32 MiB`, `addressSpaceMb: 512` — each budget admitted alone at 12×) writes ~33M newline-free astral characters (buffered, unflushed) then returns ~33M astral characters, and asserts the run settles as `output-limit` (the value is itself over its 32 MiB budget); pre-fix the unflushed log pending plus the value's build-and-encode peak added past the 512 MiB address space and OOM'd, so flushing the logs before framing the value is what lets the value check complete (Linux-only RLIMIT_AS repro; on macOS the over-budget value reports output-limit under both orders). A wide-value case (`maxValueBytes: 20 MiB`, `addressSpaceMb: 384`) returns `[0] * 6_000_000` — ~12 MB of JSON, under the 20 MiB budget, so it must round-trip; pre-fix the O(width) walk allocated ~400 MB of per-element traversal tuples and encoder stack entries (~28× the serialized size, past the 12× the gate reserves) and OOM'd on a value the meter admitted, while the O(depth) cursor keeps the only width-proportional allocation the output string itself (Linux-only RLIMIT_AS repro; the fixture stays within the address space so it is honest on macOS too). A wide-BINDING-ARGUMENT case (`addressSpaceMb: 384`) calls a binding with `[0] * 6_000_000` and asserts the length echoes back: `_lossless_json_violation` runs on model-built arguments that no child-side budget bounds first, and its per-member tuples measured 459.1 MiB against 0.0 MiB for the cursor. A backtracking case returns a 4 MiB string from a binding and asserts it round-trips: the old scalar regex retained engine state proportional to the string's width (146 MiB at 1 MiB, 557.8 MiB at 4 MiB, past the default 512 MiB), which raised MemoryError inside `_pump_replies` and stranded the call to the wall clock. A control-heavy metering case returns 8M NULs under a 16 MiB `maxValueBytes` and asserts `output-limit`, not `exception`: charging by counting instead of materializing the escaped form measured 19.1 MiB against 228.9 MiB for an identical byte count. An addressSpaceMb-baseline case asserts 64 MiB and 32 MiB reject at load with a message naming `addressSpaceMb`, rather than the budget loop's negative admissible limit. A process-identity case asserts the leader's start time reads stably on Linux and reports undefined on Darwin, the guard that keeps a recycled pgid from receiving this run's SIGTERM. A paced-replies case resolves eight 4 MiB values in one `asyncio.gather` round and asserts the frames round-trip; the peak the fix removes (32.0 MiB buffered → 0.0 MiB) lives in the host's fd-3 writable buffer, invisible through the seam, so this case pins the round-trip and the no-regression match but is measured for its peak only out-of-tree. A late-drop case settles the run on `maxWallMs` and resolves the pending binding afterwards, asserting a `timeout` result, an undefined value, and that the late path actually ran — the three assertions all hold pre-fix because `sendReply` already dropped after-settlement values, just later, so the case pins the ordering, not a seam-observable behavior. A binding-all-names case (`rebinds every name the failure path uses`) asserts a real `ValueError` survives send-done binding — a tested fix pinned by a case that rebinds `__main__.ProtocolChannel.send_sync`, `__main__.ProtocolChannel.write_encoded`, and `__main__._encode_json_plain` — the three names the shipped `send_done` would resolve late if it looked them up at call time — and pins the `done` frame against that call-time look-up skipping it; the done-value TOCTOU pre-encoding, the stray-UTF-8 budget-flush retention, and the late-rejection settled guard are counted among the ten no-fail-before fixes (reasons in the Problem section), not pinned by a fail-before test. A fragment-cap drip case writes 200 000 single-character newline-free `sys.stdout.write` calls and asserts the run completes with a truncation marker rather than a MemoryError (no-fail-before: the 25 M-scale OOM is not deterministically constructible in CI). A dual-limit CPU case runs the interpreter through a `ulimit -t 2` wrapper and busy-loops past it, asserting a `timeout` (the soft limit is lowered to 1 so SIGXCPU fires, not a `worker-exit`). A transitive-name rebind case rebinds `__main__._dump_scalar`, `__main__.os`, `__main__._os_write`, `__main__._memoryview`, and `__main__._FALLBACK_DONE_FRAME` and asserts a done frame still lands as an `exception`, not a `worker-exit` (the real message is replaced by the fixed fallback literal). A BaseException-rebind case rebinds `__main__.BaseException` to `RuntimeError` and raises `ValueError`, asserting the run still reports an `exception`, not a `worker-exit` (the catch uses a pre-program local exception class). A RuntimeError-rebind closed-loop case rebinds `__main__.RuntimeError` to `ValueError` as the first program statement and drives the closed-loop worker pattern, asserting the pump survives the dead-loop reply and delivers the later binding (the pump's `_RuntimeError` is a def-time default argument, so it captures the original before the rebind). A `_done_with_value`-rebind case rebinds `__main__._done_with_value` to a raising function and returns a legitimate value, asserting the run still reports the success (the entry name is a `_run` local bound before the program runs). A `sys.__stdout__`-flush case writes through `sys.__stdout__`/`sys.__stderr__` without an explicit flush and asserts both bytes appear in `logs` (the `-u` unbuffered child plus the settlement drain of the original std streams). The host closes the child's stdin write handle immediately after spawn (the program is an async body that reads nothing from fd 0; a live pipe would hold a host-side handle open past the run, letting a setsid-escaped descendant inheriting fd 0 keep the host process alive). The channel's frame readers bind their decode primitives (`_decode_json_plain`, `os.read`, `_READ_CHUNK_BYTES`, `bytes`, and `len`; `asyncio.get_event_loop` on the async reader) as def-time default arguments, so a `__main__` rebind cannot kill the reply pump; `_decode_json_plain` itself captures `json.loads`/the two regexes/`len`/`isinstance`/`str`/`list` the same way. The reply pump's frame reader is a BOUND METHOD captured by `_run` before the program runs and passed into `_pump_replies` as an explicit argument (a body-local `channel.read_frame_async` lookup would resolve a rebound class attribute, since the pump starts after the program's top-level statements). `send_done`'s frame-shape check uses `_run`'s bound `_str`/`_isinstance` (a program rebinding `__main__.isinstance` cannot make a legitimate success fall into the fixed-literal fallback). `_make_error_class` captures `Exception` and `setattr` as def-time defaults, and dispatch binds `_lossless_json_violation`/`asyncio.get_event_loop`/the channel's send and write primitives into `_run` locals (the frame WRITE goes through def-time bound `write_encoded`+`_encode_json_plain` rather than `send_sync`'s call-time body, and the log sink directly through the bound encode+write primitives, not send_sync) — a rebind of those names before the first binding call cannot break a legitimate call. `compile(wrapped, ..., dont_inherit=True)` stops the module's `from __future__ import annotations` from stringifying the program's type annotations. A basename `pythonBin` that does not resolve on the CURRENT process PATH now fails at LOAD ('does not resolve on PATH', like the empty/NUL checks): the child spawns with `env: {}`, so falling back to the bare name would let execvp silently start a system interpreter from the platform default PATH — a product-visible change from the old run-time ENOENT worker-exit to an early, loud configuration error. The bootstrap resets SIGXCPU to `SIG_DFL` and unblocks it before any model code runs: the child inherits the host's disposition and mask, and a host that ignores or blocks SIGXCPU would let a program run past the soft `RLIMIT_CPU` until the hard limit's SIGKILL — classifying a definite overrun as `worker-exit` instead of `timeout`. (The settle-time enforcer already restores `SIG_DFL` for a program that traps or masks the signal mid-run; this closes the inherited-state gap.) The float encoder's `Decimal(repr(value)).normalize()` runs on a fixed module-level `_FLOAT_CONTEXT = Context(prec=28)` (constructed before any model code): the process-global decimal context would otherwise let a legitimate program's `getcontext().prec = 2` silently round the completion value's digits or `traps[Inexact] = True` make the encode raise, misclassifying a successful run as an exception. A regression case mutates both knobs and asserts a float completion round-trips exactly. The host caps an fd-3 frame's RAW length at 64 MiB (`FRAME_PARSE_CAP_BYTES`) before `toString`/`JSON.parse`: the 256 MiB wire ceiling bounds the bytes, not the decoded structure, and a compact wide frame near it could decode to far more host memory. `maxLogBytes`/`maxValueBytes` are load-bounded to that parser cap so an honest child's frames always fit; a model-constructed binding argument above it is dropped (registered in the README). A frame-cap case writes 65 MiB of `A`s plus a newline and asserts a `worker-exit` with the protocol-frame-exceeded message (the pre-join counter rejects a newline-free single frame; the first-frame check rejects a newline-bearing one BEFORE the join, keeping the peak at one copy of the wire bytes — verified fail-before by reverting to the unconditional counter). A `_decode_json_plain`-rebind case asserts a binding reply still round-trips; a stdin-EOF case has the program read fd 0 and asserts it sees EOF immediately (the destroyed write handle), verified fail-before by disabling the destroy. A frame-cap case writes 65 MiB of `A`s plus a newline and asserts a `worker-exit` with the protocol-frame-exceeded message (the pre-join counter rejects a newline-free single frame; the first-frame check rejects a newline-bearing one BEFORE the join, keeping the peak at one copy of the wire bytes — verified fail-before by reverting to the unconditional counter). A multi-frame case lets two within-cap frames whose combined buffer crosses the cap both survive (the first-frame check, not the byte counter, decides), and a sealing-threshold case writes 64 MiB of 4 KiB atomic newline-free writes plus 12289 more bytes before the first newline, asserting the run is a worker-exit (sealing is the ELSE half of the newline branch, so the newline-bearing chunk always reaches the first-frame check). A pythonBin case resolves a basename against a PATH whose first entry is relative ('.') and asserts the absolute entry is used. An exact-limit case (`maxLogBytes: 64`) writes a 60-character line (62-byte JSON + 1 separator = 63 = the reserved ledger) and a 61-character line (64 > 63), asserting the first is admitted and the second truncates to the marker — the outer-array envelope reservation, pinned at the exact boundary; a companion case asserts `maxLogBytes: 61` rejects at construction. A syntax-label case asserts a parse-time syntax error carries `File ""` (ast.parse passes the same source label as compile and the runtime traceback filter). A SIGXCPU-mask case masks SIGXCPU (`pthread_sigmask`), burns past the soft limit, and returns, asserting a `timeout` (the recheck unblocks before re-raising); a trap+mask companion installs a custom handler that re-masks and asserts the same `timeout` (SIG_DFL is restored before the unblock, so the pending signal kills inside the kernel). + +## Alternatives considered + +**Leave the boot-write `/* v8 ignore */` and fix only the ordering.** Rejected: the ignore is what let the TDZ regression ship uncaught. Removing it makes the catch a measured branch, so per-file 100% coverage now proves the failure path is exercised. + +**Fix the flush race by capturing more bound methods.** Rejected: this is the approach that already failed. Binding a callable fixes reference resolution, not concurrent access to the mutable state the callable reads. Only mutual exclusion over the shared ledger closes the race. + +**Guard the residual with a size threshold (copy only large frames).** Rejected: the branch runs once per newline-bearing read, the copy is bounded by the residual's own length (always a partial line), and a threshold adds a tunable and a second code path for no measurable saving. An unconditional right-sized copy is simpler and always correct. + +**Assert the residual memory effect through the seam.** Rejected: the retained allocation is not observable through `CodeRunResult`, so a black-box test could not distinguish fixed from unfixed. Extracting `detachResidual` makes the backing-store invariant a deterministic unit test instead. + +**Reap the same-group survivor with a fire-and-forget `unref`'d SIGKILL timer alone.** Rejected: an `unref`'d timer does not keep the host alive, so a host that exits within the grace window (a one-shot run, a config subprocess) never fires the SIGKILL and the survivor is reparented to init — the same "no subprocess outlives the fiber" violation in a different shape, and `teardown`'s "await each child's exit" JSDoc would be false. Awaiting the group's death on a ref'd poll keeps the host alive exactly long enough to reap, at zero cost in the common empty-group case. + +**Assert the reap with `process.kill(pid, 0)` throwing ESRCH.** Rejected: a SIGKILL'd process lingers as a zombie until its parent `wait()`s it, and in a container whose PID 1 does not reap orphans the signal-0 probe keeps succeeding, so the assertion would false-fail cross-environment. A heartbeat file that stops advancing detects "no longer executing," which a reaped process and a zombie both satisfy. + +**Complete the cross-loop Future with a plain `set_result` and rely on the GIL.** Rejected: the GIL serializes bytecode but does not make `asyncio.Future` cross-loop-safe — completing a Future from a thread other than its loop's does not schedule its callbacks or wake the loop. `call_soon_threadsafe` on the owning loop is the documented mechanism. + +**Leave the SIGKILL timer armed after settlement (the earlier same-group fix).** Rejected: an unref'd timer left to fire up to `graceMs` after the leader was reaped can `kill(-pid)` a RECYCLED pgid, striking an unrelated group; the danger is the kill that succeeds, which `killGroup`'s ESRCH swallow cannot prevent. Clearing the timer once the group is confirmed empty bounds the reuse window to the genuine-survivor case, where the group is not empty to reuse. + +**Clamp rlimits by the inherited hard limit only.** Rejected: that silently RAISES an inherited soft limit stricter than the request, loosening the very containment the clamp exists to preserve. Clamping each side against its own inherited bound (then pinning soft under hard) keeps the strictest of configured and inherited on both. + +**Bill the host-side `capMessage` backstop by serialized cost, matching the child's `_cap_message`.** Rejected: the two caps guard different things. `_cap_message`'s output re-crosses fd 3 as a JSON string, so its escaped width is what the frame ceiling bounds — serialized billing is required there. `capMessage`'s output goes straight into `CodeRunResult.error.message` and never re-crosses a frame-bounded channel, so the honest measure of what it retains is the raw byte length of the model-visible string. An honest child has already capped by serialized cost and raw length ≤ serialized cost, so a well-formed message passes unchanged; a forged control-heavy message could serialize to ~6× its raw length, but since it travels no capped channel, billing it by that inflated wire width would truncate a legitimately-sized diagnostic for no containment gain. Each side's JSDoc documents the split and points at the other. + +**Push stray pipe output one entry per `data` chunk.** Rejected: `logs` entries are joined with `\n` downstream, so a transport chunk boundary would become a model-visible newline — a single native write split across pipe reads would read back with spurious line breaks. Aggregating by real newline (raw-chunk buffer + split on `0x0a`) matches the child's line-granular `log` frames; the ledger still bounds a newline-free flood by admitting-and-truncating the residual when it would cross the budget. + +**Enforce the fd-3 frame ceiling per-frame (split before the counter check) to avoid a batch-edge false reject.** Rejected: the ceiling check reads the byte counter BEFORE any `Buffer.concat`, precisely so a hostile program cannot force ~2× the 64 MiB frame cap of host memory (the counter and the join are a second copy of everything held). Splitting first to bill a single frame would `Buffer.concat` an over-ceiling frame before rejecting it, reintroducing that doubling — two regression tests assert the pre-concat order for exactly this reason. The batch-edge false reject the per-frame order would fix (a legitimate near-cap frame whose newline-bearing chunk also carries the next frame's leading bytes nudging the counter over the ceiling for one pipe read) is reachable only when `maxLogBytes`/`maxValueBytes` is configured within one pipe read of the 64 MiB cap — orders of magnitude past the 32/64 KiB defaults. The memory-safety bound against hostile input at any config takes precedence over a false reject reachable only at a pathological near-ceiling config; the counter's over-count and this trade-off are documented at the check. + +**Flush the two stray pipes in residual-arrival order when the combined budget crosses.** Rejected: stdout and stderr are independent OS streams whose `data` events already interleave nondeterministically with each other and with the child's own fd-3 `log` frames. The seam's `CodeRunResult.logs` JSDoc reads "in order", which the surrounding text scopes to program-emission order WITHIN a stream — ordering ACROSS concurrent streams is inherently best-effort here, since no host-side flush order can reconstruct the true interleaving the kernel already lost, so preserving a residual's arrival order at the flush buys nothing. A fixed drain order is as valid as any. Tracking a per-residual arrival tick to drain the earlier pipe first would add a branch whose two sides fire only on the relative timing of two OS pipes, which `os.sched_yield` does not make deterministic, so the branch could not be covered without a flaky test — cost with no observable contract benefit. + +**Meter the child log ledger against the address space at runtime instead of rejecting the config at load.** Rejected: an exact serialized-cost check on every child write is either a full `encode` — the very allocation an oversized write cannot afford, which the ledger's cheap pre-check exists to avoid — or a per-character Python loop, which burns the CPU budget (a 10 MB legitimate write hits SIGXCPU under `cpuSeconds: 1`). Each runtime approach trades the memory bound for another resource bound on the hot path. The address-space breach is a property of the `maxLogBytes`/`addressSpaceMb` pair, not of any particular write, so rejecting the incompatible pair once at load eliminates the whole class without any per-write cost and keeps `_LogStream`'s original character-count buffering, which is memory-safe once the budget fits the address space. + +## Consequences + +Interpreter misconfiguration fails before the service is published, every run uses the executable selected at load, and the child receives only `TMPDIR`, removing macOS startup noise without exposing host credentials. The private provider remains absent from shipped profiles while a keyless Loader snapshot pins its source-checkout PTC composition. + +The seam's resolve-don't-reject contract holds on the boot-write path and the synchronous-spawn-failure path, both with measured coverage, and neither strands a staging directory. Log capture is thread-safe at the cost of one re-entrant lock acquisition per write and flush, and stray native output is delimited by its own newlines rather than by transport chunks. Fd-3 residual memory is bounded by the actual retained bytes, and both frame readers scan an accumulating frame once rather than quadratically. The output caps admit every value a frame can carry and reject a non-integer budget at load. Disposal is genuinely quiescent against a same-group survivor — bounded by `graceMs + 2 * CLOSE_REAP_MARGIN_MS`, zero-cost when the group is already empty, with the SIGKILL timer cleared once the group empties so a stale kill cannot strike a recycled pgid — RLIMIT enforcement keeps the strictest of configured and inherited on both soft and hard (and the SIGXCPU diagnostic no longer names a budget the host cannot guarantee), bindings called from model-created threads complete instead of timing out, and the handshake frame reader no longer burns the CPU budget on a large program. Every behavioral fix carries a test that fails without it, except the eleven called out in the Problem section — the chunked frame read (a syscall-count improvement), the confirmed-empty finalize (whose only seam-observable effect freezes at signal delivery, which the pre-fix code also produced), the shared stdout/stderr budget (whose only seam-observable difference turns on nondeterministic cross-pipe arrival timing), the `flush_line` reorder (whose lowered peak stays within what the 12x gate already admits, so no config behaves differently), pacing binding replies (the 32.0 MiB → 0.0 MiB peak reduction lives inside the host's fd-3 writable buffer, unmeasurable through the seam), dropping a late binding resolution before snapshot (its three assertions all hold pre-fix, so it is not a fail-before case), the done-value TOCTOU pre-encoding (its concurrent-mutation race is not deterministically constructible through the seam, and the daemon-mutation regression's only assertion is probabilistic), the stray-UTF-8 budget-flush retention (a budget flush landing on a multibyte boundary is not schedulable through the seam — v8-ignored), and the late-rejection settled guard (a rejection arriving after settlement is not deterministically constructible from the seam), and the log-fragment seal (its 25 M-scale OOM is not deterministically constructible in CI), and the unknown-binding preview cap (its whole-target `JSON.stringify` peak is a transient allocation inside the reply path, unmeasurable through the seam) — so a future regression on the rest goes red. diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-code-runtime-python-settlement-fixes.zh.md b/.agents/notes/implemented/bug-fix/2026-07-31-code-runtime-python-settlement-fixes.zh.md new file mode 100644 index 0000000000..f2e7955a40 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-07-31-code-runtime-python-settlement-fixes.zh.md @@ -0,0 +1,151 @@ +# Agent Note: CPython 后端中的结算、分帧与生命周期修复 + +Status: implemented + +[English](2026-07-31-code-runtime-python-settlement-fixes.md) | 中文 + +## Problem + +用于 PTC mode 的 CPython 子进程后端建立在 [fd-3 帧协议](../architecture/2026-07-31-code-runtime-python-fd3-protocol.zh.md)之上，把每个程序结果都 resolve 成一个 `CodeRunResult`，仅在 seam 被误用时才 reject `run()`，并且会 dispose 到完全停稳，从而没有任何留在子进程自己进程组内的子进程存活得比 fiber 更久（一个用 `setsid()` 逃出该进程组的后代是有文档记载的例外——见该包 README 的 Known Limitations）。一连串审查暴露出一些缺陷，它们以单元测试覆盖率无法捕获的方式破坏了这些契约：每一个都藏在一处 `/* v8 ignore */` 之后、一个读起来像修复但实际并非修复的捕获可调用对象之后、一处透过 seam 不可见的内存效应之后、一处重复计数的加载期上界之后、一处存活者能够熬过的进程组升级之后、一处静默死锁的跨事件循环完成之后、一处位于结算路径之外的同步抛出之后，或者一处被当作日志边界处理的传输边界之后。大多数行为修复都附带一个在缺少它时会失败的测试；有十一处没有，并被如此标注——分块读取帧（一处系统调用次数的改进，没有可跨平台确定性断言的失败）、确认为空后的收尾（它唯一透过 seam 可观测的效应，即一个冻结的心跳，会在 SIGKILL 被投递的瞬间冻结，而修复前"投递即收尾"的代码也会产生同样的结果，用于区分的探测手段是 Alternatives 以跨环境不可靠为由否决的 signal-0 检查）、共享的 stdout／stderr 预算（它唯一透过 seam 可观测的差异，是一次流中冲刷落在哪条条目边界上，而这取决于两条相互独立的 OS 管道的相对到达时机，`os.sched_yield` 并不能使其确定；它所强化的按管道计的内存界限确实由单管道洪泛测试覆盖），`flush_line` 的 join-清空-push 重排序（它把结算期冲刷的峰值从三份副本降到两份，但 12× 加载门本就覆盖了换行路径的三副本峰值，因此每个被门放行的配置在两种顺序下都落在地址空间之内、不存在透过 seam 可观测的差异——该内存效应在 Python 子进程内部，与共享预算那处一样无法透过 seam 度量），节流 binding 回复（它的树内用例只断言分帧后的回复仍能完整往返；它所移除的峰值位于宿主 fd-3 可写缓冲内部、透过 seam 不可见，因此 32.0 MiB → 0.0 MiB 的降幅只能在树外度量），以及在快照之前丢弃迟到的 binding 解析（它的三条断言在修复前同样成立，因为 `sendReply` 本就丢弃结算之后的值——只是比快照晚），完成值的 TOCTOU 预编码（与编码竞态的并发变异无法透过 seam 确定性构造——它的 daemon 变异回归只断言结果永不为 `worker-exit`，这种断言是概率性的、不具有判别力，因此按既有的"无 fail-before 测试且有理由"先例登记为无 fail-before）、stray UTF-8 预算冲刷的扣留（正好落在多字节边界上的预算冲刷无法透过 seam 调度；它被交叉标注为 v8-ignored），以及结算后到达的迟到拒绝的 settled 先查（在运行已经结算之后才到达的拒绝无法从 seam 确定性构造），以及日志分片封存（25 M 级单字符滴灌 OOM 无法在 CI 确定性构造；树内用例只断言其完成并截断），以及 unknown-binding 预览上限（完整 target 的 `JSON.stringify` 峰值是回复路径内的一次瞬时分配——它唯一透过 seam 可观测的痕迹是伪造近上限 `global`／`name` 时的峰值内存，无法透过 seam 度量；树内用例只断言运行完成）。 + +## Decision + +若干处相互独立的修正，各自位于拥有对应缺陷的包中。 + +### unknown-binding 预览从 1 KiB 前缀转义 + +unknown-binding 回复用 `JSON.stringify` 对完整的限幅 target（`global` 加 `.` 加 `name`，各最多 `maxValueBytes` 个 code unit）构造消息——在控制字符密集字段下转义形式可达输入的约 6 倍，在 `maxValueBytes` 上限附近产生数亿字节峰值，这是任何敌意对等方边界都不会放行的。预览现在从 target 的 1 KiB 前缀转义（足以辨识 binding）；`capMessage` 仍执行回复预算。 + +### 合并的 open 日志条目只计费一次，按片段分摊 + +未结束行的显式 `flush()` 发出带 `open: true` 的 `log` 帧，宿主把下一个帧追加到同一条目（`print('a', end='', flush=True); print('b')` 读回为一条 `'ab'` 条目而不是假换行）。拆分计费算术——首片段付引号加内容加分隔符、续接与闭合帧只付内容、宿主 cap `logBudget - 1`／`logBudget + 2`、低于 2 字节的 walk guard、子进程的 `_open_started` 键控——只登记一次，见 [fd-3 协议 note 的 wire-contract 段](../architecture/2026-07-31-code-runtime-python-fd3-protocol.zh.md)。 + +### Boot-write failure no longer rejects run() + +在 [`src/index.ts`](../../../../packages/experimental/code-runtime-python/src/index.ts) 中，fd-3 引导帧写入是 `run()` 同步初始化阶段的最后一条语句。它的 `catch` 会调用 `finish()`，而 `finish()` 读取 `wallTimer` 和 `onAbort`，并通过 `settle()` 读取 `live`。这些绑定是 `const`，且声明在引导写入之后，因此在同步写入失败时，`finish()` 会在它们处于暂时性死区（temporal dead zone）时访问它们，从而抛出一个 `ReferenceError`。该错误逃出了 Promise executor 并 reject 了 `run()`，违反了 seam 的"结果一律 resolve"契约：调用方看到的是一个被抛出的错误，而不是 catch 构造的 `worker-exit`。现在引导写入代码块被放到 `wallTimer`、`onAbort` 和 `live` 初始化之后，并且那处曾把该分支从覆盖率中隐藏的 `/* v8 ignore */` 已被移除，从而使该 catch 被纳入度量。 + +### Log capture is serialized against settlement + +在 [`py/bootstrap.py`](../../../../packages/experimental/code-runtime-python/py/bootstrap.py) 中，主协程上的结算 `flush_out()`／`flush_err()` 会读取并清空各个流的 `_pending` 列表，并修改共享的 `LogBuffer` 账本。模型代码可能启动一些 daemon 线程，其 `print`／`write` 会并发地修改同一状态。捕获绑定方法（`out_stream.flush_line`）只解决了结算调用哪个可调用对象的问题，而没有解决它在执行途中读取什么的问题：一次交错的 flush 可能拼接一个正在其下被修改的 `_pending` 列表，从而破坏账本并丢失 `done` 帧，使该次运行一直拖到墙钟超时。现在 `LogBuffer` 持有一把由两个流共享的可重入锁；`_LogStream.write` 和 `flush_line`，以及 `LogBuffer.push`，都会获取该锁，因此整个读-改-写过程在多线程间是原子的。 + +### Fd-3 residual is copied, not viewed + +同样在 `src/index.ts` 中，在对待处理 fd-3 分片的 `Buffer.concat` 结果按换行符做循环之后，剩余的不完整行被以它被切出的 `subarray` 视图形式向前传递。视图会使整个 concat 的底层分配保持存活，因此一个大帧后面跟着一个极小的尾部片段，会钉住整整一帧大小的内存，而 `pendingBytes`（被设为该片段的长度）报告的值远小于实际保留的内存。现在，残余数据通过导出的 `detachResidual` 辅助函数被分离到一个大小恰当的新 `Buffer` 中，从而让 concat 分配得以被回收，并使 `pendingBytes` 成为一个诚实的度量值。 + +### Output-cap load bound is ceiling minus envelope, not divided by six + +那处在加载期拒绝比单个 fd-3 帧所能承载更大的 `maxLogBytes`／`maxValueBytes` 的检查，会把帧上限除以六以应对最坏情况下的转义膨胀。但这两项预算都是以已转义的序列化字节来计量的：宿主日志账本通过 `jsonStringCostUpTo` 按序列化开销计费（它走到上限而不分配转义后的副本），`checkDoneValue` 度量的是转义后的形式，而生产侧的 `_cap_message` 同样按序列化开销设上限，因此一个在上限之内被放行的载荷在传输时最多占用 `cap + envelope`；转义已经包含在计费之内，不能再被乘一次。现在该上界为 `FRAME_PARSE_CAP_BYTES - FRAME_ENVELOPE_BYTES`（接收路径在解码前拒绝原始长度超过 64 MiB parse cap 的帧——本次运行以 worker-exit 结算——因此预算不得超过诚实子进程的帧能穿过该解析器的值），未使用的 `MAX_JSON_ESCAPE_EXPANSION` 常量已被删除。旧的上界并非不安全（它是放行不足），但它静默地禁止了合法的大上限。这同一处加载检查还会拒绝一个非整数的 `maxLogBytes`／`maxValueBytes`：子进程通过 `int(...)` 读取每一项预算，而 `int(...)` 会对浮点数向下取整，因此 `maxLogBytes: 3.5` 会在子进程侧截断在 3 字节，而宿主却把小数部分也计入——两侧因此强制着不同的公开配置。在加载期拒绝该浮点数使两侧保持一致，与 worker 后端相符。 + +### Same-group survivors are reaped before the fiber goes quiescent + +模型程序可能在子进程自己的进程组里（没有 `setsid`，因此 `kill(-pid)` 能到达它）留下一个后代，它忽略 SIGTERM，但释放了继承而来的 stdout／stderr／fd-3 管道。随后 leader 退出，由于管道已被抽空，它的 `close` 触发，于是结算在那个后代仍存活时运行。`kill()` 在 SIGTERM 之后装设一个 `unref` 的 SIGKILL 定时器；本次修复是，当有一次升级正在进行时，`settle()` 既不立即 resolve 该次运行的 `finished` promise，也不立即把该运行从 `live` 中移除。取而代之的是，当 `killing` 被置位且进程组尚未为空时（`process.kill(-pid, 0)` 不抛出 ESRCH），它在一个 ref 的定时器上轮询该进程组，以 `graceMs + CLOSE_REAP_MARGIN_MS` 为界，仅当进程组已清空后才把该运行从 `live` 移除并 resolve `finished`。这个 ref 的轮询是承重部分：它让宿主事件循环保持存活，直到 SIGKILL 真正回收了该进程组，因此即使是一个短命的宿主（一次性的 headless 运行、一个配置子进程）也无法退出并把存活者 reparent 给 init。把 `live` 的移除推迟，正是让一个与刚返回的 `run()` 竞争的 `dispose()` 仍会 await 该存活者的原因：若在回收之前就把运行从 `live` 移除，teardown 会快照到一个空集合并在后代仍存活时返回。在正常情况下（leader 是唯一成员），第一次探测返回 ESRCH，结算以零附加延迟完成收尾。`teardown()` 会 await 每次运行的 `finished`，因此 dispose 是真正完全停稳的，与其 JSDoc 相符——包括对一个已经 resolve 的运行也是如此。 + +结算还会在进程组被确认为空的那一刻取消 SIGKILL 定时器（正常路径，以及轮询看到存活者已消失时）。让它继续处于装设状态会暴露一个 PID 复用隐患：一个在 leader 被回收后仍挂起长达 `graceMs` 的 `kill(-pid)`，可能在内核复用了 leader 的 pid 之后击中一个被回收（recycled）的 pgid，从而 SIGKILL 掉一个无关的进程组（`killGroup` 吞掉 ESRCH 并无帮助——危险恰恰是那次针对被复用进程组成功执行的 kill）。在空进程组探测时清除它，把复用窗口收窄到只剩真正存在存活者的情形，此时进程组不可能为空以供复用。 + +被清除的定时器覆盖不到的那段窗口，由 `killGroup` 内部的**身份校验**封死。它发出的每个信号都是裸 `process.kill(-child.pid, sig)`——与 `child.kill()` 不同，它没有 handle 守卫——因此在 leader 被回收到 `close` 触发之间的那段间隔里（实测有一个持有管道的后代时可达 3039 毫秒），信号会打到一个被复用的 pgid 上。所以 leader 的启动时刻在 spawn 时读取一次（`/proc//stat` 第 22 字段），并在每次发信号前重读，有两条裁定：读数**存在且不同**意味着该数字现在属于另一个进程，于是扣下信号；读数**缺失**意味着 leader 已被回收，而这正是每次升级的常态——它的 `/proc` 条目已消失，而它曾领导的进程组仍可能持有本次 teardown 要回收的那个存活者——于是信号照常发出。缺失也是无 `/proc` 平台上的恒定读数，那里守卫处于惰性状态、保持原有行为。把缺失读作身份不符并非假想：第一版就是这样做的，它扣下了宽限期的 SIGKILL 与轮询截止分支的 SIGKILL，导致三个同组心跳用例在 Linux coverage lane 上变红，而在 Darwin 上因读取器恒返回 undefined 而通过。 + +回收轮询还会处理宿主事件循环被阻塞、越过两个定时器的情形。如果一次同步计算从轮询被调度之前一直占住事件循环、直到越过它的截止时间，那么当事件循环恢复时，轮询定时器和宽限窗口的 SIGKILL 定时器都已逾期，而 Node 会先运行更早调度的轮询——因此宽限窗口的 SIGKILL 可能从未触发。为此截止时间分支会自己发送 SIGKILL（若定时器已运行则该操作幂等），而不是取消尚未触发的升级，随后再额外给予一个 `CLOSE_REAP_MARGIN_MS`，并持续轮询直到进程组被确认为空，因为仅凭信号投递就收尾会在进程组仍在消亡时宣告完全停稳。因此等待的外层上界为 `graceMs + 2 * CLOSE_REAP_MARGIN_MS`。若这段额外余量耗尽而进程组仍非空，一个最终的硬性上界会收尾；该分支带有一处 `/* v8 ignore */`，因为它仅在一个被 SIGKILL 的存活者作为僵尸进程滞留且从未被 `wait()`——一个 PID 1 不回收孤儿进程的容器——时才可达，而这无法在各 CI 平台上确定性地构造出来。该 ignore 的理由陈述的是这种环境依赖性，而不是声称该分支不可能运行，并交叉引用 Alternatives 中以同样理由否决 signal-0 回收断言的那一条。 + +### RLIMIT clamps against the inherited soft limit, not only the hard + +在 [`py/bootstrap.py`](../../../../packages/experimental/code-runtime-python/py/bootstrap.py) 中，`_clamped` 仅用继承而来的 HARD 限制来约束一个请求的 `(soft, hard)` rlimit 对。一个继承了低于请求值的软限制的部署——比如继承 `(100, 200)`、请求 `(150, 160)`——会拿回 `(150, 160)`，把有效软限制从 100 抬高到 150：对 `RLIMIT_AS` 而言这放松了内存上限，对 `RLIMIT_CPU` 而言它推迟了 SIGXCPU，两者都违反了"取配置值与继承值中最严格者"。现在 `_clamped` 用每一侧各自继承而来的对应值来约束该侧（`RLIM_INFINITY` 不施加任何上限），随后把 soft 钉在 hard 之下，因此 `setrlimit` 绝不会看到一个倒置的对。结算时的 CPU 复查（`die_if_cpu_exhausted`）遵循同一规则：它把已消耗的 CPU 与实际生效的、被夹紧的 `cpu_soft` 比较，而不是与配置的 `cpuSeconds` 比较，因此一个捕获 SIGXCPU、在返回前消耗超过更严格的继承软限制的程序会被报告为 timeout，而非误判为成功。 复查会在解除程序屏蔽的 SIGXCPU（`pthread_sigmask(SIG_UNBLOCK, ...)`，import 期捕获）之前先恢复 SIG_DFL：一个既安装自定义 handler 又屏蔽信号的程序，否则会在 unblock 的瞬间让那个挂起的 handler 以模型代码身份运行（可重新屏蔽或抛出），所以信号被释放时处置必须已是 SIG_DFL；SIG_DFL 在前时，挂起的信号在内核内直接致死、无字节码窗口，而 `kill` 重投递是给从未挂起情形的兜底。SIGXCPU 诊断不再把配置的 `cpuSeconds` 说成实际生效的预算——在一个更严格的继承软限制之下那个数字是错的——而是报告 CPU 时间是在"至多配置的 N 秒"处被耗尽，这一表述无论哪个限制先触发都成立。 + +### 并发 binding 回复对 fd 3 做节流 + +`sendReply` 忽略了 `proto.write` 的 `false` 返回值，因此一个在一轮 `asyncio.gather` 中解析多个大值的程序，会把每条回复都在同一个 turn 内编码、并全部排入 fd 3 的可写缓冲。binding 回复在 seam 层没有字节上限可以约束它，而且这个失败杀掉的是**宿主进程**而不是让本次运行失败：在 highWaterMark 为 64 KiB 的管道上实测，八条 4 MiB 回复会同时缓冲 32.0 MiB。现在回复走一个队列，一次编码并写出一帧，管道写满时等待 `drain`——同样形状实测峰值为 0.0 MiB。编码放在循环内部，因此一条运行已不再需要的排队回复会被 `settled` 检查丢弃，根本不会被序列化。同一个 `settled` 谓词还在 `await fn(...)` 之后、`snapshotJsonValue` 之前守护回复回调，因此一个在结算之后才 resolve 的宽值会在走完它的宽度之前被丢弃——宿主不会为一个结果已定的 run 展开一个迟到的宽值。 + +节流不改变任何模型可观测的行为。子进程通过一个持续读取 fd 3 的 pump、按 id 把每条回复匹配到它自己的 `call`，因此到达顺序从来不可观测，而各 binding 本身仍然并发执行——只有宿主的峰值内存与冲刷时延改变。这也正是为什么串行化并不构成对 seam 并发契约的收窄，而那恰是最初推迟此项的理由；那个理由是错的。 + +### Binding replies complete on the calling loop's thread + +同样在 `py/bootstrap.py` 中，一个绑定回复 Future 是在运行 `dispatch` 的那个事件循环上创建的。当模型通过 `asyncio.run(tools.x(...))` 从一个工作线程调用某个绑定时，该 Future 属于该线程的事件循环，而不是 `_pump_replies` 读取回复的主事件循环。`asyncio.Future` 不是线程安全的：从另一个线程完成它并不会唤醒它自己的事件循环，因此直接的 `set_result`／`set_exception` 会让那个正在等待的线程被搁置，该次运行退化为墙钟超时。现在每个待处理条目都会在记录 Future 的同时记录其 Future 所属的事件循环，`_pump_replies` 通过该事件循环的 `call_soon_threadsafe` 来完成它。共享的 `pending`／`next_id` 状态由一把 `threading.Lock` 保护，该锁跨越 id 认领、fd-3 写入和计数器推进这三步持有，因此并发调用方无法以违反宿主所要求的 id 顺序来交错帧。对一个已经关闭的事件循环（工作线程已结束、在回复到达前放弃了它的调用）调用 `call_soon_threadsafe` 会抛出 `RuntimeError`；该调度被包裹起来，使这个已无意义的回复被丢弃，而不是让异常终结 pump 任务并搁置此后的每一个回复。 + +### The blocking frame reader reads in chunks, not byte by byte + +`ProtocolChannel.read_frame`（用于 `boot` 和 `run` 握手帧）过去通过在无缓冲（`buffering=0`）fd 上的 `FileIO.readline()` 读取，这会为每个字节发起一次 `os.read(1)`。`run` 帧在 `RLIMIT_CPU` 生效之后才到达，因此一个合法的数兆字节程序会在 `ast.parse` 运行之前，在数以百万计的单字节系统调用中烧掉数秒 CPU——有可能仅在读取这一步就耗尽预算。现在它以 `_READ_CHUNK_BYTES` 为单位分块读取，写入异步读取器已经使用的那同一个 `_pending` 残余缓冲区（包裹用的 `os.fdopen` 对象已被移除；两个读取器都直接调用 `os.read(self._fd, ...)`），因此读取开销微不足道，并且越过换行符的预读也为下一帧保留了下来。两个读取器都跟踪一个持续推进的扫描偏移（`find(b"\n", scanned)`），使一个跨多个分块累积起来的大帧只被扫描一次，而不是每来一个分块就从索引 0 重新扫描——分块式重扫会把逐字节的开销换成同一大帧路径上 O(N²) 的 memchr 开销。 + +### Synchronous spawn failure resolves worker-exit, not reject + +同样在 `src/index.ts` 中，`spawn` 是在结算 Promise 的 executor 存在之前被调用的。Node 只把一组固定的 spawn errno（EACCES、EAGAIN、EMFILE、ENFILE、ENOENT）推迟为一个异步的 `error` 事件，而结算路径已经把它转成一个 `worker-exit`；其余每一个 errno 都会从 `spawn` 同步抛出。一个长度超过平台 PATH_MAX 的 `pythonBin` 能通过加载期校验（非空、无 NUL），却会让 `spawn` 在此处抛出 `ENAMETOOLONG`——在 executor 之外——因此 `run()` 会 reject 而不是 resolve，违反了"只 resolve、不 reject"，并且由于只有 `settle()` 才会移除本次运行刚物化出来的暂存目录，它会把该目录留在磁盘上。现在 `spawn` 调用和 fd-3 收窄被包裹起来：一次同步抛出会移除暂存目录，并 resolve 与异步 `error` 事件所产生的同一类 `worker-exit`（`python spawn error: …`）。 + +### 解释器选择与子进程环境在加载期固定 + +`pythonBin` 在插件加载期解析为一个可执行绝对路径，并在与运行时相同的受限环境中完成版本探测。提供方要求 CPython 3.10 或更高版本并保留该确切路径，因此后续 `PATH` 或工作目录变化不能切换解释器；不是可执行普通文件的显式路径、无法解析的裸名或不受支持的解释器都会在 `ctx.codeRuntime` 注册前失败。同步探测有固定的五秒期限，并在期限到达时发送 `SIGKILL`，因此忽略 `SIGTERM` 的包装脚本不能阻塞插件加载。每次探测与运行只接收 `TMPDIR`：macOS 系统 Python 需要它来避免向被捕获的 stderr 发出启动警告，而凭证、`PATH`、`HOME` 与其他宿主环境值均不会进入模型代码。若已校验的可执行文件在激活后消失，普通 spawn 结算仍 resolve 为 `worker-exit`。 + +### Stray pipe output is aggregated by line, not by transport chunk + +同样在 `src/index.ts` 中，原生 stdout／stderr 字节（C 扩展写入、越过管道缓冲区的 `os.write`）过去每来一个 Node `data` 分片就被推入 `logs` 一条条目。`logs` 条目在下游（PTC mode）会用 `\n` 拼接，因此一次大于单次管道读取、且不含换行符的写入——它以若干个 `data` 分片到达——回读时会在任意传输边界处被插入模型可见的换行符。现在捕获会累积原始 `Buffer` 分片（与 fd-3 读取器同一形态，出于同样的原因：一个字符串 `+=` 累加器会为每个分片重新复制整份残余数据，而每个分片都从索引 0 扫描它则是第二重平方——在一次大的不含换行符的写入上二者都是 O(N²)），在原始的 `0x0a` 字节处切分，并为每个完整行准入一条条目。换行符绝不会出现在一个 UTF-8 多字节序列内部，因此对每个切出的行做解码无需流式解码器即可安全进行。三条相互独立的界限使残余数据不至于耗尽宿主内存，每一条都与 fd-3 读取器相对应：分片列表在越过 `MAX_PENDING_CHUNKS` 后会封存（SEAL）为已完成的块，因此一个以单字节 `os.write` 控速的程序无法累积起数以百万计的存活 Buffer 对象（其逐对象开销是任何字节计数都看不到的）；当两个管道合并（COMBINED）的持续推进序列化（SERIALIZED）开销——通过 `accrueStrayCost` 跟踪，它跨分片按结构解码 UTF-8，因此一个渲染为 U+FFFD 的字节会被计入该替换字符序列化后的三个字节——将要越过预算时，残余数据会被冲刷，因此一场控制字符或非法 UTF-8 的洪泛会在原始字节的一小部分处就冲刷，而不是先累积起满满一个预算份额的原始字节，并且 stdout 与 stderr 是合并计量的，而不是各自对照完整预算（那样会让两者同时各保留将近一个预算份额，使峰值翻倍）；而一旦账本已经截断，缓冲便停止，从而不会为永远无法被准入的输出累积任何内容。`accrueStrayCost` 按 U+FFFD 宽度对非法字节计费，正是针对一个从不作为合法序列开头的字节（0x80–0xC1、0xF5–0xFF）、一个在完成前断裂的多字节序列，或一个结构完整但非法（ILLEGAL）的序列的修复：`toString('utf8')` 会把其中每一个这样的字节都渲染为它自己的 U+FFFD（3 字节），因此它校验每个前导字节的首个后续字节范围（WHATWG：`E0`→A0-BF、`ED`→80-9F、`F0`→90-BF、`F4`→80-8F，其余为 80-BF），并对任何落在该范围之外的序列按每字节 3 计费。按原始的 1 计费会把一场 `b"\xff"` 洪泛少计三倍，而只按结构宽度计费同样廉价地把一个 CESU-8 代理项（`ED A0 80`）或过长编码（`E0 80 80`）少计三倍，让残余数据在冲刷前增长到满满一个预算份额的原始字节，并且在一个较大的 `maxLogBytes` 附近，在冲刷的 concat 加 `toString` 中膨胀到约 1 GiB 的峰值。被准入字符串的每条条目计费通过 `jsonStringCostUpTo` 按序列化开销计量，它把字符串走到上限即停止——先前的 `Buffer.byteLength(JSON.stringify(text))` 会先分配出整份转义后的形式，因此在一个较大的 `maxLogBytes` 之下，一行接近预算、控制字符密集的内容，仅仅为了度量它就可能瞬时分配超过一 GB。`jsonStringCostUpTo`（走字符串的那个函数，由一个伪造的、其文本经 `JSON.parse` 产生的 `log` 帧到达）给一个孤立（LONE）代理项计满六个转义字节（在 ES2019 良构 `JSON.stringify` 下为 `\uXXXX`），而不是 `Buffer.byteLength` 为其 U+FFFD 渲染所报告的三个字节，因此一场 `\ud800` 洪泛不会被少计一半；`accrueStrayCost` 走原始字节，从不把一个代理项当作代理项看到——一个 CESU-8 编码的代理项到达它时是三个字节，被它的逐前导字节范围检查所拒绝，每个计 3（共 9），与 `toString('utf8')` 所渲染的相符。该残余数据会在管道 `end` 时冲刷，也会在 `closeDeadline` 处理器销毁流之前被显式冲刷：一个持有管道不放的 `setsid` 逃逸者会迫使结算在没有 `end` 的情况下走那条路径，因此 leader 在退出前发出的最后一次不含换行符的 `os.write(1, …)` 否则会从 `logs` 中被丢弃。 + +### An incompatible output-budget/addressSpaceMb pair is rejected at load + +子进程（[`py/bootstrap.py`](../../../../packages/experimental/code-runtime-python/py/bootstrap.py)）在 `RLIMIT_AS` 之下构建、计费并分帧一条 `maxLogBytes` 的日志条目或一个 `maxValueBytes` 的完成值，而两个账本都是按字符计数、对照一个序列化字节预算触发的。一个星芒面字符是一个字符，但占 CPython `str` 存储的四个字节以及四个 UTF-8 字节，且最重的路径峰值时有三份这样的副本同时存活：一次 `sys.stdout.write(line + "\n")` 会持有调用方的 `text` 实参（在整个 `write` 调用期间存活，约 4 倍）、交给 `LogBuffer.push` 的行切片（约 4 倍）、以及 `_push_locked` 为计费和发送而取的 `text.encode("utf-8")` 副本（约 4 倍）——峰值约为预算的 12 倍。结算期的 `flush_line` 路径只持有两份（它的 `"".join(...)` 与那份 encode 副本——它在 push 之前先丢弃 pending 分块），因此换行路径才是起约束作用的最坏情况。当一项预算逼近 `addressSpaceMb` 时，一次合法的、接近预算的输出会在那次构建加编码期间突破地址空间，并作为 `worker-exit`（日志）而不是截断而终止，或作为 `output-limit`（值）而失败。在运行时对每次子进程写入按地址空间计量是错误的修复：热路径上一次精确的序列化开销检查，要么是一次完整的 `encode`（正是要避免的那次分配），要么是一个逐字符的 Python 循环（它会烧掉 CPU 预算——一次 10 MB 的合法写入会在 `cpuSeconds: 1` 之下触发 SIGXCPU）。两者都是拿一种资源界限换另一种。取而代之，[`src/index.ts`](../../../../packages/experimental/code-runtime-python/src/index.ts) 在加载期（LOAD）拒绝这个不兼容的组合：每项预算乘以 `OUTPUT_BUDGET_WORST_CASE_ADDRESS_SPACE_MULTIPLE`（十二——换行路径三份同时存在的约 4 倍副本）必须放得进为解释器自身占用预留一份固定的 `INTERPRETER_BASELINE_BYTES` 之后剩下的地址空间，并用一个 `>=`，使得一项其最坏情况峰值恰好等于那片余量的预算也会被拒绝（该峰值加上预留的基线正好是整个地址空间，即 `RLIMIT_AS` 边界）。`flush_line` 也被改为在 push 之前先丢弃 pending 分块，与换行路径的 join-清空-push 顺序一致，使它至多只持有 join 及其 encode 副本，而非三份副本。该基线是与倍数分开（SEPARATELY）预留的，因为它是一项固定开销，而非随预算伸缩的开销：把它折进倍数会让一项恰好为 `addressSpaceMb / 12` 大小的预算被放行，而其峰值加上解释器仍会越界。`maxLogBytes` 和 `maxValueBytes` 都被对称地门控；值路径的构建加编码是同一形态。该检查在每个平台上都运行，而不仅在强制 `RLIMIT_AS` 的平台上：这种不兼容是那些配置值的属性，因此一个 Linux 部署无论由哪个宿主组装配置都会 OOM，而一致的加载期拒绝正是 fail-loud 契约（Darwin 仅跳过运行时的 `setrlimit`）。这在配置 seam 处消除了这一类问题，而不是给写入路径打补丁，因此 `_LogStream` 保留它原有的按字符计数的缓冲（一个对序列化开销有效的下界，一旦预算放进地址空间就是内存安全的）。值路径在第二处施加同样的纪律：`_check_done_value`（字节计量器）与 `_encode_json_plain`（帧编码器）都以 O(DEPTH) 而非 O(width) 遍历。每个容器只压入一个游标帧、逐个拉取子元素，而不是每个子元素一个遍历元组或栈条目——一个扁平的 `[0] * 6_000_000` 序列化后约 12 MB，但逐元素遍历会分配约 400 MB 的簿记（约为序列化尺寸的 28 倍，远超门预留的 12 倍），于是一个被计量器放行的值可能因遍历自身的帧而 OOM。改用游标后，唯一与宽度成正比的分配就是计量器已界定的输出字符串。 + +宿主门控是对照配置的（CONFIGURED）`addressSpaceMb` 校验的，但一个启动环境可能继承一个更严格的（STRICTER）`RLIMIT_AS`（一个低于 `addressSpaceMb` 的 `ulimit -v` 包装层），而 bootstrap 的 `_clamped` 会正确地把有效（EFFECTIVE）限制降到该值——从而让这些预算是按一个子进程永远得不到的上限来定尺寸的。因此 `bootstrap.py` 在应用该有效被夹紧的软限制之后，会对照它重新检查两项预算，镜像宿主门控的倍数与基线，并在引导期抛出（被 setrlimit 阶段的处理器捕获，并作为 `exception` 上报——与任何其他资源限制应用失败同属一类），而不是任由一次接近预算的输出在运行途中 OOM。这两个子进程侧常量与宿主侧的保持一致，靠的是共享的推理，而不是一个 wire 字段。 + +在此之外还一并修复了一处残余写入路径的复制，它与配置门控相互独立：`_LogStream.write` 的换行分支会在冲刷触发器能够对其设界之前，先把最后一个换行符之后整个未结束的尾部（`text[pos:]`）缓冲进 `_pending`，因此一个早出现的换行符后跟一个巨大的尾部（`"\n" + "A" * 30 MiB`）会对模型自身的字符串再做一份完整副本——正是这条路径存在所要规避的那次 `RLIMIT_AS` 死亡，而且是配置门控无法覆盖的一次，因为该尾部可能远超 `maxLogBytes`。现在该尾部被切到一个 `remaining + 4` 字符的前缀（超过 `remaining` 字符的任何内容都无法被准入，因为字符计数是序列化开销的下界），随后冲刷触发器会用标记将它拒绝。 + +### 完成值与错误在其校验点处预编码 + +在 [`py/bootstrap.py`](../../../../packages/experimental/code-runtime-python/py/bootstrap.py) 中，`_done_with_value` 现在会在成功路径上把整个终止帧作为一个已预编码的 JSON 字符串返回：被准入的值在这里、在 `_run` 的 `try` 之内的校验点处恰好序列化一次，即 `'{"type": "done", "value": ' + _encode_json_plain(value) + "}"`。程序在返回之后仍可能从 daemon 线程或信号处理器继续变异它返回的 list／dict，因此在一个更晚的点上做第二次遍历会构成一次 TOCTOU——如果一次变异让一个并发变异的后续编码在结算处理器之外抛出，就会把一次已结算的运行在宿主侧降级成 `worker-exit`。在这里、在包裹该调用的 `try` 之内恰好序列化一次，就关上了这个窗口：如果一次并发变异导致编码抛出，异常处理器会把它如实分类为 `exception`；一旦字符串产生出来，该帧就会被逐字写走、不再触碰任何活对象。`_run` 在程序运行前把 `_done_with_value` 的入口名绑成局部（`done_with_value_bound`），而 `_done_with_value` 自身把 `_check_done_value` 与 `_encode_json_plain` 绑定为 def 期默认参数——因此模型执行后对入口名或这两个名字的 `__main__` 重绑无法把一个合法成功改写为 `exception`。日志账本（宿主 `logBudget` 与子进程 `_remaining`）从预算低 1 字节起算，预留序列化外层数组的外壳（两条括号与 n-1 个逗号，覆盖 n 条目的分隔符），因此恰好耗尽账本的结果序列化后仍在配置上限之内。 一旦账本已截断，宿主会整体清空两条 stray 管道的缓冲输出（之后的每个字节都会被 `admit` 变成 no-op，保留它只会把宿主内存花在永远无法准入的输出上）；子进程以 `-u` 运行，使 `sys.__stdout__`/`sys.__stderr__` 的写入对 stray 捕获立即可见，而结算 flush 仍在 done 帧前排空原始 std 流（防御 `sys.__stdout__ = boom` 重绑后残留的缓冲包装）。构造器拒绝低于 64 的 `maxLogBytes`（能为截断标记自身序列化形式留出一字节余量的最小预算）；`maxValueBytes` 只保留正整数要求，因为完成值可以只有一字节、且 done 帧外壳是 seam 协议成本。标记仍是 envelope，因此带已放行条目的截断运行序列化后至多为 `maxLogBytes + marker + envelope`（已记录在包 README）。但编码器到达的一个传递依赖（例如 `_dump_scalar`/`_dump_string`/`json`/`io`——非穷举清单）重绑仍可让编码抛出、把成功降级为 `exception`，这在包 README 中被登记为已接受残余。 + +`send_done`（`_run` 内部的一个局部函数）通过绑定的 `channel.write_encoded` 写出已预编码的字符串，并在写之前用绑定的 `_encode_json_plain` 编码一个 dict 错误帧——它绝不经 `channel.send_sync`，因为后者的函数体会在调用时刻重新解析 `self.write_encoded` 和模块级的 `_encode_json_plain`。`_encode_json_plain` 与 `channel.write_encoded` 在程序运行前就被绑定进局部变量，理由与 `flush_out`／`flush_err`／`safe_model_traceback` 被绑定相同：程序以 `__main__` 运行，因此 `import __main__; __main__.ProtocolChannel.send_sync = boom` 或 `__main__._encode_json_plain = boom` 本会在调用时刻把发送／编码重新解析成被替换的可调用对象，当该替换抛出时跳过 `done` 帧、把已结算的结论降级成宿主侧的 `worker-exit`。 + +### 预算触发的冲刷会扣留下一个未完成的多字节尾序列 + +同样在 [`src/index.ts`](../../../../packages/experimental/code-runtime-python/src/index.ts) 中，`flushStray(stray, retainPartialTail)` 在预算触发的冲刷（`captureStray` 中的合并成本阈值）上会把一个未完成的多字节尾部从解码中扣留：当残余以部分 UTF-8 前导序列结束（`stray.utf8.expected > 0`）时，前导字节加上迄今已消耗的续字节（≤3 字节）会从该帧中分离出来作为新的残余，只有完整的前缀被准入并解码。当整个残余就是单个未完序列时什么都不准入，因此一个合法、未完成的字符绝不会在一个被放行、未截断的条目里被渲染成 U+FFFD，也不会推进一条虚假的空条目。被扣留的尾部会从一个全新的 `stray.utf8` 状态重新累计——若用冲刷后 `expected > 0` 的状态来计量它，会把那些被承载下来的前导字节当作非法续字节计费——从而下一个分块能正确地继续推进，且该管道在保留的尾部之上重建其成本／UTF-8 状态。`end`／`closeDeadline` 路径传入 `false`，原样解码整个残余，因为在那里一个不完整的尾序列是真实的坏输入，U+FFFD 才是如实呈现。 + +### 迟到的 binding 拒绝在格式化错误之前就返回 + +同样在 `src/index.ts` 中，binding 拒绝的 catch 分支现在会在格式化 `messageOf(error)` **之前**检查 `settled` 并返回。一次在 `maxWallMs`、abort 或 dispose 已经把该运行结算之后才到达的拒绝，本会让 `messageOf(error)` 在这之前运行敌意的 `toString`／`message` getter——为一个结局已定的运行花费宿主堆与时间——然后 `sendReply` 才去窥探 `settled`。及早丢弃这一条已分帧的回复省下了这笔开销。运行中那条本就大致线性的回复排空改用队头游标按数组下标读取、而非 `shift()` 逐项弹出，因此一大轮等待 fd 3 的 `drain` 的宽 binding 的 `asyncio.gather` 会以线性时间排空，而不是因反复切片退化成 O(n²)。 + +### 无换行滴灌会封存其分片；CPU 软限制保持在硬限制之下；done 帧回退到固定字面量；回复队列清空已消费槽位 + +在 [`py/bootstrap.py`](../../../../packages/experimental/code-runtime-python/py/bootstrap.py) 中，`_LogStream` 现在会在待处理分片列表越过一个上限时封存它：无换行、每次 `write` 一个字符的滴灌会每次调用累积一个 list 槽位（以及一个 str 对象），在一个大的 `maxLogBytes` 下，25 M 次单字符洪泛会在字节预算达到之前，于其自身记账上 OOM（加上 `_push_bounded_prefix` 随后构造的同规模列表）。越过 `_PENDING_MAX_CHUNKS` 后，当前分片被 join 成一个块并移入 `_pending_blocks` 列表（字符数不变），把存活的碎片数量限制在宿主侧 `captureStray` 封存所做的同等水平；该 join 只针对 ≤cap 的当前分片，从不针对整个累积缓冲，因此大的滴灌保持 O(B)，而不是以 O(B²/cap) 次反复复制不断增长的块。宿主侧 open hold 镜像同样的封存：预算内的单字符 open 洪泛（`print('x', end='', flush=True)` 循环是诚实子进程可达路径）否则会为每帧累积一个片段数组槽位加字符串对象头——约 30× 字节计数看不到的开销，在 `maxLogBytes` 装载上限附近最高约 2 GB 宿主辅助堆。越过 `MAX_PENDING_CHUNKS` 后持有的片段并入 `openSealed`；闭合帧合并、`truncateLogs` 与 `finish` 残段都读取 sealed 加当前片段并清空封存。 + +`_clamped` 还会把钳制出的、与硬限制相等的 RLIMIT_CPU 软限制降低一个单位（当硬限制至少为 2 时）。`ulimit -t N` 会同时设置两者，而当 soft == hard 时，内核会在同一 tick 检查硬限制并直接 SIGKILL 一个忙循环，因此 SIGXCPU 永远不会送达——而宿主只在 `signal === 'SIGXCPU'` 时把 CPU 超限分类为超时，所以一次确定的预算耗尽会被误报为 `worker-exit`。把软限制降低一个单位给 SIGXCPU 一个触发窗口，因此超限会被报告为超时。这仅限定于 RLIMIT_CPU（在 RLIMIT_AS 上的一字节软差异只会让子进程实际应用的限制与宿主预算门失步，没有需要保留的信号）。`hard >= 2` 守卫留下了 `hard == 1` 盲区——一个 1 秒的双限制无法把软限制降到 0，因此那里的确定超限仍被报告为 `worker-exit`。 + +`send_done` 将其 encode+write 包进 try，任何来自被重绑的传递名（`_dump_scalar`/`os`）的抛出都会写入一条固定的预编码 done 帧，经由 `_run` 局部绑定的 `_os_write`/`_memoryview`/`_FALLBACK_DONE_FRAME`——因此一个已结算的 `exception` 判决绝不会被降级为 `worker-exit`，宿主仍会拿到一个判决。回复队列的头游标排空会清除每个已消费槽位，因此一个已写出的宽 payload 会被立即释放，把宿主内存限制在持续的 fd-3 背压下的当前积压量。结算路径各 `except` 子句所捕获的异常类同样在任何模型代码运行之前绑定：`_BaseException` 是 `_run` 的局部与 `_make_failure_reporter` 的闭包单元；`_RuntimeError`、`_BindingRejection`、`str` 与 `bool` 是 `_pump_replies` 的 def 期默认参数（函数体内的 `X = X` 绑定太晚——模型顶层语句会先于泵体首步执行）。重绑 `__main__.BaseException` 无法让程序异常逃出处理器、丢失 `done` 帧；重绑 `__main__.RuntimeError`（或 `_BindingRejection`/`str`/`bool`）无法让闭环调度失败逃出泵的捕获、把每条后续回复搁浅到墙钟超时。 + +## Testing + +- `tests/runtime.spec.ts` 在加载期拒绝缺失、不可执行、非 CPython、低于 3.10 或无响应的解释器配置；在激活后更改 `PATH`，证明已解析的可执行文件保持固定；在激活后删除该文件，保留迟到的 `worker-exit` 路径；并断言运行中的程序能看到 `TMPDIR`，但看不到 `PATH`、`HOME` 或 `DEEPSEEK_API_KEY`。原生输出用例分别固定每个来源流内的顺序，而不要求独立通道之间存在总顺序；Darwin 资源限制用例明确说明或跳过平台特有的 `RLIMIT_AS` 行为。 +- `snapshots/session/ptc-python-turn` 通过真实 Loader 把 headless PTC worker 提供方替换为私有 Python 提供方，经真实 bash binding 重放一个 Python `run_code` 程序，并固定 Python SDK prompt、tool schema、dispatch event、捕获日志与完成值。 +- `tests/boot-write-failure.spec.ts` 对 `spawn` 做 mock，使 fd-3 管道在引导写入时抛出异常（这是真实子进程无法被迫进入的唯一路径），并断言 `run()` resolve 出一个 `worker-exit` 而非 reject。一个同级用例让被 mock 的 `spawn` 同步抛出，并断言 `run()` 仍然 resolve 出一个 `worker-exit`，且会移除它的暂存目录——以被 mock 的 `spawn` 在其 argv 中收到的确切引导路径为准，因此一个同级 worker 的并发暂存不会让它变得不稳定。两者都被隔离在这个 spec 中，因此真实子进程测试套件不受影响。 +- `tests/residual-detach.spec.ts` 对 `detachResidual` 做单元测试：向前传递的副本与残余数据相等、拥有一个大小与其自身长度一致的底层存储（fixture 保持在 Node 的 Buffer 池阈值之上），并且不与源帧的 `ArrayBuffer` 共享。 +- `tests/runtime.spec.ts`：output-cap 用例断言 `parse-cap - envelope` 上界（67108800）及其消息。一个 daemon 线程用例驱动四个线程穿过结算的 flush 发出未结束的写入。一个 native-write 用例在抬高后的 `maxLogBytes` 之下，通过 `os.write` 写入 200 KiB 且不含换行符，断言它回读时恰好是一条日志条目（证明散逸输出是按行聚合的，而不是在管道分片边界处被切开）；一个配套用例写入 `b"one\ntwo\nthree"`，断言得到三条条目（证明真正的换行符仍然起分隔作用）。一个 newline-free-flood 用例在一个 4 KiB 的 `maxLogBytes` 之下写入 2 MiB，断言捕获终止于截断标记且保持在预算之内（证明残余数据受账本约束，而不是被整体缓冲）；一个 NUL-flood 配套用例在同一预算之下写入 4000 个不含换行符的 NUL，断言发生截断（证明残余数据是按序列化开销计费的，约为原始的 6 倍，且在度量时不分配转义后的副本）；一个 illegal-UTF-8 用例在一个 3072 字节的预算之下控速发出单字节 `\xff` 写入，并对 `Buffer.concat` 做包装以度量峰值合并缓冲区，断言它保持在 2048 之下（按 U+FFFD 宽度 3 计费时残余数据在约 1024 原始字节处冲刷；一次原始字节的少计会让它达到约 3072，因此该界限具有区分力）；一个 CESU-8/overlong 用例把结构良构但非法的 `ED A0 80` 一次一个字节地控速发出，断言同样的峰值界限（按每序列真实的 9 计费时它提前冲刷；按结构宽度 3 计费会使峰值增至三倍，因此把逐前导字节范围检查回退会使它变红）；一个 broken-multibyte 用例在分开的分片里先写入一个 3 字节的前导字节、再写入一个新的 ASCII 字节，断言同时捕获到一个 `A` 和一个 U+FFFD（覆盖 `accrueStrayCost` 的跨分片断裂序列分支）；一个 post-truncation 用例写入一个 108 字节的载荷（小于最小的 PIPE_BUF，因此是一次原子写入），其首行耗尽一个 64 字节的预算，断言第二行被丢弃（覆盖单次 `data` 回调中的截断后准入空操作，无需 v8-ignore）；一个 short-escape 用例写入一行混合了制表符、引号、反斜杠、一个 `\uXXXX` 控制字符、一个多字节字符和 ASCII 的内容，断言它原样完成往返（覆盖 `jsonStringCostUpTo` 的每一条分支）；一个 reassembly 用例写入一个跨越每个合法多字节前导字节类别（E0 范围、普通 3 字节、F0 和 F4）、越过管道缓冲区的载荷，断言它原样完成往返且不含 U+FFFD（覆盖 `accrueStrayCost` 的逐前导字节范围与跨分片重组）；一个 lone-surrogate 用例在一个 4 KiB 预算之下伪造一个以 1000 个 `\ud800` 转义洪泛的 fd-3 `log` 帧，断言发生截断（该计数正落在计 3 字节会放行、计 6 字节则截断的窗口内，证明该代理项是按其完整转义宽度计费的）；一个 stray-sealing 用例在抬高后的预算之下控速发出 60000 次单字节、不含换行符的 `os.write(1, …)` 调用，并对 `Buffer.concat` 做包装以度量复制量，断言这股细流合并为一条条目、且累积复制量保持在一个实测的 256 KiB 阈值之下（封存后的形态复制约 120 KB，重新合并的形态复制约 538 KB，因此把封存回退成重新合并会使该断言变红——证明分片列表在越过 `MAX_PENDING_CHUNKS` 后封存为块）。一个 closeDeadline-flush 用例让 leader 写入一段不含换行符的诊断，随后 spawn 一个持有管道不放的 `setsid` 孤儿进程，断言该诊断在 `logs` 中存留下来（证明残余数据在截止时间销毁流之前被冲刷）。same-group 回收用例 spawn 一个忽略 SIGTERM 的同进程组后代，它释放管道并递增一个心跳文件；该测试断言在宽限窗口的 SIGKILL 之后心跳停止：无论被杀死的后代是被回收还是作为僵尸进程滞留，这个断言都成立，因此它在 PID 1 不 wait() 孤儿进程的环境下同样成立。一个 dispose-after-resolve 用例断言，对一个已完成、且存在同进程组存活者的运行调用 `dispose()`，只有在该存活者停止执行之后才返回（证明该运行会一直留在 `live` 中，直到它的进程组被回收），并带有一个 `expect(afterDispose).toBeGreaterThan(0)` 守卫，使得当心跳文件从未被写入时，冻结心跳的断言不会被空洞地通过。一个 deadline 用例忙阻塞事件循环越过两个定时器，断言该存活者的心跳冻结（证明轮询的截止时间分支自身发送 SIGKILL，而不是取消尚未触发的升级）。cross-loop 用例在主协程通过 `await asyncio.sleep` 让出时，从一个工作线程自己的 `asyncio.run` 事件循环运行一个绑定，断言该回复完成往返而不是超时；一个配套用例放弃某个线程的调用，使其事件循环关闭，随后在一个后续绑定之前回答它——断言 pump 在关闭事件循环上的 `call_soon_threadsafe` 之后仍然存活（由宿主门控的顺序使其具有确定性，未修复时会把后续绑定拖到墙钟上挂起）。inherited-soft-limit 用例通过一个 `ulimit -S -t` 包装脚本运行解释器，将 CPU 软限制设为低于 `cpuSeconds`，并断言实际应用的 `RLIMIT_CPU` 软限制是继承来的值，而不是配置的值（用 CPU 而非地址空间，因为 macOS 忽略 `ulimit -v`）。一个配套用例继承 1 秒的 CPU 软限制，让程序捕获 SIGXCPU 并忙循环越过它，断言结算复查报告 timeout——证明复查用的是实际生效的软限制，而不是配置的 `cpuSeconds`。一个 control-heavy-diagnostic 用例在一个较小的 `maxValueBytes` 之下抛出一个 NUL 洪泛异常，断言序列化后的帧能放得下（证明该诊断是按序列化开销计量的）。一个 tail-copy 用例（`maxLogBytes: 256`、`addressSpaceMb: 384`）让程序在一个变量里构建一个尾部并写入 `"\n" + tail`，其中 `tail` 为 150 MiB——构建峰值约 2 倍（约 300 MiB，落在地址空间之内，因此模型自身的分配会成功，任何 OOM 都属于缺陷路径），而修复前的整尾重新缓冲会加上第三份约 150 MiB 的副本、越过 384 MiB；切片后的前缀让该次运行得以截断并完成（仅 Linux 的 RLIMIT_AS 复现，macOS 走顺利路径——fixture 自身的构建必须放进地址空间，这是这些 RLIMIT_AS 用例的一条通用规则）。一个 output-budget/address-space 用例断言一个 50 MB 的 `maxLogBytes` 和一个 50 MB 的 `maxValueBytes` 各自对照一个 256 MiB 的 `addressSpaceMb` 在加载期被拒绝（乘以最坏情况的 12 之后超过解释器基线之后剩下的余量），而默认的各项上限对照 512 MiB 则加载成功，对两项预算对称地门控；一个具区分力的用例断言一个 48 MiB 的 `maxLogBytes` 对照一个 512 MiB 的 `addressSpaceMb` 被拒绝——48×8 = 384 MiB 放得进 448 MiB 的可预算余量（旧的 8× 倍数会错误放行），但 48×12 = 576 MiB 放不进。该倍数覆盖的约 12× 峰值来自换行路径上一次接近预算的写入——调用方自己的字符串、行切片与 encode 副本同时存活；结算期 flush 已不再是承重者，因为 `flush_line` 在 push 之前就丢弃了 pending 分块，因此只持有两份副本而非三份。一个 inherited-RLIMIT_AS 用例通过一个 `ulimit -v 131072` 包装层运行解释器，配以一个配置的 512 MiB `addressSpaceMb` 所允许的 32 MiB `maxLogBytes`，断言引导期的重新检查把它作为 `exception` 拒绝、且其消息点名了继承的 RLIMIT_AS（128 MiB 的继承限制在基线之后剩下的太少；仅 Linux，macOS 忽略 `ulimit -v`，该次运行会继续）。一个 non-integer-budget 用例断言一个小数的 `maxLogBytes`／`maxValueBytes` 在加载期被拒绝。一个 combined-peak 用例（`maxLogBytes: 32 MiB`、`maxValueBytes: 32 MiB`、`addressSpaceMb: 512`——每项预算单独都被 12× 门放行）写入约 33M 个不含换行符的星芒面字符（缓冲、未冲刷）后返回约 33M 个星芒面字符，断言该次运行以 `output-limit` 结算（该值本身就超过它 32 MiB 的预算）；修复前未冲刷的日志 pending 加上值的构建加编码峰值会一起越过 512 MiB 地址空间而 OOM，因此在分帧值之前先冲刷日志正是让值检查得以完成的原因（仅 Linux 的 RLIMIT_AS 复现；在 macOS 上超预算的值在两种顺序下都报 output-limit）。一个宽完成值用例（`maxValueBytes: 20 MiB`、`addressSpaceMb: 384`）返回 `[0] * 6_000_000`——JSON 约 12 MB、低于预算，因此必须成功往返；修复前 O(width) 的遍历为每个元素分配遍历元组与编码器栈项（约为序列化尺寸的 28×，超出门保留的 12×），在一个计量器已放行的值上 OOM，而 O(depth) 游标使唯一按宽度分配的只剩输出字符串本身。一个宽 binding 实参用例（`addressSpaceMb: 384`）以 `[0] * 6_000_000` 调用 binding 并断言长度回传：`_lossless_json_violation` 运行在模型构造的实参上，子进程侧没有任何字节预算先行约束，其逐元素元组实测 459.1 MiB，而游标为 0.0 MiB。一个回溯用例从 binding 返回一个 4 MiB 字符串并断言其成功往返：旧的标量正则保留的引擎状态与字符串宽度成正比（1 MiB 时 146 MiB，4 MiB 时 557.8 MiB，超过默认的 512 MiB），会在 `_pump_replies` 内抛出 MemoryError 并把该次调用搁置到墙钟。一个 control-heavy 计费用例在 16 MiB 的 `maxValueBytes` 之下返回 8M 个 NUL，断言得到 `output-limit` 而非 `exception`：以计数替代物化转义形式来计费，在字节数完全相同的前提下实测 19.1 MiB 对 228.9 MiB。一个 addressSpaceMb 下界用例断言 64 MiB 与 32 MiB 在加载期被拒绝，且消息点名 `addressSpaceMb`，而不是预算循环给出的负数上限。一个进程身份用例断言 leader 的启动时刻在 Linux 上可稳定读取、在 Darwin 上报告 undefined，这正是使被复用的 pgid 不会收到本次运行 SIGTERM 的那道守卫。一个 paced-replies 用例在一轮 `asyncio.gather` 中 resolve 八条 4 MiB 的值，断言这些帧能够往返；本修复移除的峰值（32.0 MiB 缓冲 → 0.0 MiB）位于宿主 fd-3 可写缓冲内部、透过 seam 不可见，因此该用例钉住的是往返与无回归匹配，其峰值只能在树外度量。一个 late-drop 用例让该次运行在 `maxWallMs` 上结算，随后才 resolve 那个 pending 的 binding，断言得到一个 `timeout` 结果、一个 undefined 值、且迟到路径确实被执行过——这三条断言在修复前也全部成立，因为 `sendReply` 本就丢弃结算之后的值、只是更晚，因此该用例钉住的是顺序，而非一个透过 seam 可观测的行为。一个"绑定全部用名"用例（`rebinds every name the failure path uses`）断言一个真实的 `ValueError` 在 send-done 绑定之后仍然存活——这是一个有测修复，由一个逐名重绑 `__main__.ProtocolChannel.send_sync`、`__main__.ProtocolChannel.write_encoded` 与 `__main__._encode_json_plain` 的用例钉住——这三个名字正是 shipped 的 `send_done` 若做调用时刻查找时会迟解析的那三个——并钉住 `done` 帧不被一次调用时刻的查找跳过；而完成值的 TOCTOU 预编码、stray UTF-8 预算冲刷的扣留与结算后到达的迟到拒绝的 settled 先查，都被计入那十处无 fail-before 修复（理由见 Problem 段），不由 fail-before 测试钉住。 一个 fragment-cap 滴灌用例写入 200 000 次单字符无换行的 `sys.stdout.write` 调用，断言该次运行以截断标记完成而非 MemoryError（no-fail-before：25 M 规模的 OOM 无法在 CI 中确定性构造）。一个 dual-limit CPU 用例通过一个 `ulimit -t 2` 包装脚本运行解释器并忙循环越过它，断言得到 `timeout`（软限制被降到 1，因此 SIGXCPU 触发，而非 `worker-exit`）。一个传递名重绑用例重绑 `__main__._dump_scalar`、`__main__.os`、`__main__._os_write`、`__main__._memoryview` 与 `__main__._FALLBACK_DONE_FRAME`，断言仍有一帧 done 以 `exception` 落地，而非 `worker-exit`（真实消息被固定兜底字面量替换）。 一个 BaseException 重绑用例把 `__main__.BaseException` 重绑为 `RuntimeError` 并抛出 `ValueError`，断言该次运行仍报告 `exception`，而非 `worker-exit`（catch 用的是程序运行前的局部异常类）。一个 RuntimeError 重绑闭环用例把 `__main__.RuntimeError` 重绑为 `ValueError` 作为程序首条语句，并驱动闭环 worker 模式，断言泵存活于死循环回复、投递后续 binding（泵的 `_RuntimeError` 是 def 期默认参数，因此在重绑前捕获原始值）。 一个 `_done_with_value` 重绑用例把 `__main__._done_with_value` 重绑为一个抛出函数并返回合法值，断言该次运行仍报告成功（入口名是程序运行前绑定的 `_run` 局部）。 一个 `sys.__stdout__` flush 用例不经显式 flush 直接通过 `sys.__stdout__`/`sys.__stderr__` 写入，断言两个字节都出现在 `logs` 中（`-u` 无缓冲子进程加上结算对原始 std 流的排空）。 宿主在 spawn 后立即关闭子进程的 stdin 写句柄（程序是不读 fd 0 的 async 函数体；存活的管道会在运行结束后继续持有宿主侧句柄，让继承 fd 0 的 setsid 逃逸后代拖住宿主进程）。通道的帧读取器把解码原语（`_decode_json_plain`、`os.read`、`_READ_CHUNK_BYTES`、`bytes`，以及 `len`；异步读取器还有 `asyncio.get_event_loop`）绑定为 def 期默认参数，因此 `__main__` 重绑无法杀死回复泵；`_decode_json_plain` 自身以同样方式捕获 `json.loads`/两个正则/`len`/`isinstance`/`str`/`list`。回复泵的帧读取器是 `_run` 在程序运行前捕获的绑定方法，以显式参数传入 `_pump_replies`（函数体内的 `channel.read_frame_async` 查找会解析被重绑的类属性，因为泵在程序顶层语句之后才启动）。 `send_done` 的帧形判别使用 `_run` 绑定的 `_str`/`_isinstance`（程序重绑 `__main__.isinstance` 无法让合法成功落入固定字面量兜底）。 `_make_error_class` 把 `Exception` 与 `setattr` 捕获为 def 期默认值，dispatch 把 `_lossless_json_violation`/`asyncio.get_event_loop`/通道的 send 与 write 原语绑定进 `_run` 局部（帧写入走 def 期绑定的 `write_encoded`+`_encode_json_plain`，而非 `send_sync` 的调用期函数体；日志 sink 直接走绑定的 encode+write 原语，而非 send_sync）——在首次 binding 调用前重绑这些名字无法破坏合法调用。`compile(wrapped, ..., dont_inherit=True)` 阻止本模块的 `from __future__ import annotations` 把程序的类型注解字符串化。裸名 `pythonBin` 在 CURRENT 进程 PATH 上无法解析时现在于 LOAD 期失败（'does not resolve on PATH'，与空／NUL 检查一致）：子进程以 `env: {}` spawn，回退到裸名会让 execvp 从平台默认 PATH 静默启动一个调用方从未要求的系统解释器——这是从旧的运行期 ENOENT worker-exit 到早期、响亮的配置错误的可见行为变更。bootstrap 在任何模型代码运行前把 SIGXCPU 重置为 `SIG_DFL` 并解除屏蔽：子进程继承宿主的处置与掩码，忽略或屏蔽 SIGXCPU 的宿主会让程序越过软 `RLIMIT_CPU` 一直跑到硬限的 SIGKILL——把确定的超限分类成 `worker-exit` 而非 `timeout`。（结算期 enforcer 已为在运行中 trap 或屏蔽信号的程序恢复 `SIG_DFL`；这里补上继承态的缺口。）浮点编码器的 `Decimal(repr(value)).normalize()` 运行在模块加载期构造的固定 `_FLOAT_CONTEXT = Context(prec=28)` 上（在任何模型代码之前）：进程全局 decimal context 否则会让合法程序的 `getcontext().prec = 2` 静默舍入完成值的数字，或让 `traps[Inexact] = True` 使编码抛异常、把成功运行误判为 exception。一个回归用例同时改动两个旋钮并断言浮点完成值精确往返。 宿主在 `toString`/`JSON.parse` 之前把 fd-3 帧的原始长度限制在 64 MiB（`FRAME_PARSE_CAP_BYTES`）：256 MiB 线上上限约束的是字节而非解码后的结构，接近它的紧凑宽帧解码后可能占用远超线上字节的宿主内存。`maxLogBytes`/`maxValueBytes` 在加载期被限制到该解析器上限，因此诚实子进程的帧总能放得下；模型构造的超限 binding 实参被丢弃（已登记在 README）。一个帧上限用例写入 65 MiB 的 `A` 加一个换行，断言得到携带 protocol-frame-exceeded 消息的 `worker-exit`（pre-join 计数拒绝无换行的单帧；第一帧检查在 join 之前拒绝带换行的帧，使峰值保持在线上字节的一份拷贝——通过回退到无条件计数验证 fail-before）。一个 `_decode_json_plain` 重绑用例断言 binding 回复仍能往返；一个 stdin-EOF 用例让程序读取 fd 0 并断言它立即看到 EOF（被销毁的写句柄），通过禁用销毁验证 fail-before。一个帧上限用例写入 65 MiB 的 `A` 加一个换行，断言得到携带 protocol-frame-exceeded 消息的 `worker-exit`（pre-join 计数拒绝无换行的单帧；第一帧检查在 join 之前拒绝带换行的帧，使峰值保持在线上字节的一份拷贝——通过回退到无条件计数验证 fail-before）。一个多帧用例让两个都在上限内、但合并缓冲越过上限的帧都存活（由第一帧检查而非字节计数决定）；一个 sealing 阈值用例写入 64 MiB 的 4 KiB 原子无换行写，再加首个换行前的 12289 字节，断言该次运行为 worker-exit（sealing 是换行分支的 ELSE 半支，因此带换行的 chunk 总是抵达第一帧检查）。一个 pythonBin 用例把一个裸名解析到 PATH 首项为相对条目（`.`）的路径，断言使用绝对条目。 一个 exact-limit 用例（`maxLogBytes: 64`）写入一个 60 字符行（62 字节 JSON + 1 分隔符 = 63 = 预留后的账本）与一个 61 字符行（64 > 63），断言前者放行、后者截断为仅标记——钉住外层数组外壳预留的精确边界；一个配套用例断言 `maxLogBytes: 61` 在构造期被拒绝。一个语法标签用例断言解析期语法错误携带 `File ""`（`ast.parse` 与 compile 及运行期 traceback 过滤使用同一来源标签）。一个 SIGXCPU 屏蔽用例屏蔽 SIGXCPU（`pthread_sigmask`）、越过软限并返回，断言得到 `timeout`（复查在重投递前解除屏蔽）；一个 trap+mask 配套用例安装一个重新屏蔽的自定义 handler 并断言同样的 `timeout`（SIG_DFL 在 unblock 前恢复，因此挂起信号在内核内致死）。 + +## Alternatives considered + +**保留引导写入处的 `/* v8 ignore */`，只修复顺序。** 已否决：正是那处 ignore 让这个 TDZ 回归得以未被发现地进入代码库。移除它使该 catch 成为被度量的分支，因此按文件计的 100% 覆盖率现在能证明该失败路径确实被执行。 + +**通过捕获更多绑定方法来修复 flush 竞态。** 已否决：这正是已经失败过的做法。绑定一个可调用对象解决的是引用解析，而不是对该可调用对象所读取的可变状态的并发访问。只有对共享账本施加互斥才能消除该竞态。 + +**用大小阈值来保护残余数据（只复制大帧）。** 已否决：该分支在每次包含换行符的读取时运行一次，复制的规模受残余数据自身长度约束（始终是一个不完整行），而阈值会引入一个可调参数和第二条代码路径，却换不来任何可度量的节省。无条件地做大小恰当的复制更简单，且始终正确。 + +**通过 seam 断言残余数据的内存效应。** 已否决：被保留的分配透过 `CodeRunResult` 不可观测，因此黑盒测试无法区分已修复与未修复。转而抽取出 `detachResidual`，把底层存储的不变量变成一个确定性的单元测试。 + +**仅用一个发后不理的 `unref` SIGKILL 定时器来回收同进程组存活者。** 已否决：`unref` 的定时器不会让宿主保持存活，因此一个在宽限窗口内退出的宿主（一次性运行、一个配置子进程）永远不会触发 SIGKILL，存活者被 reparent 给 init，这是同一个"没有子进程存活得比 fiber 更久"的违规换了个形态，而且 `teardown` 的"await 每个子进程退出"的 JSDoc 会变为不实。在一个 ref 的轮询上 await 进程组的消亡，让宿主恰好保持存活足够长以完成回收，在常见的空进程组情形下代价为零。 + +**用 `process.kill(pid, 0)` 抛出 ESRCH 来断言回收。** 已否决：一个被 SIGKILL 的进程会作为僵尸进程滞留，直到它的父进程 `wait()` 它，而在一个 PID 1 不回收孤儿进程的容器里，signal-0 探测会持续成功，因此该断言会在跨环境时误报失败。一个停止推进的心跳文件检测的是"不再执行"，而被回收的进程和僵尸进程都满足这一点。 + +**用一个普通的 `set_result` 完成跨事件循环的 Future 并依赖 GIL。** 已否决：GIL 序列化字节码，但并不使 `asyncio.Future` 跨事件循环安全：从一个并非其事件循环所属的线程完成一个 Future，不会调度它的回调，也不会唤醒该事件循环。在拥有该 Future 的事件循环上调用 `call_soon_threadsafe` 才是有文档记载的机制。 + +**在结算之后让 SIGKILL 定时器继续处于装设状态（早先的同进程组修复）。** 已否决：一个被留待在 leader 被回收后长达 `graceMs` 才触发的 `unref` 定时器，可能 `kill(-pid)` 一个被回收（recycled）的 pgid，击中一个无关的进程组；危险是那次成功执行的 kill，而 `killGroup` 吞掉 ESRCH 无法阻止它。在进程组被确认为空后清除该定时器，把复用窗口收窄到真正存在存活者的情形，此时进程组不为空以供复用。 + +**只用继承而来的硬限制来约束 rlimit。** 已否决：那会静默地抬高一个比请求更严格的继承软限制，放松了该约束本应保持的那种收束。用每一侧各自继承而来的界来约束该侧（随后把 soft 钉在 hard 之下），在 soft 和 hard 两者上都保持配置值与继承值中的最严格者。 + +**按序列化开销对宿主侧的 `capMessage` 兜底做计费，与子进程的 `_cap_message` 相符。** 已否决：这两处上限守护的是不同的东西。`_cap_message` 的输出会作为一个 JSON 字符串再次穿过 fd 3，因此帧上限约束的是它转义后的宽度——那里必须按序列化计费。`capMessage` 的输出直接进入 `CodeRunResult.error.message`，绝不会再次穿过一个受帧上限约束的通道，因此对它所保留内容的诚实度量是模型可见字符串的原始字节长度。一个诚实的子进程已经按序列化开销设过上限，而原始长度 ≤ 序列化开销，因此一条格式良好的消息会原样通过；一条伪造的、控制字符密集的消息可能序列化到其原始长度约 6 倍，但由于它不经过任何受上限约束的通道，按那个被抬高的传输宽度对它计费只会截断一条尺寸合法的诊断，而换不来任何收束上的收益。每一侧的 JSDoc 都记录了这一区分，并指向另一侧。 + +**每来一个 `data` 分片就把散逸的管道输出推入一条条目。** 已否决：`logs` 条目在下游会用 `\n` 拼接，因此一个传输分片边界会变成一个模型可见的换行符——一次被拆散在多次管道读取中的原生写入会带着无端的换行回读。按真正的换行符聚合（原始分片缓冲 + 在 `0x0a` 处切分）与子进程的按行粒度的 `log` 帧相符；账本仍然通过在残余数据将要越过预算时把它准入并截断，来约束一场不含换行符的洪泛。 + +**逐帧强制 fd-3 帧上限（在计数器检查之前先切分）以避免一次批次边缘的误拒。** 已否决：帧上限检查在任何 `Buffer.concat` 之前读取字节计数器，正是为了让一个敌意程序无法迫使宿主内存达到 64 MiB 帧上限的约 2 倍（计数器与那次拼接是所持全部内容的第二份副本）。先切分以对单个帧计费，会在拒绝一个超上限的帧之前就 `Buffer.concat` 它，从而重新引入那种翻倍——正是出于这个原因，有两个回归测试断言了先计数后拼接的顺序。逐帧顺序本会修复的那次批次边缘误拒（一个合法的接近上限的帧，其携带换行符的分片同时也带上了下一帧的起始字节，在一次管道读取中把计数器推过上限）只有当 `maxLogBytes`／`maxValueBytes` 被配置到距 64 MiB 帧上限一次管道读取以内时才可达——比 32／64 KiB 的默认值高出好几个数量级。在任何配置下都抵御敌意输入的内存安全边界，优先于一个仅在病态的接近上限配置下才可达的误拒；计数器的超额计数与这一权衡都记录在该检查处。 + +**当合并预算被越过时，按残余数据到达顺序冲刷两个散逸管道。** 已否决：stdout 与 stderr 是相互独立的 OS 流，它们的 `data` 事件本就彼此之间、以及与子进程自己的 fd-3 `log` 帧之间不确定地交错。seam 的 `CodeRunResult.logs` JSDoc 写着「in order」，周围的文字把它限定为一条流之内的程序发出顺序——跨并发流的顺序在这里本质上是尽力而为，因为没有任何宿主侧的冲刷顺序能够重建内核已经丢失的真实交错，因此在冲刷处保留一条残余数据的到达顺序换不来任何东西。一个固定的排空顺序与任何顺序一样有效。跟踪一个逐残余数据的到达计次以先排空较早的管道，会增加一个分支，它的两侧只在两个 OS 管道的相对时机上才触发，而 `os.sched_yield` 并不使之具有确定性，因此该分支无法在不写一个不稳定测试的情况下被覆盖——有成本却没有可观测的契约收益。 + +**在运行时按地址空间对子进程日志账本计量，而不是在加载期拒绝该配置。** 已否决：对每次子进程写入做一次精确的序列化开销检查，要么是一次完整的 `encode`——正是一次超大写入负担不起、而账本那处廉价预检本就为规避它而存在的那次分配——要么是一个逐字符的 Python 循环，它会烧掉 CPU 预算（一次 10 MB 的合法写入会在 `cpuSeconds: 1` 之下触发 SIGXCPU）。每一种运行时做法都是在热路径上拿内存界限换另一种资源界限。地址空间的突破是 `maxLogBytes`／`addressSpaceMb` 组合的属性，而不是任何一次具体写入的属性，因此在加载期一次性拒绝这个不兼容的组合，能在没有任何逐次写入代价的情况下消除整类问题，并保留 `_LogStream` 原有的按字符计数的缓冲，它一旦预算放进地址空间就是内存安全的。 + +## Consequences + +解释器误配置会在服务发布前失败，每次运行都使用加载期选定的可执行文件，并且子进程只接收 `TMPDIR`，从而消除 macOS 启动噪声而不暴露宿主凭证。私有提供方仍不进入已发布 profile，同时由一个 keyless 真实 Loader 快照固定其源码检出 PTC 组合。 + +seam 的"只 resolve、不 reject"契约在引导写入路径和同步 spawn 失败路径上都得以成立，两者的覆盖率都是被度量的，且两者都不会遗留一个暂存目录。日志捕获是线程安全的，代价是每次写入和 flush 都要获取一次可重入锁，并且散逸的原生输出由它自己的换行符来分隔，而不是由传输分片来分隔。fd-3 残余数据的内存受实际保留的字节数约束，并且两个帧读取器都以一次而非平方级的方式扫描一个不断累积的帧。输出上限放行一个帧所能承载的每一个值，并在加载期拒绝一个非整数的预算。dispose 面对同进程组存活者是真正完全停稳的（以 `graceMs + 2 * CLOSE_REAP_MARGIN_MS` 为界，在进程组已为空时代价为零，并且一旦进程组清空就清除 SIGKILL 定时器，从而一次滞留的 kill 无法击中一个被回收的 pgid），RLIMIT 强制在 soft 和 hard 两者上都保持配置值与继承值中的最严格者（并且 SIGXCPU 诊断不再把一个宿主无法保证的预算说出来），并且从模型创建的线程调用的绑定会完成而不是超时，而且握手帧读取器不再在一个大程序上烧掉 CPU 预算。每处行为修复都附带一个在缺少它时会失败的测试，除了 Problem 一节点出的那十一处——分块读取帧（一处系统调用次数的改进）、确认为空后的收尾（它唯一透过 seam 可观测的效应会在信号投递时冻结，而修复前的代码也会产生同样的结果）、共享的 stdout／stderr 预算（它唯一透过 seam 可观测的差异取决于不确定的跨管道到达时机），`flush_line` 重排序（它降低后的峰值仍落在 12× 门本就放行的范围内，因此没有配置会有不同表现）、节流 binding 回复（32.0 MiB → 0.0 MiB 的峰值降幅位于宿主 fd-3 可写缓冲内部、透过 seam 不可度量），在快照之前丢弃迟到的 binding 解析（它的三条断言在修复前同样成立，因此不是一处缺代码即变红的用例），完成值的 TOCTOU 预编码（它的并发变异竞态无法透过 seam 确定性构造，而 daemon 变异回归的唯一断言是概率性的），stray UTF-8 预算冲刷的扣留（落在多字节边界上的预算冲刷无法透过 seam 调度——v8-ignored），以及结算后到达的迟到拒绝的 settled 先查（结算之后才到达的拒绝无法从 seam 确定性构造），以及日志分片封存（其 25 M 规模的 OOM 无法在 CI 中确定性构造），以及 unknown-binding 预览上限（其完整 target 的 `JSON.stringify` 峰值是回复路径内的一次瞬时分配，无法透过 seam 度量）——因此其余各处未来若发生回归都会变红。 diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-resume-selector-batch-projection.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-07-31-resume-selector-batch-projection.i18n.yaml index a1a69bb1af..a5e97e8e8e 100644 --- a/.agents/notes/implemented/bug-fix/2026-07-31-resume-selector-batch-projection.i18n.yaml +++ b/.agents/notes/implemented/bug-fix/2026-07-31-resume-selector-batch-projection.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-07-31-resume-selector-batch-projection.md -2026-07-31-resume-selector-batch-projection.md: e4809575e03bbd74522b26a8a170ac558d6eee41 -2026-07-31-resume-selector-batch-projection.zh.md: 04646d266c87b96b7c28692663540ffe808d0082 +2026-07-31-resume-selector-batch-projection.md: aa99ecf323b44432e360402f072d89436b2778bd +2026-07-31-resume-selector-batch-projection.zh.md: ebc9d43677edc23004c32736db989d39e59ef152 diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-resume-selector-batch-projection.md b/.agents/notes/implemented/bug-fix/2026-07-31-resume-selector-batch-projection.md index e4809575e0..aa99ecf323 100644 --- a/.agents/notes/implemented/bug-fix/2026-07-31-resume-selector-batch-projection.md +++ b/.agents/notes/implemented/bug-fix/2026-07-31-resume-selector-batch-projection.md @@ -26,7 +26,7 @@ No session-query or session-persistence surface changed. The shipped TUI composi **Fix only the O(N²) listing inside `SessionCorpus.load()`.** Rejected as the primary fix: the per-candidate full decompress, replay validation, and triple clone dominated on large logs. The redundant pre-listing in `load()` remains a candidate cleanup with error-semantics implications. -**Surface a last-modified time through `listSnapshots`/`SessionRecord`.** Cleanest seam-wise, but touches the persistence contract, provider, and query record type for what the TUI can already derive from `locate()` plus one stat. Reintroduce if a second consumer needs metadata activity times. +**Surface a last-modified time through `list()`/`SessionRecord`.** Cleanest seam-wise, but touches the persistence contract, provider, and query record type for what the TUI can already derive from the stored log's file metadata. Reintroduce if a second consumer needs metadata activity times. **A bespoke persisted title index or TUI-local title cache.** Rejected: the session-projection cache already is the owned durable checkpoint system with an invalidation contract (`stateVersion`, identity binding, shrunk-log anchoring); mounting it beats adding a parallel cache. diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-resume-selector-batch-projection.zh.md b/.agents/notes/implemented/bug-fix/2026-07-31-resume-selector-batch-projection.zh.md index 04646d266c..ebc9d43677 100644 --- a/.agents/notes/implemented/bug-fix/2026-07-31-resume-selector-batch-projection.zh.md +++ b/.agents/notes/implemented/bug-fix/2026-07-31-resume-selector-batch-projection.zh.md @@ -26,7 +26,7 @@ session-query 与 session-persistence 的任何接口都未改变。随附的 TU **只修复 `SessionCorpus.load()` 内部的 O(N²) 列表查询。** 作为主要修复被否决：在大日志上，按候选行执行的完整解压、回放验证和三重克隆才是主要开销。`load()` 中的冗余预列表查询仍是一个候选清理项，但涉及错误语义。 -**通过 `listSnapshots`/`SessionRecord` 暴露最后修改时间。** 从 seam 角度最干净，但要触碰持久化约定、provider 和查询记录类型，而 TUI 已能用 `locate()` 加一次 stat 得到同样的信息。若出现第二个需要元数据活动时间的消费方再引入。 +**通过 `list()`/`SessionRecord` 暴露最后修改时间。** 从 seam 角度最干净，但要触碰持久化约定、provider 和查询记录类型，而 TUI 已能从已存日志的文件元数据得到同样的信息。若出现第二个需要元数据活动时间的消费方再引入。 **专门的持久化标题索引或 TUI 本地标题缓存。** 否决：session-projection 缓存本身就是自有的持久 checkpoint 系统，并已带失效约定（`stateVersion`、身份绑定、日志收缩锚定）；挂载它优于再造一套并行缓存。 diff --git a/.agents/notes/implemented/bug-fix/2026-08-04-load-pre-react-loop-sessions.md b/.agents/notes/implemented/bug-fix/2026-08-04-load-pre-react-loop-sessions.md deleted file mode 100644 index e95817ee60..0000000000 --- a/.agents/notes/implemented/bug-fix/2026-08-04-load-pre-react-loop-sessions.md +++ /dev/null @@ -1,40 +0,0 @@ -# Agent Note: Load sessions from the pre-react-loop format - -Status: implemented - -English | [中文](2026-08-04-load-pre-react-loop-sessions.zh.md) - -## Problem - -The react-loop simplification changed durable events while retaining `SESSION_FORMAT_VERSION` 0. Stored sessions from the change's base contain `steering/message` and `turn/start.trigger`; their terminal reasons also use coarse `aborted`, separate `disposed`, and two older error payloads. Current surface and turn invariants cannot replay those records directly. - -The new durable inbox is not part of this compatibility problem. The base emitted process-local inbox notifications but no `agent/inbox/*` session events, so replaying old history as pending work would resurrect already claimed or discarded prompts. - -## Decision - -`PersistenceCoordinator` recognizes the exact pre-react-loop shapes after backend decoding and projects them into the current read view. It removes the obsolete `turn/start.trigger`, converts `steering/message` to the same identified `user/message`, maps old failure facts into the current structured error, folds `disposed` into an aborted turn with the `disposed` cause, and represents coarse aborted records with the persistence-only `{ kind: 'legacy' }` cause because their caller is unavailable. - -The coordinator applies the projection to `load`, `inspect`, adoption, HMR prefix comparison, and `readFrom`. A seek-capable `readFrom` normally reads only its suffix; when that suffix contains a legacy event needing an earlier replacement identity, the coordinator loads and normalizes the complete prefix before returning the requested seq range. - -The importer does not synthesize inbox splices. A resumed pre-react-loop agent begins with empty pending lists, matching the base runtime's inability to persist pending inbox work. The stored artifact remains append-only and later events use the current format. - -## Alternatives considered - -**Treat the same-version records as unsupported.** This follows the pre-release default but strands sessions produced by the PR base even though the removed steering content and terminal facts have complete mappings. - -**Replay old inbox notifications into durable splices.** Those notifications were not session events and do not provide a trustworthy pending-state snapshot. Inferring insertions without every claim and discard would re-run consumed work. - -**Assign coarse aborted records to an existing caller.** Mapping them to `user`, `parent`, or `hook` would invent a caller that the old record did not name. A dedicated `legacy` cause keeps the stop classification without making a false audit claim. - -**Rewrite stored JSONL records.** A rewrite would violate the append-only contract and require atomic migration machinery for a read compatibility boundary. - -## Consequences - -Sessions written in the refactor's base format resume through the current AgentLoop with their steering content, turn boundaries, error facts, and stop classification intact. The shared coordinator contract covers in-memory and JSONL `load`/`inspect`/`readFrom`; an assembled JSONL Agent resume verifies that the historical transcript is visible while both new inbox lists start empty. - -This exception supports the base format, not intermediate formats produced during development of the refactor. In particular, it defines no migration for earlier experimental `agent/inbox/spliced` payloads. Exact-shape recognition keeps malformed current-looking records on their rejection path instead of guessing them into validity. - -## Related - -- [Load sessions persisted before message identity](2026-07-28-load-pre-identity-session-messages.md) — owns deterministic identities and the general read-only import boundary for another same-version format change. -- [Session persistence as an abstract service](../architecture/2026-06-14-session-persistence.md) — owns append-only backend storage and resume. diff --git a/.agents/notes/implemented/bug-fix/2026-08-04-load-pre-react-loop-sessions.zh.md b/.agents/notes/implemented/bug-fix/2026-08-04-load-pre-react-loop-sessions.zh.md deleted file mode 100644 index 98fb1f530f..0000000000 --- a/.agents/notes/implemented/bug-fix/2026-08-04-load-pre-react-loop-sessions.zh.md +++ /dev/null @@ -1,40 +0,0 @@ -# Agent Note: 加载 react-loop 重构前格式的会话 - -Status: implemented - -[English](2026-08-04-load-pre-react-loop-sessions.md) | 中文 - -## 问题 - -react-loop 简化在保持 `SESSION_FORMAT_VERSION` 为 0 的同时更改了持久事件。该变更基线所存储的会话包含 steering（中途引导）事件 `steering/message`，以及 `turn/start.trigger` 字段；其终止原因还使用粗粒度 `aborted`、独立的 `disposed` 和两种旧版错误载荷。当前表层和轮次不变量无法直接回放这些记录。 - -新的持久 inbox 不属于此兼容性问题。该基线会发出进程本地 inbox 通知，但不会产生 `agent/inbox/*` 会话事件，因此将旧历史回放为待处理工作会让已经领取或丢弃的提示词再次执行。 - -## 决策 - -`PersistenceCoordinator` 会在后端解码后识别 react-loop 重构前的确切形状，并将其投影为当前读取视图。它移除已废弃的 `turn/start.trigger`，把 `steering/message` 转换为同一条带标识的 `user/message`，将旧版失败事实映射为当前结构化错误，把 `disposed` 折叠为带 `disposed` 原因的已中止轮次，并用仅供持久化导入使用的 `{ kind: 'legacy' }` 原因表示粗粒度中止记录，因为无法获得其调用方。 - -协调器会把该投影应用于 `load`、`inspect`、接管、HMR（热模块替换）前缀比较和 `readFrom`。可寻址的 `readFrom` 通常只读取后缀；如果后缀包含需要更早替换标识的旧版事件，协调器会先加载并规范化完整前缀，再返回所请求的 seq 范围。 - -导入器不会合成 inbox splice。恢复后的 react-loop 重构前 agent（智能体）从空的待处理列表开始，这与基线运行时无法持久化待处理 inbox 工作的行为一致。已存储产物仍然仅追加，后续事件使用当前格式。 - -## 考虑过的替代方案 - -**将同版本记录视为不受支持。** 这符合预发布阶段的默认立场，但会使 PR（Pull Request）基线产生的会话无法恢复，尽管已移除的 steering 内容和终止事实都有完整映射。 - -**将旧 inbox 通知回放为持久 splice。** 这些通知不是会话事件，也无法提供可信的待处理状态快照。如果无法获知每一次领取和丢弃，就推断插入操作，会让已消费的工作再次执行。 - -**将粗粒度中止记录归因于现有调用方。** 将其映射到 `user`、`parent` 或 `hook` 会凭空指定旧记录未注明的调用方。专用的 `legacy` 原因既能保留停止分类，也不会产生虚假的审计事实。 - -**重写已存储的 JSONL 记录。** 重写会违反仅追加约定，并要求为读取兼容边界建立原子迁移机制。 - -## 后果 - -以重构基线格式写入的会话可以通过当前 AgentLoop 恢复，并完整保留 steering 内容、轮次边界、错误事实和停止分类。共享协调器约定覆盖内存与 JSONL 的 `load`／`inspect`／`readFrom`；组装后的 JSONL agent 恢复用例会验证历史 transcript（文本记录）可见，同时两个新 inbox 列表都从空状态开始。 - -此例外支持基线格式，不支持重构开发期间产生的中间格式。具体而言，它没有为更早的实验性 `agent/inbox/spliced` 载荷定义迁移。通过确切形状识别，当前格式外观相似但结构错误的记录仍会走拒绝路径，不会被猜测性地转换为有效记录。 - -## 相关资料 - -- [加载消息标识机制引入前持久化的会话](2026-07-28-load-pre-identity-session-messages.zh.md)：负责另一项同版本格式变更的确定性标识和通用只读导入边界。 -- [以抽象服务实现会话持久化](../architecture/2026-06-14-session-persistence.zh.md)：负责仅追加后端存储和恢复。 diff --git a/.agents/notes/implemented/bug-fix/2026-08-05-context-meter-blind-to-compaction.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-08-05-context-meter-blind-to-compaction.i18n.yaml index 0bf2cfa13f..abaae859ca 100644 --- a/.agents/notes/implemented/bug-fix/2026-08-05-context-meter-blind-to-compaction.i18n.yaml +++ b/.agents/notes/implemented/bug-fix/2026-08-05-context-meter-blind-to-compaction.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-08-05-context-meter-blind-to-compaction.md -2026-08-05-context-meter-blind-to-compaction.md: daf11b41236943d5fb35e9363c128ce470c7f4e9 -2026-08-05-context-meter-blind-to-compaction.zh.md: 6ea2f8814c4731fdf7c08d8b867cd7086ceccdbf +2026-08-05-context-meter-blind-to-compaction.md: a23e51771c74e350fd3fb04e3a23ab3cdfb6092e +2026-08-05-context-meter-blind-to-compaction.zh.md: 12fd41af0d8c8864a120f26d91148543f74b2e2b diff --git a/.agents/notes/implemented/bug-fix/2026-08-05-context-meter-blind-to-compaction.md b/.agents/notes/implemented/bug-fix/2026-08-05-context-meter-blind-to-compaction.md index daf11b4123..a23e51771c 100644 --- a/.agents/notes/implemented/bug-fix/2026-08-05-context-meter-blind-to-compaction.md +++ b/.agents/notes/implemented/bug-fix/2026-08-05-context-meter-blind-to-compaction.md @@ -27,7 +27,7 @@ This reverses the "the ring, header, and bar length stay provider-exact" half of ## Alternatives considered -**Project `measure().totalTokens` instead.** The measurement service already composes exactly this (`baseline` anchor plus signed `surfaceDeltaTokens`), and it reacts correctly — measured at 4383 → 304 across the same compaction. But it is a service over private replay state, not a pure fold, and a projection cannot call it. Reproducing its anchor inside a `ProjectionDefinition` needs `_estimateProviderAssistant`'s random access to the chunk events cited by seq (`session.events[seq]`), which `apply(state, event)` does not have. Anchoring on the sampled surface total is the same idea reachable from a pure per-event fold. +**Project `measure().totalTokens` instead.** The measurement service already composes exactly this (`baseline` anchor plus signed `surfaceDeltaTokens`), and it reacts correctly — measured at 4383 → 304 across the same compaction. But it is a service over private replay state, not a pure fold, and a projection cannot call it. Reproducing its anchor inside a `ProjectionDefinition` needs `_estimateProviderAssistant`'s random access to the chunk events cited by seq (`session.eventAt(seq)`), which `apply(state, event)` does not have. Anchoring on the sampled surface total is the same idea reachable from a pure per-event fold. **Emit a synthetic usage record at the end of compaction.** Would move `pressureTokens` itself, but the only usage compaction holds is the summarization request's own — a different prompt entirely. Recording it as the conversation's prompt size would be a lie in the durable log rather than in one display. diff --git a/.agents/notes/implemented/bug-fix/2026-08-05-context-meter-blind-to-compaction.zh.md b/.agents/notes/implemented/bug-fix/2026-08-05-context-meter-blind-to-compaction.zh.md index 6ea2f8814c..12fd41af0d 100644 --- a/.agents/notes/implemented/bug-fix/2026-08-05-context-meter-blind-to-compaction.zh.md +++ b/.agents/notes/implemented/bug-fix/2026-08-05-context-meter-blind-to-compaction.zh.md @@ -27,7 +27,7 @@ AFTER compact: ring=4% header=~4227/100000 rows=[system 18, tools 0, messag ## 备选方案 -**改为投影 `measure().totalTokens`。** 测量服务本来就合成了正是这个量（`baseline` 锚点加有符号的 `surfaceDeltaTokens`），而且反应正确——同一次压缩前后实测为 4383 → 304。但它是一个建立在私有重放状态上的服务，不是纯折叠，投影无法调用它。要在 `ProjectionDefinition` 内部复现它的锚点，需要 `_estimateProviderAssistant` 对按 seq 引用的分片事件进行随机访问（`session.events[seq]`），而 `apply(state, event)` 拿不到。以取样时的表层总量作为锚点，是同一个思路在纯逐事件折叠中可达的版本。 +**改为投影 `measure().totalTokens`。** 测量服务本来就合成了正是这个量（`baseline` 锚点加有符号的 `surfaceDeltaTokens`），而且反应正确——同一次压缩前后实测为 4383 → 304。但它是一个建立在私有重放状态上的服务，不是纯折叠，投影无法调用它。要在 `ProjectionDefinition` 内部复现它的锚点，需要 `_estimateProviderAssistant` 对按 seq 引用的分片事件进行随机访问（`session.eventAt(seq)`），而 `apply(state, event)` 拿不到。以取样时的表层总量作为锚点，是同一个思路在纯逐事件折叠中可达的版本。 **在压缩结束时补写一条合成的用量记录。** 这确实能推动 `pressureTokens` 本身，但压缩手上唯一的用量是摘要请求自己的用量——那是完全另一个提示词。把它记成本对话的提示词规模，等于把谎言写进持久日志，而不只是写进某一处展示。 diff --git a/.agents/notes/implemented/bug-fix/2026-08-10-child-agents-join-their-parent-preset.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-08-10-child-agents-join-their-parent-preset.i18n.yaml index 197f0d5f7a..622596ee8c 100644 --- a/.agents/notes/implemented/bug-fix/2026-08-10-child-agents-join-their-parent-preset.i18n.yaml +++ b/.agents/notes/implemented/bug-fix/2026-08-10-child-agents-join-their-parent-preset.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-08-10-child-agents-join-their-parent-preset.md -2026-08-10-child-agents-join-their-parent-preset.md: 5a3d0d1c5e2b492b10a0274204c10e244c37196f -2026-08-10-child-agents-join-their-parent-preset.zh.md: ce5d69404d4f8b6a9879668d2ed0cb1c9d133a40 +2026-08-10-child-agents-join-their-parent-preset.md: 5321d99bad35cf26ebc0e7842dc68c2775bc384b +2026-08-10-child-agents-join-their-parent-preset.zh.md: f0d5b59c3787ef69c2c983da0c8ec0ff4a43e6e7 diff --git a/.agents/notes/implemented/bug-fix/2026-08-10-child-agents-join-their-parent-preset.md b/.agents/notes/implemented/bug-fix/2026-08-10-child-agents-join-their-parent-preset.md index 5a3d0d1c5e..5321d99bad 100644 --- a/.agents/notes/implemented/bug-fix/2026-08-10-child-agents-join-their-parent-preset.md +++ b/.agents/notes/implemented/bug-fix/2026-08-10-child-agents-join-their-parent-preset.md @@ -22,7 +22,7 @@ This is a bind, not a mount, and both differences are load-bearing. The child ge `dsh-subagent` reaches the roster through `ctx.get('agentPresets')` with a type-only import and an optional peer dependency — the documented opportunistic-consumption pattern it already uses for `sandboxPolicy` and `approval`. -Giving the child its parent's tools exposed a second defect the same agent-plane move introduced: `ToolRuntime` exempted SCOPED registrations from a restriction and filtered only the global layer, so once every model-facing row became an ancestor contribution, a child's `toolFilter` stopped constraining anything — and, with the global layer empty, `restrict()` rejected every name it was given as unknown, failing the child outright. The exempt set is the tools a scope registers ITSELF, not the tools that happen to live in the global layer; reading it the second way held only while those two sets coincided. `view()` now filters everything a scope inherits — the global layer and every ancestor layer — and exempts only its own. The own-layer exemption is load-bearing rather than incidental: the delegation runtime registers a child's `report` and structured-output tools into the child's own layer, and a filter naming the capabilities the child may use must not strip the machinery it answers through. +Giving the child its parent's tools exposed a second defect the same agent-plane move introduced: `ToolRuntime` exempted SCOPED registrations from a restriction and filtered only the global layer, so once every model-facing row became an ancestor contribution, a child's `toolFilter` stopped constraining anything — and, with the global layer empty, `restrict()` rejected every name it was given as unknown, failing the child outright. The exempt set is the tools a scope registers ITSELF, not the tools that happen to live in the global layer; reading it the second way held only while those two sets coincided. `view()` now filters everything a scope inherits — the global layer and every ancestor layer — and exempts only its own. The own-layer exemption is load-bearing rather than incidental: the delegation runtime registers a child's structured-output tool into the child's own layer, and a filter naming the capabilities the child may use must not strip the machinery it answers through. ## Alternatives considered @@ -30,11 +30,11 @@ Giving the child its parent's tools exposed a second defect the same agent-plane **Bind the child's key to the PARENT's key rather than to the standing mount.** Rejected because it changes what a child inherits: the parent's own scope layer carries its per-agent restrictions, which would then intersect into every descendant, and a child outliving its parent would hang off a disposed agent's key. Joining the standing mount gives the child its parent's composition and nothing else. -**Extend the continuable activation setup registry to cover one-shot children.** Rejected because that registry's contribution type is synchronous `(childCtx) => () => void` with per-installation revocation, modelling deployment capabilities that come and go, while a preset join is a one-time bind with no revocation of its own. Widening it would have made the omission possible again for any driver that skipped the registry. +**Introduce one shared child-setup registry for both drivers.** Rejected because a synchronous, revocable contribution models deployment capabilities that come and go, while a preset join is a one-time bind with no revocation of its own. Routing composition through an optional registry would make the omission possible again for any driver that skipped it. **Let `dsh-subagent` import `resolveSessionPreset` and mount by the resolved id.** Rejected because it makes the preset roster a hard module edge for a package that must work without one, and it lands back on the remount semantics above. -**Filter every layer on the chain, including the scope's own.** Rejected because it makes a per-child capability filter delete that child's reporting and structured-output tools, which the delegation runtime registers into the child's own layer — an `allow` naming the capabilities a child may use would leave it unable to answer at all. +**Filter every layer on the chain, including the scope's own.** Rejected because it makes a per-child capability filter delete that child's structured-output tool, which the delegation runtime registers into the child's own layer — an `allow` naming the capabilities a child may use would leave it unable to produce the requested result. **Leave the durable header alone and fix only the live join.** Rejected because the live child and the same child read cold would then disagree about which composition produced its history — the same class of defect, moved rather than fixed. diff --git a/.agents/notes/implemented/bug-fix/2026-08-10-child-agents-join-their-parent-preset.zh.md b/.agents/notes/implemented/bug-fix/2026-08-10-child-agents-join-their-parent-preset.zh.md index ce5d69404d..f0d5b59c37 100644 --- a/.agents/notes/implemented/bug-fix/2026-08-10-child-agents-join-their-parent-preset.zh.md +++ b/.agents/notes/implemented/bug-fix/2026-08-10-child-agents-join-their-parent-preset.zh.md @@ -22,7 +22,7 @@ Status: implemented `dsh-subagent` 以类型级导入加可选 peer 依赖的方式，通过 `ctx.get('agentPresets')` 触达 roster——这正是它对 `sandboxPolicy` 与 `approval` 已在使用的、有明确文档的机会性消费模式。 -把父方的工具交给子 agent 之后，暴露出同一次 agent 平面搬迁引入的第二个缺陷：`ToolRuntime` 把**作用域级**注册排除在限制之外、只过滤全局层，因此当所有面向模型的行都变成祖先贡献之后，子 agent 的 `toolFilter` 就不再约束任何东西——而且全局层为空时，`restrict()` 会把收到的每个名字都判为未知并直接让子 agent 创建失败。豁免集合应当是作用域**自己注册**的工具，而不是恰好位于全局层的工具；后一种读法只在这两个集合重合时才成立。`view()` 现在过滤作用域继承来的一切——全局层与每个祖先层——只豁免它自己那层。这条自身层豁免是承重的而非顺带的：委派运行时把子 agent 的 `report` 与结构化输出工具注册进子 agent 自己那层，而一个只点名子 agent 可用能力的过滤器绝不能把它回报所依赖的机制一并剥掉。 +把父方的工具交给子 agent 之后，暴露出同一次 agent 平面搬迁引入的第二个缺陷：`ToolRuntime` 把**作用域级**注册排除在限制之外、只过滤全局层，因此当所有面向模型的行都变成祖先贡献之后，子 agent 的 `toolFilter` 就不再约束任何东西——而且全局层为空时，`restrict()` 会把收到的每个名字都判为未知并直接让子 agent 创建失败。豁免集合应当是作用域**自己注册**的工具，而不是恰好位于全局层的工具；后一种读法只在这两个集合重合时才成立。`view()` 现在过滤作用域继承来的一切——全局层与每个祖先层——只豁免它自己那层。这条自身层豁免是承重的而非顺带的：委派运行时把子 agent 的结构化输出工具注册进子 agent 自己那层，而一个只点名子 agent 可用能力的过滤器绝不能把它产出请求结果所依赖的机制一并剥掉。 ## 考虑过的替代方案 @@ -30,11 +30,11 @@ Status: implemented **把子 agent 的 key 绑到**父方的** key 而不是常驻挂载上。** 否决，因为这改变了子 agent 继承的内容：父方自己的 scope 层携带其逐 agent 限制，那些限制会就此与每个后代求交，而活得比父方久的子 agent 会挂在一个已 dispose 的 agent key 上。加入常驻挂载给到子 agent 的是父方的组装，仅此而已。 -**扩展可继续 activation setup 注册表以覆盖一次性子 agent。** 否决，因为该注册表的贡献类型是同步的 `(childCtx) => () => void` 并带有逐次安装的撤销，建模的是会来会走的部署能力，而 preset 加入是一次性认父、自身没有撤销可言。扩展它反而会让任何绕过该注册表的驱动重新具备遗漏的可能。 +**为两个驱动引入一份共享 child setup 注册表。** 否决，因为同步且可撤销的贡献建模的是会来会走的部署能力，而 preset 加入是一次性认父、自身没有撤销可言。让组合经由可选注册表完成，反而会让任何绕过它的驱动重新具备遗漏的可能。 **让 `dsh-subagent` 导入 `resolveSessionPreset` 并按解析出的 id 挂载。** 否决，因为这会给一个必须在没有 roster 时也能工作的包引入硬模块边，而且最终仍落回上述的重新挂载语义。 -**过滤链上的每一层，包括作用域自身那层。** 否决，因为那会让逐子 agent 的能力过滤器把该子 agent 的回报与结构化输出工具一并删掉——它们由委派运行时注册进子 agent 自己那层——于是一个点名"子 agent 可用哪些能力"的 `allow` 会让它彻底无法回报。 +**过滤链上的每一层，包括作用域自身那层。** 否决，因为那会让逐子 agent 的能力过滤器删掉该子 agent 的结构化输出工具——它由委派运行时注册进子 agent 自己那层——于是一个点名“子 agent 可用哪些能力”的 `allow` 会让它无法产出所请求的结果。 **只修活着的加入，不动持久化 header。** 否决，因为那样活着的子 agent 与冷读同一个子 agent 会对"哪份组装产出了这段历史"给出不同答案——同一类缺陷，只是被搬了个地方而不是被修掉。 diff --git a/.agents/notes/implemented/bug-fix/2026-08-13-bounded-cold-blank-verification.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-08-13-bounded-cold-blank-verification.i18n.yaml index 94a7268a28..51a7388005 100644 --- a/.agents/notes/implemented/bug-fix/2026-08-13-bounded-cold-blank-verification.i18n.yaml +++ b/.agents/notes/implemented/bug-fix/2026-08-13-bounded-cold-blank-verification.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-08-13-bounded-cold-blank-verification.md -2026-08-13-bounded-cold-blank-verification.md: ab2f3b5a0534a98e02a3e2494ca2fff1223efe81 -2026-08-13-bounded-cold-blank-verification.zh.md: 5dc4b62f7ac43ebd3c4a8cef58f5da9af367520a +2026-08-13-bounded-cold-blank-verification.md: 8244b93641cfe576cd2f5b0c618ae69fbb215dd5 +2026-08-13-bounded-cold-blank-verification.zh.md: 2c7f392fa8fbd3fed0641fe2b95e717f76cd78fb diff --git a/.agents/notes/implemented/bug-fix/2026-08-13-bounded-cold-blank-verification.md b/.agents/notes/implemented/bug-fix/2026-08-13-bounded-cold-blank-verification.md index ab2f3b5a05..8244b93641 100644 --- a/.agents/notes/implemented/bug-fix/2026-08-13-bounded-cold-blank-verification.md +++ b/.agents/notes/implemented/bug-fix/2026-08-13-bounded-cold-blank-verification.md @@ -14,15 +14,15 @@ The same cold list used the JSONL artifact mtime for `updatedAt`. Opening a Sess `dsh-api-session-controller` registers `sessionListMetadata`, a projection containing `blank` and `lastPromptAt`. The attached summary folds the same functions directly over the live log. `blank` changes only from true to false on `turn/start`; `lastPromptAt` changes only on a `user/message` whose source kind is `user`. -A cold summary trusts cached `blank: false`, because a checkpoint prefix containing `turn/start` remains non-blank. Cached `blank: true` and a cache miss do not prove the current log is blank. When persistence exposes a physical artifact through `locate()` and its observed size is at most the `coldBlankProbeMaxBytes` eligibility threshold (default 1 KiB per Session), the gateway calls `readFrom(id, 0)` and folds exact list metadata from the stored prefix. Files above the threshold, backends without a location, vanished artifacts, and failed reads all produce `blank: false`, keeping the Session visible. +A cold summary trusts cached `blank: false`, because a checkpoint prefix containing `turn/start` remains non-blank. Cached `blank: true` and a cache miss do not prove the current log is blank and are served `blank: false`, keeping the Session visible. The earlier physical-size probe — a `locate()` path plus a `coldBlankProbeMaxBytes` eligibility threshold gating an exact `readFrom(id, 0)` fold — is removed with the seam's path query ([export and pre-release trims](../simplification/2026-08-27-persistence-export-and-pre-release-trims.md)); persistence snapshot metadata (`eventCount`/`sizeBytes` on `stat()`/`list()`) is the reintroduction path for exact cold verification. -`updatedAt` is the later of `createdAt` and `lastPromptAt`. An eligible artifact read supplies exact `lastPromptAt` at no additional I/O cost; other cache misses or stale checkpoints order the Session too old rather than promoting it from an unrelated file write. After each asynchronous cold read, the gateway checks the live store again and replaces the cold result with an attached summary when another request resumed that Session meanwhile. +`updatedAt` is the later of `createdAt` and `lastPromptAt`. A cache miss or stale checkpoint orders the Session too old rather than promoting it from an unrelated file write. ## Alternatives considered **Trust cached `blank: true`.** Rejected because the projection cache deliberately permits a persisted log to advance beyond its checkpoint. A crash or fail-soft write failure after the first `turn/start` would hide a real conversation and could make the client reuse it as New Session. -**Read every cold log.** Rejected because list latency and I/O would scale with total stored conversation bytes. The physical-size eligibility check targets small historical artifacts that can be checked cheaply and degrades larger unknowns toward visibility. It intentionally does not add a persistence operation solely to make the threshold atomic with the read: concurrent growth may increase one probe's read cost, but the additional events can only preserve visibility or change a blank result to non-blank. +**Read every cold log.** Rejected because list latency and I/O would scale with total stored conversation bytes; unverified cold entries degrade toward visibility instead. **Store blankness and recency in an authoritative persistence index.** Deferred because the shipped JSONL provider has an immutable first line and would require a second durable artifact with ordered updates. An out-of-tree provider may use its own index only with defined update atomicity, versioning, and recovery. The broader exact-index design remains in the [last-activity proposal](../../proposed/architecture/2026-07-29-durable-last-activity-index.md). @@ -30,8 +30,6 @@ A cold summary trusts cached `blank: false`, because a checkpoint prefix contain ## Consequences -Existing small blank JSONL artifacts are hidden without depending on projection-cache availability, and a stale cache cannot hide a stored `turn/start`. A cold list may read each artifact whose observed physical size is within the configured threshold when its cache does not already prove non-blank. The default threshold compares compressed bytes for the shipped Zstandard JSONL backend. +A stale cache cannot hide a stored `turn/start`, and a cold list performs no artifact I/O: cold rows are served from cached projections only. Blank cold Sessions without a cached non-blank projection remain visible, and missing or delayed recency cache entries fall back to `createdAt`. These are conservative degradations: the UI may show an extra empty row or order a Session too low, but it does not hide a conversation or promote one because it was merely opened. -Blank artifacts above the threshold and blank Sessions on location-less backends remain visible. Missing or delayed recency cache entries for artifacts that are not read fall back to `createdAt`. These are conservative degradations: the UI may show an extra empty row or order a Session too low, but it does not hide a conversation or promote one because it was merely opened. - -The gateway-owned projection is an effect of the gateway fiber; unloading the gateway removes the key. Unit coverage pins exact-threshold eligibility, stale-true rejection, monotonic false reuse, exact small-log recency, live-attachment races, fallback direction, human-prompt recency, and fiber disposal. A keyless Web snapshot boots the shipped compressed JSONL composition, seeds a small cold blank artifact without a cache row, and verifies that the sidebar omits it. +The gateway-owned projection is an effect of the gateway fiber; unloading the gateway removes the key. Unit coverage pins stale-true rejection, monotonic false reuse, cache-miss visibility, human-prompt recency, and fiber disposal. diff --git a/.agents/notes/implemented/bug-fix/2026-08-13-bounded-cold-blank-verification.zh.md b/.agents/notes/implemented/bug-fix/2026-08-13-bounded-cold-blank-verification.zh.md index 5dc4b62f7a..2c7f392fa8 100644 --- a/.agents/notes/implemented/bug-fix/2026-08-13-bounded-cold-blank-verification.zh.md +++ b/.agents/notes/implemented/bug-fix/2026-08-13-bounded-cold-blank-verification.zh.md @@ -14,15 +14,15 @@ Web 会话树会隐藏空白 Session，并把当前选中的空白项复用为 N `dsh-api-session-controller` 注册 `sessionListMetadata` 投影，其中包含 `blank` 与 `lastPromptAt`。已附加摘要直接用同一组函数折叠实时日志。`blank` 只在 `turn/start` 时从 true 单调变为 false；`lastPromptAt` 只在来源 kind 为 `user` 的 `user/message` 上更新。 -冷摘要信任缓存的 `blank: false`，因为已包含 `turn/start` 的 checkpoint 前缀会始终保持非空。缓存的 `blank: true` 和 cache miss 都无法证明当前日志为空。当 persistence 通过 `locate()` 暴露物理工件，且其观测大小不超过 `coldBlankProbeMaxBytes` 资格阈值（默认每个 Session 1 KiB）时，网关调用 `readFrom(id, 0)`，从已存前缀折叠精确列表元数据。超过阈值的文件、不提供位置的后端、已消失的工件和读取失败都产生 `blank: false`，让 Session 保持可见。 +冷摘要信任缓存的 `blank: false`，因为已包含 `turn/start` 的 checkpoint 前缀会始终保持非空。缓存的 `blank: true` 和 cache miss 都无法证明当前日志为空，因而按 `blank: false` 提供，让 Session 保持可见。早先的物理大小探测——`locate()` 路径加上门控一次精确 `readFrom(id, 0)` 折叠的 `coldBlankProbeMaxBytes` 资格阈值——随该 seam 的路径查询一并移除（[导出与预发布裁剪](../simplification/2026-08-27-persistence-export-and-pre-release-trims.zh.md)）；persistence 快照元数据（`stat()`/`list()` 上的 `eventCount`/`sizeBytes`）是重新引入精确冷验证的路径。 -`updatedAt` 取 `createdAt` 与 `lastPromptAt` 中较晚者。符合资格的工件读取无需额外 I/O 即可提供精确 `lastPromptAt`；其他 cache miss 或陈旧 checkpoint 只会让 Session 排得偏旧，而不会因无关的文件写入被提升。每次异步冷读取后，网关都会再次检查实时 store；若另一请求期间已恢复该 Session，则用已附加摘要替换冷结果。 +`updatedAt` 取 `createdAt` 与 `lastPromptAt` 中较晚者。cache miss 或陈旧 checkpoint 只会让 Session 排得偏旧，而不会因无关的文件写入被提升。 ## Alternatives considered **信任缓存的 `blank: true`。** 拒绝，因为 projection cache 有意允许持久日志前进到 checkpoint 之后。首个 `turn/start` 之后若发生崩溃或 fail-soft 写入失败，真实对话就会被隐藏，客户端还可能把它复用为 New Session。 -**读取每一份冷日志。** 拒绝，因为列表延迟与 I/O 会随所有已存对话的总字节数增长。物理大小资格检查只针对能够低成本核验的小型历史工件，更大的未知项则向保持可见降级。该检查有意不为“让阈值与读取原子化”单独新增 persistence 操作：并发增长可能增加一次探测的读取成本，但新增事件只会保持可见，或把空白结果改为非空。 +**读取每一份冷日志。** 拒绝，因为列表延迟与 I/O 会随所有已存对话的总字节数增长；未经核验的冷条目转而向保持可见降级。 **把空白状态与最近时间存入权威 persistence index。** 暂缓，因为交付的 JSONL provider 首行不可变，需要增加带有顺序写入要求的第二份持久工件。仓库外 provider 只有定义更新原子性、版本与恢复语义后才可使用自己的索引。更广泛的精确索引设计仍由[最后活动提案](../../proposed/architecture/2026-07-29-durable-last-activity-index.zh.md)负责。 @@ -30,8 +30,6 @@ Web 会话树会隐藏空白 Session，并把当前选中的空白项复用为 N ## Consequences -既有的小型空白 JSONL 工件无需依赖 projection cache 是否存在即可被隐藏，陈旧 cache 也无法隐藏已存的 `turn/start`。对于 cache 尚不能证明非空，且观测物理大小在配置阈值内的每个 Session，冷列表可能读取其工件。对默认交付的 Zstandard JSONL 后端，该阈值比较压缩后的字节数。 +陈旧 cache 无法隐藏已存的 `turn/start`，且冷列表不做任何工件 I/O：冷行只从缓存投影提供。没有缓存非空投影的空白冷 Session 保持可见，缺失或延迟的最近时间 cache 条目回退到 `createdAt`。这些都是保守降级：UI 可能多显示一条空记录，或把 Session 排得偏低，但不会隐藏真实对话，也不会因为单纯打开而把会话提升到前面。 -超过阈值的空白工件，以及来自不提供位置的后端的空白 Session 会保持可见。对于未被读取的工件，缺失或延迟的最近时间 cache 会回退到 `createdAt`。这些都是保守降级：UI 可能多显示一条空记录，或把 Session 排得偏低，但不会隐藏真实对话，也不会因为单纯打开而把会话提升到前面。 - -网关自有投影是网关 fiber 的 effect；卸载网关会移除该 key。单元覆盖固定了临界大小资格、拒绝陈旧 true、复用单调 false、小日志精确最近时间、实时附加竞态、回退方向、真人 prompt 最近时间和 fiber 销毁。无密钥 Web snapshot 会启动发行版的压缩 JSONL 组合，在没有 cache row 的情况下播种一份小型冷空白工件，并验证侧栏不展示它。 +网关自有投影是网关 fiber 的 effect；卸载网关会移除该 key。单元覆盖固定了拒绝陈旧 true、复用单调 false、cache miss 保持可见、真人 prompt 最近时间和 fiber 销毁。 diff --git a/.agents/notes/implemented/bug-fix/2026-08-13-feedback-note-editor-popover.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-08-13-feedback-note-editor-popover.i18n.yaml index c7289d869f..ac3e62bf89 100644 --- a/.agents/notes/implemented/bug-fix/2026-08-13-feedback-note-editor-popover.i18n.yaml +++ b/.agents/notes/implemented/bug-fix/2026-08-13-feedback-note-editor-popover.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-08-13-feedback-note-editor-popover.md -2026-08-13-feedback-note-editor-popover.md: 42c08e39cfb054db689503e23306c5049a97b6cb -2026-08-13-feedback-note-editor-popover.zh.md: 553029f8a42dae42a38e909d716b41e2c6dd252e +2026-08-13-feedback-note-editor-popover.md: 8f51f090cc24292ad02d96fefb1f45bc05df08c9 +2026-08-13-feedback-note-editor-popover.zh.md: e64e95ac6599be2d7c343b4432415f0e99c6ef9b diff --git a/.agents/notes/implemented/bug-fix/2026-08-13-feedback-note-editor-popover.md b/.agents/notes/implemented/bug-fix/2026-08-13-feedback-note-editor-popover.md index 42c08e39cf..8f51f090cc 100644 --- a/.agents/notes/implemented/bug-fix/2026-08-13-feedback-note-editor-popover.md +++ b/.agents/notes/implemented/bug-fix/2026-08-13-feedback-note-editor-popover.md @@ -18,7 +18,7 @@ The note editor does not enter the row's flex layout at all. It is a popover: a **The action strip.** The like/dislike buttons and the note trigger stay in the row, unchanged. The trigger is a plain button (`aria-haspopup="dialog"`, `aria-expanded` while open) that shows "Add a note" before a note exists and the note text afterward. -**The popover.** While open, the panel contains the textarea plus Save and Cancel, and any note-save failure, as `role="dialog"` with a title distinct from the textarea's own label so both are addressable by name. It opens beneath the trigger (4px gap), clamps to 12px from the viewport edges, auto-focuses the textarea, and closes on Escape or an outside pointer-down. Closing returns focus to the trigger only when the panel was really open, never on the initial mount (a freshly rendered rated message must not pull focus into its action row). A rating action during an open editor closes the panel. The four undefined tokens are replaced with the ones the theme actually defines, matching the primitives' precedent: `border-l2` and `bg-layer-1` for the input, `button-primary-fill` with `label-primary-foreground` plus a `button-primary-hover` state for Save; the panel surface reuses the Menu card recipe (`--dsw-specific-menu`, `--dsw-shadow-lv3`, inverted hairline `--dsw-alias-border-inverted`, `border-radius: 12px`). +**The popover.** While open, the panel contains the textarea plus Save and Cancel, and any note-save failure, as `role="dialog"` with a title distinct from the textarea's own label so both are addressable by name. It opens beneath the trigger (4px gap), clamps to 12px from the viewport edges, auto-focuses the textarea, and closes on Escape or an outside pointer-down. Closing returns focus to the trigger only when the panel was really open, never on the initial mount (a freshly rendered rated message must not pull focus into its action row). A rating action during an open editor closes the panel. The four undefined tokens are replaced with the ones the theme actually defines, matching the primitives' precedent: `border-l2` and `bg-layer-1` for the input, `button-primary-fill` with `label-primary-foreground` plus a `button-primary-hover` state for Save; the panel surface reuses the Menu card surface recipe (`--dsw-specific-menu`, the `--dsw-elevation-prominent` shadow with the `--dsw-alias-border-l1` stroke rebind and `border: 0`) at `border-radius: 12px`. **Failure surfaces split by where the human is looking.** A rating or list-load failure shows beside the buttons in the row, legible whether or not the popover is open. A note-save failure shows inside the popover, next to Save/Cancel, and the panel stays open so the draft survives to be corrected. diff --git a/.agents/notes/implemented/bug-fix/2026-08-13-feedback-note-editor-popover.zh.md b/.agents/notes/implemented/bug-fix/2026-08-13-feedback-note-editor-popover.zh.md index 553029f8a4..e64e95ac65 100644 --- a/.agents/notes/implemented/bug-fix/2026-08-13-feedback-note-editor-popover.zh.md +++ b/.agents/notes/implemented/bug-fix/2026-08-13-feedback-note-editor-popover.zh.md @@ -18,7 +18,7 @@ Status: implemented **操作条。** 点赞/点踩按钮与备注触发按钮保持原样留在行内。触发按钮是普通 `button`（`aria-haspopup="dialog"`，打开时 `aria-expanded`），在没有备注时显示「补充说明」，已有备注时显示备注文本。 -**浮层。** 打开时，面板内含 textarea、Save 与 Cancel，以及任何备注保存失败提示，作为 `role="dialog"`，其标题与 textarea 自身的标签不同，以便两者都能按名称寻址。它在触发按钮下方打开（4px 间距），钳制到距视口边缘 12px，自动聚焦 textarea，并在 Escape 或外部 pointer-down 时关闭。关闭时仅当面板确实曾经打开才把焦点还给触发按钮，绝不会在初始挂载时（新渲染出的一条已评分消息不得把焦点拉进其操作条）。编辑器打开时进行评分操作会关闭面板。四个未定义 token 换成主题确实定义的那些，与 primitives 的既有做法一致：输入框用 `border-l2` 与 `bg-layer-1`，Save 用 `button-primary-fill` 配 `label-primary-foreground` 并加 `button-primary-hover` 状态；面板表面复用 Menu 卡片的配方（`--dsw-specific-menu`、`--dsw-shadow-lv3`、反色发丝线 `--dsw-alias-border-inverted`、`border-radius: 12px`）。 +**浮层。** 打开时，面板内含 textarea、Save 与 Cancel，以及任何备注保存失败提示，作为 `role="dialog"`，其标题与 textarea 自身的标签不同，以便两者都能按名称寻址。它在触发按钮下方打开（4px 间距），钳制到距视口边缘 12px，自动聚焦 textarea，并在 Escape 或外部 pointer-down 时关闭。关闭时仅当面板确实曾经打开才把焦点还给触发按钮，绝不会在初始挂载时（新渲染出的一条已评分消息不得把焦点拉进其操作条）。编辑器打开时进行评分操作会关闭面板。四个未定义 token 换成主题确实定义的那些，与 primitives 的既有做法一致：输入框用 `border-l2` 与 `bg-layer-1`，Save 用 `button-primary-fill` 配 `label-primary-foreground` 并加 `button-primary-hover` 状态；面板表面复用 Menu 卡片的表面配方（`--dsw-specific-menu`、`--dsw-elevation-prominent` 投影配 `--dsw-alias-border-l1` 描边重绑与 `border: 0`），圆角取 `border-radius: 12px`。 **失败提示按人的视线所落之处拆分。** 评分或列表加载失败显示在按钮旁的图标行里，无论浮层是否打开都清晰可读。备注保存失败显示在浮层内、Save/Cancel 旁，且面板保持打开，以便草稿留存待修正。 diff --git a/.agents/notes/implemented/bug-fix/2026-08-17-subagent-message-settlement-ordering.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-08-17-subagent-message-settlement-ordering.i18n.yaml new file mode 100644 index 0000000000..c30e2a0d4e --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-17-subagent-message-settlement-ordering.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-08-17-subagent-message-settlement-ordering.md +2026-08-17-subagent-message-settlement-ordering.md: cdc996643c84c5f50a3bd1836e82645660dc8c57 +2026-08-17-subagent-message-settlement-ordering.zh.md: 1143da1560e4969dcc4f6a0c6d5ca18060b56191 diff --git a/.agents/notes/implemented/bug-fix/2026-08-17-subagent-message-settlement-ordering.md b/.agents/notes/implemented/bug-fix/2026-08-17-subagent-message-settlement-ordering.md new file mode 100644 index 0000000000..cdc996643c --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-17-subagent-message-settlement-ordering.md @@ -0,0 +1,44 @@ +# Agent Note: Child Agent messages precede their settlement notices + +Status: implemented + +English | [中文](2026-08-17-subagent-message-settlement-ordering.zh.md) + +## Problem + +A continuable child can send selected content and later produce an unconditional manager-authored settlement notice. If those two messages enter queues with different claim priority, the later settlement notice can reach the parent model before the earlier child message. The first step of a turn claims the complete `next-step` batch before one `next-turn` message, so mixing a FIFO later-turn send with a next-step settlement reverses causal order. [Issue #2600](https://github.com/deepseek-harness/deepseek-harness/issues/2600) records the defect. + +The child instruction says to send a finding whenever it changes what the parent should do next. Deferring that message to a later turn contradicts its scheduling meaning and separates causally ordered messages across queues with different claim priority. + +## Decision + +Every model-authored adjacent-Agent message uses fixed Steer delivery through `SubagentRuntime.sendMessage()`. A running parent reads the child message at its nearest safe step boundary and an idle parent starts a turn. There is no quiet or next-turn model delivery option. + +The continuation manager retains `sendWaking()` and `admitWaking()` around messages delivered to resident continuable parents. Their purpose is waking-send admission accounting: the receiving Activation remains live between synchronous inbox insertion and the microtask that observes the wake. + +### Ordering across parent states + +A running parent receives an accepted child message and the child's later settlement notice in the same `next-step` FIFO. If the parent becomes idle before settlement arrives, it has already claimed the child message; settlement may then open a later turn without reversing observed order. + +During parent maintenance, the child message occupies `next-step` and latches a wake, while settlement may occupy `next-turn` because maintenance reports idle status. The initial claim still takes next-step input before the queued turn. Waking input submitted after cancellation follows the core Agent's cancellation convergence rather than bypassing it. + +### Verification + +The control-tool suite holds a parent inside an active model request, submits child messages, settles the child, and verifies sender identity, Steer admission, FIFO batching, and preservation after settlement. Continuation coverage pins waking admission accounting for a resident continuable parent and keeps the runtime-owned settlement source distinct from `agent-message`. + +The keyless continuable-subagent snapshot uses the shipped fixed delivery. Its child-visible tool schema is the same as the parent's, and the accepted child message precedes the later settlement notice without a scheduling overlay. + +## Alternatives considered + +**Offer quiet delivery.** A quiet message can remain unread after an idle parent parks. It also gives equivalent model-authored messages different liveness semantics and reopens deployment-dependent ordering. + +**Offer next-turn delivery.** A later next-step settlement notice can still overtake it. Preserving message-before-settlement would require a cross-queue ordering barrier, and no current model operation requires later-turn isolation strongly enough to own that mechanism. + +**Move settlement notices to `next-turn`.** Settlement batching uses the next-step queue so several children finishing together cost one parent step instead of one turn each. Moving settlement would increase latency and model work to retain an unnecessary message scheduling mode. + +## Consequences + +- A child message may extend an open parent turn. It never interrupts the active model request or tool execution; the agent loop admits it only at a step boundary. +- Messages accepted together share one next-step batch, preserving FIFO order and limiting turn amplification. +- Model callers cannot choose a delivery mode, so ordering and wake behavior do not vary by deployment or call. +- A child-to-parent send still requires the direct parent to remain live; the service provides no durable parent mailbox. diff --git a/.agents/notes/implemented/bug-fix/2026-08-17-subagent-message-settlement-ordering.zh.md b/.agents/notes/implemented/bug-fix/2026-08-17-subagent-message-settlement-ordering.zh.md new file mode 100644 index 0000000000..1143da1560 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-17-subagent-message-settlement-ordering.zh.md @@ -0,0 +1,44 @@ +# Agent Note: Child Agent 消息先于其结算通知 + +Status: implemented + +[English](2026-08-17-subagent-message-settlement-ordering.md) | 中文 + +## 问题 + +可继续 child 可以发送选中内容，之后还会产生一条由管理器编写且无条件投递的结算通知。如果这两条消息进入领取优先级不同的队列，较晚的结算通知可能先于较早的 child 消息到达 parent 模型。一个轮次的第一个 step 会先领取完整 `next-step` 批次，再领取一条 `next-turn` 消息，因此混用 FIFO 后续轮次发送与 next-step 结算会颠倒因果顺序。[Issue #2600](https://github.com/deepseek-harness/deepseek-harness/issues/2600)记录了该缺陷。 + +child 指令要求在发现会改变 parent 下一步动作时发送该发现。把这条消息推迟到后续轮次既违背其调度含义，也会把具有因果顺序的消息拆到领取优先级不同的队列。 + +## 决策 + +每条模型编写的相邻 Agent 消息都通过 `SubagentRuntime.sendMessage()` 使用固定 Steer 投递。运行中的 parent 在最近安全 step 边界读取 child 消息，空闲 parent 则启动一个轮次。模型没有静默或 next-turn 投递选项。 + +继续执行管理器在投递到驻留可继续 parent 的消息周围保留 `sendWaking()` 与 `admitWaking()`。它们负责唤醒发送准入记账：接收方 Activation 会在同步 inbox 插入与观察到唤醒的微任务之间保持在线。 + +### 不同 parent 状态下的顺序 + +运行中的 parent 在同一条 `next-step` FIFO 中接收已接受的 child 消息与该 child 随后的结算通知。如果 parent 在结算到达前变为空闲，它已经领取 child 消息；结算随后可以开启后续轮次，而不会颠倒观察顺序。 + +parent 处于 maintenance 时，child 消息占用 `next-step` 并锁存一次唤醒，而结算可能因 maintenance 报告空闲状态而占用 `next-turn`。初始领取仍会先取 next-step 输入，再取排队轮次。取消后提交的唤醒输入遵循核心 Agent 的取消收敛，而不会绕过它。 + +### 验证 + +控制工具测试套件让 parent 保持在活跃模型请求中，提交 child 消息、结算 child，并验证 sender 身份、Steer 准入、FIFO 批处理与结算后保留。继续执行覆盖固定驻留可继续 parent 的唤醒准入记账，并让 runtime 所有的结算来源与 `agent-message` 保持不同。 + +无密钥可继续 subagent 快照使用随附的固定投递。其 child 可见工具 schema 与 parent 相同，且已接受的 child 消息先于之后的结算通知，无需调度 overlay。 + +## 考虑过的替代方案 + +**提供静默投递。** 空闲 parent 停驻后可能永远不读取静默消息。它还会让等价的模型编写消息具有不同存活语义，并重新引入依赖部署的顺序。 + +**提供 next-turn 投递。** 较晚的 next-step 结算通知仍可能越过它。保持消息先于结算需要跨队列顺序屏障，而当前没有模型操作对后续轮次隔离的需求强到足以承担该机制。 + +**把结算通知移到 `next-turn`。** 结算批处理使用 next-step 队列，使多个 child 同时结束只消耗 parent 的一个 step，而不是每个 child 一个轮次。移动结算会为了保留不必要的消息调度模式而增加延迟与模型工作。 + +## 后果 + +- child 消息可以延长开放的 parent 轮次。它绝不会中断活跃模型请求或工具执行；agent loop 只在 step 边界接纳它。 +- 一起被接受的消息共享一个 next-step 批次，保持 FIFO 顺序并限制轮次放大。 +- 模型调用方不能选择投递模式，因此顺序与唤醒行为不会随部署或调用而变化。 +- child 到 parent 的发送仍要求直接 parent 保持在线；服务不提供持久 parent mailbox。 diff --git a/.agents/notes/implemented/bug-fix/2026-08-17-subagent-report-settlement-ordering.md b/.agents/notes/implemented/bug-fix/2026-08-17-subagent-report-settlement-ordering.md deleted file mode 100644 index 30dfab5e96..0000000000 --- a/.agents/notes/implemented/bug-fix/2026-08-17-subagent-report-settlement-ordering.md +++ /dev/null @@ -1,44 +0,0 @@ -# Agent Note: Subagent reports precede their settlement notices - -Status: implemented - -English | [中文](2026-08-17-subagent-report-settlement-ordering.zh.md) - -## Problem - -A continuable child can explicitly report selected content and later produce an unconditional manager-authored settlement notice. Report delivery used `Agent.followup()` and entered the parent's `next-turn` queue, while settlement delivery to a running parent used `Agent.steer()` and entered `next-step`. The first step of a turn claims the complete `next-step` batch before one `next-turn` message, so the later settlement notice could reach the model before the earlier report. The assembled report scenario required `reportDelivery: quiet` to avoid that nondeterministic interleaving. [Issue #2600](https://github.com/deepseek-harness/deepseek-harness/issues/2600) records the defect. - -The report tool tells a child to report whenever a finding changes what its parent should do next. Deferring that message to a later turn contradicted the tool's scheduling meaning and separated causally ordered messages across queues with different claim priority. - -## Decision - -`SubagentReportDelivery` is `'quiet' | 'next-step'`, and `next-step` is the default. Next-step delivery calls `parent.steer()`, so a running parent reads the report at its nearest safe step boundary and an idle parent starts a turn. Quiet delivery continues to call `parent.inject()` and enters the same queue without waking an idle parent. - -The continuation manager retains `sendWaking()` and `admitWaking()` around next-step reports delivered to resident continuable parents. Their purpose is waking-send admission accounting, independent of whether the message targets a step or a turn: the receiving Activation remains live between synchronous inbox insertion and the microtask that observes the wake. - -### Ordering across parent states - -A running parent receives an accepted report and the child's later settlement notice in the same `next-step` FIFO. If the parent becomes idle before settlement arrives, it has already claimed the report; settlement may then open a later turn without reversing the observed order. - -During parent maintenance, the report occupies `next-step` and latches a wake, while settlement may occupy `next-turn` because maintenance reports idle status. The initial claim still takes next-step input before the queued turn. Waking input submitted after cancellation is redirected by `Agent.send()` to `next-turn`, so report and settlement follow the core agent's cancellation convergence rather than bypassing it. - -### Verification - -The report package holds a parent inside an active model request, submits a child report, settles that child, and asserts the pending parent batch is ordered `subagent-report`, then `subagent-settled`, with no queued later turn. Separate coverage pins repeated reports as one FIFO next-step batch, idle-parent wakeup, and waking admission accounting for a continuable parent. - -The assembled ACP report scenario uses the shipped default. Its scheduling fence keeps the child behind the parent's delegation turn and holds the parent in maintenance until settlement follows the report. The report latches the wake while the settlement notice queues a turn; when maintenance ends, the parent claims next-step input before next-turn input and observes both notices in causal order without a quiet-delivery overlay. - -## Alternatives considered - -**Keep the `wakeup` name but change its implementation to `steer()`.** The existing public description defined `wakeup` as one later parent turn. Reusing the value for a different inbox target would leave configuration unable to state the behavior it selects. The pre-release configuration instead names `next-step` directly. - -**Expose `quiet | next-step | next-turn`.** A next-turn report still permits a later next-step settlement notice to overtake it. Preserving report-before-settlement would require a cross-queue ordering barrier, and no current deployment requires next-turn isolation strongly enough to own that mechanism. - -**Move settlement notices to `next-turn`.** Settlement batching deliberately uses the next-step queue so several children finishing together cost one parent step instead of one turn each. Moving settlement would increase latency and model work to retain a report scheduling mode with no current consumer. - -## Consequences - -- A report may extend an open parent turn. It never interrupts the active model request or tool execution; the agent loop admits it only at a step boundary. -- Reports accepted together share one next-step batch, preserving FIFO order and reducing the turn amplification of the former one-turn-per-report behavior. -- The `wakeup` configuration value is rejected rather than retained as an alias. This repository has no external pre-release compatibility promise for Cordis configuration. -- `quiet` remains the deployment escape for reports that must not wake a parked parent, with the existing risk that no model reads them until another waking input arrives. diff --git a/.agents/notes/implemented/bug-fix/2026-08-17-subagent-report-settlement-ordering.zh.md b/.agents/notes/implemented/bug-fix/2026-08-17-subagent-report-settlement-ordering.zh.md deleted file mode 100644 index 658eb18e3a..0000000000 --- a/.agents/notes/implemented/bug-fix/2026-08-17-subagent-report-settlement-ordering.zh.md +++ /dev/null @@ -1,44 +0,0 @@ -# Agent Note: Subagent report 先于其结算通知 - -Status: implemented - -[English](2026-08-17-subagent-report-settlement-ordering.md) | 中文 - -## 问题 - -可继续 child 可以显式上报选中内容，之后还会产生一条由管理器撰写且无条件投递的结算通知。报告投递曾使用 `Agent.followup()` 并进入 parent 的 `next-turn` 队列，而面向运行中 parent 的结算投递使用 `Agent.steer()` 并进入 `next-step`。一个轮次的第一个 step 会先领取完整 `next-step` 批次，再领取一条 `next-turn` 消息，因此较晚的结算通知可能先于较早的报告到达模型。整体组装的报告场景必须使用 `reportDelivery: quiet`，才能避开这种不确定交错。[Issue #2600](https://github.com/deepseek-harness/deepseek-harness/issues/2600)记录了该缺陷。 - -report 工具要求 child 在发现会改变 parent 下一步动作的信息时上报。把这条消息推迟到后续轮次，既违背了工具的调度含义，也让具有因果顺序的消息分散到领取优先级不同的队列中。 - -## 决策 - -`SubagentReportDelivery` 为 `'quiet' | 'next-step'`，默认值为 `next-step`。Next-step 投递调用 `parent.steer()`，因此运行中的 parent 会在最近的安全 step 边界读取报告，空闲 parent 则会启动一个轮次。静默投递继续调用 `parent.inject()`，进入同一队列但不唤醒空闲 parent。 - -对于投递到驻留可继续 parent 的 next-step 报告，继续执行管理器会保留外围的 `sendWaking()` 与 `admitWaking()`。它们负责唤醒发送的准入记账，与消息面向 step 还是 turn 无关：接收方 Activation 在同步插入 inbox 与观察该唤醒的微任务之间保持在线。 - -### 不同 parent 状态下的顺序 - -运行中的 parent 会在同一个 `next-step` FIFO 中接收已接受的报告和该 child 稍后的结算通知。若 parent 在结算到达前变为空闲，它已经领取了报告；结算随后可以开启一个更晚的轮次，而不会反转观察顺序。 - -parent 处于 maintenance 时，报告占据 `next-step` 并锁存一次唤醒，而结算可能因为 maintenance 呈现空闲状态而占据 `next-turn`。首次领取仍会先取 next-step 输入，再取排队轮次。取消后提交的唤醒输入会由 `Agent.send()` 重定向到 `next-turn`，因此报告和结算会遵循核心 agent 的取消收敛，而不会绕过它。 - -### 验证 - -report 包把 parent 保持在一个活动模型请求中，提交 child 报告，再让该 child 结算，并断言等待中的 parent 批次按 `subagent-report`、`subagent-settled` 排序，且没有排队的后续轮次。独立覆盖还会固定重复报告形成一个 FIFO next-step 批次、空闲 parent 唤醒，以及可继续 parent 的唤醒准入记账。 - -整体组装的 ACP 报告场景使用随附默认值。调度围栏让 child 等到 parent 的委派轮次之后，并让 parent 保持 maintenance，直至结算跟在报告之后到达。报告会锁存唤醒，结算通知则排入后续轮次；maintenance 结束时，parent 先领取 next-step 输入、再领取 next-turn 输入，因此无需静默投递 overlay 也能按因果顺序观察两条通知。 - -## 备选方案 - -**保留 `wakeup` 名称，但把其实现改为 `steer()`。** 既有公开描述把 `wakeup` 定义为一个后续 parent 轮次。让该值复用于不同的 inbox 目标，会使配置无法准确说明自己选择的行为。预发布配置因此直接使用 `next-step` 名称。 - -**暴露 `quiet | next-step | next-turn`。** Next-turn 报告仍可能被稍后的 next-step 结算通知超越。要保住报告先于结算，需要跨队列顺序屏障；当前没有任何部署对 next-turn 隔离的需求强到足以承担该机制。 - -**把结算通知移到 `next-turn`。** 结算批处理刻意使用 next-step 队列，使多个一起结束的 child 只花费 parent 的一个 step，而不是各自一个轮次。移动结算会增加延迟和模型工作量，只为保留一个没有当前消费方的报告调度模式。 - -## 后果 - -- 报告可能延长已打开的 parent 轮次。它绝不会打断活动模型请求或工具执行；agent loop 只会在 step 边界准入它。 -- 一起接受的报告会共享一个 next-step 批次，保持 FIFO 顺序，并减少原先每份报告各占一个轮次所造成的轮次放大。 -- `wakeup` 配置值会被拒绝，而不是保留为别名。本仓库对预发布 Cordis 配置不作外部兼容承诺。 -- 对于不得唤醒停驻 parent 的报告，`quiet` 仍是部署退路，同时保留既有风险：在另一条唤醒输入到达之前，没有模型会读取这些报告。 diff --git a/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-call-backlog-and-binding-metadata-snapshot.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-call-backlog-and-binding-metadata-snapshot.i18n.yaml new file mode 100644 index 0000000000..cf4c994853 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-call-backlog-and-binding-metadata-snapshot.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-call-backlog-and-binding-metadata-snapshot.md +2026-08-29-code-runtime-python-call-backlog-and-binding-metadata-snapshot.md: afdad301a4853754184b75668d167e71420c2480 +2026-08-29-code-runtime-python-call-backlog-and-binding-metadata-snapshot.zh.md: 48d6e0748979b09053aa51a94788fdd95d997179 diff --git a/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-call-backlog-and-binding-metadata-snapshot.md b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-call-backlog-and-binding-metadata-snapshot.md new file mode 100644 index 0000000000..afdad301a4 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-call-backlog-and-binding-metadata-snapshot.md @@ -0,0 +1,70 @@ +# Agent Note: Bound in-flight binding calls, snapshot binding metadata, compact the reply queue, and meter wide completions with cursors in the CPython backend + +Status: implemented + +English | [中文](2026-08-29-code-runtime-python-call-backlog-and-binding-metadata-snapshot.zh.md) + +## Problem + +A further review round on the CPython subprocess backend (packages/experimental/code-runtime-python) surfaced seven findings on the binding-dispatch, validation, completion-metering, and frame-parse paths. First, the reply-backlog cap counts only RESOLVED calls — `pendingReplies` grows after the binding's `await` resolves — so a child flooding calls against a binding whose promise never settles accumulates one async closure per frame until the wall clock without ever tripping the cap. Second, `validateBindings` reads `errorClass.name`, `errorClass.memberNameProperty`, and `namespace.global` several times and retains the original errorClass object for the boot frame, whose `JSON.stringify` re-reads it after validation: a getter that returns a valid value during validation and then throws or returns a conflicting value at stringify time turns the seam-misuse rejection into a worker-exit, or injects a different name than validation approved. Third, `replyQueue` never shrinks mid-drain: the drain loop clears consumed slots to `undefined` but leaves `length` (and the backing store) growing, so a child that reads replies just fast enough to keep the drain alive but never empty grows the array linearly with cumulative throughput. Fourth, the completion meter `checkDoneValue` pushes every member of an open container onto an explicit work stack, so a wide completion value near the frame cap (millions of members) copies that many references onto the stack — O(width) auxiliary memory on top of the already-parsed value — OOMing the host after the parse succeeded. Fifth, a done frame processed in the SAME data event as more than 1024 call frames settles the run before the post-macrotask call-backlog check runs (which no-ops once settled), so a child could finish successfully while leaving the outstanding closures behind. Sixth, the child's `_dump_string` folds a spelled-out surrogate pair into its astral code point, so two DIFFERENT Python dict keys — `"\ud83d\ude00"` and `"\U0001f600"` — encode to the SAME JSON member and the host's `JSON.parse` silently drops one of them, violating the lossless-JSON promise for completions and binding arguments. Seventh, the load gate bounds the CHILD's build-and-encode under `RLIMIT_AS` but not the HOST's `JSON.parse`: a legitimately configured wide completion near the frame cap (e.g. a 3-million-key dict under a 50 MiB budget) materializes several times its raw bytes in the host's property storage, so a constrained host heap (e.g. `--max-old-space-size=256`) dies with a process-level OOM during the parse — before `checkDoneValue` (which only sees the already-parsed value) could reject it. + +## Decision + +### In-flight binding calls are capped at 1024, checked once per macrotask after the microtasks drain + +`case 'call'` counts the outstanding binding calls before dispatch (`pendingCalls`) and releases the slot in the async body's `finally`, covering the reply-written, resolution-rejected, and settled-drop exits. The data handler schedules ONE post-batch check per macrotask via `setImmediate` (deduped by a flag): it runs after the current macrotask's microtasks, so it sees the TRUE outstanding count — the live count is inflated by the batch's own frames (the finallys have not run yet), and a per-event snapshot is stale when flowing mode fires several `data` events within one macrotask before any microtask drains. When the count passes `MAX_PENDING_REPLIES` — strictly greater, so exactly 1024 outstanding calls are allowed — the run settles as a `worker-exit` with a call-backlog message. The check no-ops once `settled`, so a `done` or `log` frame wins over the cap: a program that returns with binding calls it started but never awaited still completes with its value. The `done` handler independently re-checks the count before accepting the frame, closing the window where a done in the SAME batch as a flood would settle the run before the post-macrotask check could fire. This is a count bound, not a byte bound. + +### Binding metadata is snapshotted into plain values before validation and the boot frame + +`validateBindings` reads `namespace.global`, `errorClass.name`, and `errorClass.memberNameProperty` each exactly once into a plain local, validates the copies, and stores a plain `{ name, memberNameProperty }` object in the bindings map. The boot frame serializes that stored copy, so validation and the boot frame see identical values regardless of getter state; a stateful getter cannot change or throw between the two stages. + +### The reply queue compacts its consumed prefix mid-drain + +`drainReplies` compacts the consumed prefix (`replyQueue.splice(0, head); head = 0`) once `head` reaches `MAX_PENDING_REPLIES`. The splice is O(head) once per bound of consumed frames — amortized O(1) per reply — bounding the backing store to O(backlog + bound) for a drain that never empties. + +### The completion meter walks wide values with one cursor per nesting level + +`checkDoneValue` now holds one cursor per OPEN container (a values iterator for the root and arrays, an entries iterator for objects whose key escapes are metered when the entry is reached), the same shape `hasNonLosslessNumber` and the child's `_check_done_value` already use. The byte budget still bounds the walk: each member is metered as its cursor yields it, and the width lower-bound checks bail an over-budget container before the cursor descends. The auxiliary state is O(depth), not O(width), so a wide completion near the frame cap meters exactly instead of copying millions of references. `encodeJsonPlain` keeps its per-container task stack, which is O(width) but holds only references while the encoded output is itself O(total bytes) — same-order as its result, so the exemption is documented in its comment. + +### The frame parse cap is bounded by the host's heap + +The raw-byte frame cap does not protect the host process: `JSON.parse` of a wide-object frame materializes several times the raw bytes in property storage. The WORST shape is a dict of many short unique keys, which forces V8's dictionary-mode property storage plus one interned string per key — measured 6.4x for a 3,000,000-key frame (~31 MB raw) on a 1 GiB heap, trending up with key count (a flat unique-key array is ~4x, a repeated-key dict ~3x); a 256 MiB heap OOMs on that frame outright. The effective cap each instance enforces is `min(protocol cap, floor((heap_size_limit - HOST_PARSE_BASELINE_BYTES) / HOST_PARSE_WORST_CASE_MULTIPLE))` with a 16x multiple — ~2.5x over the measured worst shape — derived from the host's configured heap limit (`--max-old-space-size` honored via `v8.getHeapStatistics().heap_size_limit`). A default Node heap (~4 GiB) never binds; a constrained host lowers the cap and the load gate rejects any budget whose frame could not be parsed safely, failing loud at load instead of OOMing the host mid-parse. The child's `RLIMIT_AS` gate is a separate resource and stays unchanged. + +### Dict keys that fold to one JSON member are rejected as non-lossless + +The child's `_dump_string` folds a spelled-out surrogate pair into its astral code point so the host's UTF-16 strings (where the two code units and the single character are the SAME string) meter at the same cost. Python can hold both spellings as distinct keys, so a dict containing `"\ud83d\ude00"` and `"\U0001f600"` would emit two members with the same JSON key and the host's `JSON.parse` would silently drop one. Both lossless-JSON walks (`_lossless_json_violation` for binding arguments, `_check_done_value` for completions) now track each dict's combined keys in a per-dict seen-set — O(keys), the same order as the dict itself — and reject a collision as non-lossless before any encoding. + +## Testing + +- `tests/runtime.spec.ts` — a hostile child floods 5000 sequential calls against a binding that never settles (`await new Promise(() => {})`); the run settles as `worker-exit` with the call-backlog message long before `maxWallMs`. Verified fail-before: without the cap the run times out at the wall clock. +- `tests/runtime.spec.ts` — a legitimate `asyncio.gather` of 1025 instant calls completes with all 1025 results: the post-macrotask check sees the count after the finallys drained, where a per-frame check could trip on the 1025th frame of a single 64 KiB read. +- `tests/runtime.spec.ts` — a program that schedules 1024 slow bindings (still pending) and returns `"done"` completes with its value: the check no-ops once the done frame settles the run, and the strict threshold allows exactly 1024 outstanding calls. Verified fail-before: an unconditional event-boundary check failed this exact case. +- `tests/runtime.spec.ts` — a single 62 KiB write of 1025 compact calls against a never-settling binding settles as `worker-exit` long before `maxWallMs`, even though no further frames ever arrive: the per-macrotask check fires after the batch. Verified fail-before: a per-event admission snapshot never re-checks without further frames and the run waited out the wall clock. +- `tests/runtime.spec.ts` — a single write of 1025 compact calls PLUS a done frame in the same batch settles as `worker-exit`: the done handler re-checks the count before accepting the frame, where the post-macrotask check would no-op after the done settled the run. Verified fail-before: without the done re-check the run completed successfully with the outstanding closures left behind. +- `tests/runtime.spec.ts` — a burst of 1300 instant calls whose frames split across pipe reads completes with all results: the check runs after all of a macrotask's finallys, where a per-event snapshot could see a stale in-flight count when flowing mode fires several events before any microtask drains. +- `tests/runtime.spec.ts` — a completion value and binding arguments whose dict contains both `"\ud83d\ude00"` and `"\U0001f600"` as keys are rejected as non-lossless (invalid-output / a lossless-JSON call rejection): the two spellings fold to one JSON member, which the host's JSON.parse would silently collapse. Verified fail-before: without the collision check both round-tripped with one key dropped. +- `tests/protocol.spec.ts` — `hostFrameParseCeiling` derives the effective parse cap from a simulated heap: the protocol cap binds on a default heap, a ~304 MiB host limit yields a 15 MiB cap, and a tiny heap leaves almost no parse room. +- `tests/runtime.spec.ts` — a child node with a 128 MiB old space rejects a 50 MiB completion budget at load (`maxValueBytes must not exceed`), where the address-space gate alone would admit it. Verified fail-before: with the heap bound ignored the budget loaded. +- `tests/runtime.spec.ts` — a child node with a 128 MiB old space builds a wide-unique-key dict whose frame is AT the derived cap and parses it, surviving. Verified fail-before: with the parse multiple at 8 the derived cap doubles and the same subprocess OOMs during the parse. +- The suite's temp fixtures (`dsh-bad-bin-`, `dsh-fake-bin-`, `dsh-rlimit-*`, `dsh-staging-`, heartbeat dirs, wrapper scripts) are now registered and removed after each test, so repeated runs do not accumulate `dsh-*` artifacts in the shared tmpdir. +- Two namespace-shape tests — `errorClass.name`/`errorClass.memberNameProperty` and `namespace.global` exposed through getters that throw or change on a second read; the run boots and completes, and each field is read exactly once (asserted). Verified fail-before: without the snapshot, the errorClass getter threw inside validation and the global getter injected a different name, failing the program with `NameError`. +- `tests/runtime.spec.ts` — a child floods calls whose replies exceed the writable high-water mark, blocking the first drain write; the resumed drain consumes a backlog past the compaction bound while a second wave of calls is still pending, and the child reads fd 3 itself (blocking the reply pump) to verify all 1524 replies arrive. No fixed sleep: the child's reads pace at the drain's delivery rate, and the host finishes pushing a wave within milliseconds, so the queue is always full at the splice; newlines are counted per chunk (each reply carries exactly one), never by re-scanning the accumulated total, which would be O(n²). Verified fail-before: a splice that removed pending frames dropped the second wave and the run hung to the wall clock. +- `tests/protocol.spec.ts` — a 2,000,000-element array and a 100,000-key object meter at their exact serialized size, reject one byte under, and still find a `-0` tail element, pinning the cursor walk's breadth behavior. + +## Alternatives considered + +**Pause the fd-3 read side instead of counting in-flight calls.** Rejected: pausing reads would also stall processing of `done` and `log` frames the child may send after its last call, changing settlement timing; a count cap is deterministic and matches the existing frame-cap pattern. + +**Check the in-flight count per frame.** Rejected: the finallys run on the microtask queue, which drains only when the macrotask ends, so a single event carrying more than `MAX_PENDING_REPLIES` legitimate call frames would trip a per-frame check even though every binding settled immediately. + +**Check the count at call admission against a per-event snapshot.** Rejected twice: an unconditional event-boundary check reclassifies a `done` frame as worker-exit when a program returns with calls it never awaited, and a snapshot that refreshes per `data` event is stale when flowing mode fires several events within one macrotask (a legitimate burst whose second chunk carries more in-flight calls than the cap would be killed). Checking the true count once per macrotask, after the microtasks drain, is chunking-independent on both axes, and the strict threshold lets exactly `MAX_PENDING_REPLIES` outstanding calls complete normally. + +**Read metadata once but keep the original errorClass object.** Rejected: the boot frame's `JSON.stringify` re-invokes the getters; only a plain stored copy guarantees both stages read the same values. + +**Rely on the drain's `finally` reset for queue memory.** Rejected: the reset runs only when the drain ends; a drain that never empties keeps growing. Mid-drain compaction bounds the backing store while the drain is alive. + +**Keep the completion meter's explicit member stack.** Rejected: the byte budget bounds the WALK but not the stack's reference count, which is O(width) — a wide value near the frame cap copies millions of references and can OOM the host after the parse succeeded; the per-level cursor shape keeps O(depth) state. + +## Consequences + +In-flight binding closures are bounded like the reply backlog, so a child flooding calls against a never-settling binding fails the run early instead of accumulating closures until the wall clock, while a legitimate large concurrent gather is unaffected (the cap is checked after the microtask queue drains). The boot frame serializes exactly the metadata validation approved, regardless of getter state. The reply queue's backing store stays bounded during sustained partial drains; the compaction is internal memory hygiene with no observable behavior change. The completion meter keeps its exact byte accounting with O(depth) auxiliary state, so a wide completion value meters without a host OOM. diff --git a/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-call-backlog-and-binding-metadata-snapshot.zh.md b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-call-backlog-and-binding-metadata-snapshot.zh.md new file mode 100644 index 0000000000..48d6e07489 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-call-backlog-and-binding-metadata-snapshot.zh.md @@ -0,0 +1,70 @@ +# Agent Note: 在 CPython 后端限制在途 binding 调用、快照 binding 元数据、压缩回复队列并用游标计量宽完成值 + +Status: implemented + +[English](2026-08-29-code-runtime-python-call-backlog-and-binding-metadata-snapshot.md) | 中文 + +## Problem + +对 CPython 子进程后端（packages/experimental/code-runtime-python）的又一轮评审在 binding 分发、校验、完成值计量与帧解析路径上浮出七项发现。其一，回复积压上限只计数已解析的调用——`pendingReplies` 在 binding 的 `await` 解析后才增长——因此向 promise 永不结算的 binding 洪泛调用的子进程会每个帧累积一个异步闭包直到墙钟，却始终不触发该上限。其二，`validateBindings` 多次读取 `errorClass.name`、`errorClass.memberNameProperty` 与 `namespace.global`，并把原始 errorClass 对象保留到引导帧，其 `JSON.stringify` 在校验后重读该对象：getter 在校验时返回合法值、在序列化时抛错或返回冲突值，会把 seam 误用拒绝变成 worker-exit，或注入一个未经校验批准的名字。其三，`replyQueue` 在排空进行中从不收缩：排空循环把已消费槽位清成 `undefined`，但 `length`（及其后备存储）继续增长，因此以恰好能让排空持续存活却永不排空的速率读取回复的子进程，会让数组随累计吞吐量线性增长。其四，完成值计量器 `checkDoneValue` 把开放容器的每个成员压入显式工作栈，因此接近帧上限的宽完成值（数百万成员）会把同等数量的引用复制进栈——在已解析值之上再占 O(width) 辅助内存——在解析成功后 OOM 终止宿主。其五，与超过 1024 个调用帧处于同一 data 事件的 done 帧会在批后调用积压检查运行之前结算运行（该检查在 settled 后为空操作），因此子进程可以「成功」完成却留下未结算闭包。其六，子端 `_dump_string` 把拼写出来的代理项对折叠成星面码点，因此两个不同的 Python 字典键——`"\ud83d\ude00"` 与 `"\U0001f600"`——编码成同一个 JSON 成员，宿主的 `JSON.parse` 会静默丢弃其一，违反完成值与 binding 参数的 lossless JSON 承诺。其七，加载门限制的是子端在 `RLIMIT_AS` 下的构建与编码，而非宿主的 `JSON.parse`：合法配置的接近帧上限宽完成值（如 50 MiB 预算下的 300 万键字典）会在宿主属性存储中物化其原始字节的若干倍，因此受限堆宿主（如 `--max-old-space-size=256`）会在解析期间以进程级 OOM 死亡——早于 `checkDoneValue`（它只能看到已解析值）拒绝它。 + +## Decision + +### 在途 binding 调用限制为 1024，每个宏任务在微任务排空后检查一次 + +`case 'call'` 在分发前对在途 binding 调用计数（`pendingCalls`），并在异步体的 `finally` 中释放槽位，覆盖回复已写入、解析被拒绝与结算后丢弃三种出口。data 处理器经 `setImmediate`（用标志去重）为每个宏任务调度一次批后检查：它在该宏任务的微任务之后运行，因此看到的是真实在途计数——实时计数被本批自身帧抬高（`finally` 尚未运行），而按 `data` 事件刷新的快照在 flowing 模式下同一宏任务内连续发出多个事件（微任务尚未排空）时会过期。计数**严格大于** `MAX_PENDING_REPLIES`（恰好 1024 个在途调用允许）时，运行以带 call-backlog 消息的 `worker-exit` 结算。检查在 `settled` 后为空操作，因此 `done` 或 `log` 帧优先于上限：启动了 binding 调用却未 await 就返回的程序仍以其值正常完成。`done` 处理器在接受帧之前独立复查计数，封住与洪泛同批的 done 会在批后检查触发前结算运行的窗口。这是计数上限而非字节上限。 + +### binding 元数据在校验与引导帧之前快照为纯值 + +`validateBindings` 把 `namespace.global`、`errorClass.name` 与 `errorClass.memberNameProperty` 各恰好读取一次到普通局部变量，对副本做校验，并在 bindings 映射中存入普通 `{ name, memberNameProperty }` 对象。引导帧序列化该存储副本，因此无论 getter 处于何种状态，校验与引导帧看到的都是相同的值；有状态的 getter 无法在两个阶段之间改变或抛错。 + +### 回复队列在排空进行中压缩已消费前缀 + +`drainReplies` 在 `head` 达到 `MAX_PENDING_REPLIES` 时压缩已消费前缀（`replyQueue.splice(0, head); head = 0`）。该 splice 为 O(head)，每消费一上限的帧执行一次——均摊到每条回复为 O(1)——使永不排空的排空把后备存储限制在 O(积压 + 上限)。 + +### 完成值计量器每层持一个游标遍历宽值 + +`checkDoneValue` 现在为每个开放容器持一个游标（根与数组用 values 迭代器，对象用 entries 迭代器——key 的转义字节在该 entry 到达时计量），与 `hasNonLosslessNumber` 及子端 `_check_done_value` 已用的形态一致。字节预算仍然限制遍历：每个成员在游标产出时计量，宽度下界检查会在游标下降之前拒绝超预算容器。辅助状态为 O(depth) 而非 O(width)，因此接近帧上限的宽完成值精确计量，而不是复制数百万引用。`encodeJsonPlain` 保留其每容器任务栈——该栈为 O(width) 但只持有引用，而编码输出本身即 O(total bytes)，与结果同量级，豁免已在注释中说明。 + +### 帧解析上限受宿主堆约束 + +原始字节帧上限并不保护宿主进程：`JSON.parse` 一个宽对象帧会在属性存储中物化其原始字节的若干倍。**最坏形态是大量短唯一键的字典**——迫使 V8 进入字典模式属性存储并为每个键内化一个字符串——1 GiB 堆上 3,000,000 键帧（约 31 MB 原始）实测 6.4 倍且随键数上升（平铺唯一键数组约 4 倍、重复键字典约 3 倍）；256 MiB 堆直接在该帧上 OOM。每个实例执行的有效上限为 `min(协议上限, floor((heap_size_limit - HOST_PARSE_BASELINE_BYTES) / HOST_PARSE_WORST_CASE_MULTIPLE))`，系数为 16——实测最坏形态的约 2.5 倍安全余量——由宿主配置的堆上限推导（`--max-old-space-size` 经 `v8.getHeapStatistics().heap_size_limit` 生效）。默认 Node 堆（约 4 GiB）永不收紧；受限宿主会降低上限，加载门拒绝任何帧无法被安全解析的预算，在加载期响亮失败而非在解析中途 OOM 宿主。子端的 `RLIMIT_AS` 门是另一资源，保持不变。 + +### 折叠为同一 JSON 成员的字典键按非 lossless 拒绝 + +子端 `_dump_string` 把拼写出来的代理项对折叠成星面码点，使宿主的 UTF-16 字符串（两个码元与单个字符是同一字符串）按相同成本计量。Python 可以把两种拼写作为不同键持有，因此包含 `"\ud83d\ude00"` 与 `"\U0001f600"` 的字典会发出两个同键成员，宿主的 `JSON.parse` 会静默丢弃其一。两条 lossless-JSON 遍历（binding 参数的 `_lossless_json_violation` 与完成值的 `_check_done_value`）现在用每字典 seen 集跟踪合并后的键——O(keys)，与字典本身同量级——在编码前把冲突判为非 lossless。 + +## Testing + +- `tests/runtime.spec.ts`——敌意子进程向永不结算的 binding（`await new Promise(() => {})`）洪泛 5000 个连续调用；运行在远早于 `maxWallMs` 时以带 call-backlog 消息的 `worker-exit` 结算。已实测失败前置：没有该上限时运行在墙钟处超时。 +- `tests/runtime.spec.ts`——合法的 `asyncio.gather` 并发 1025 个即时调用并全部完成：批后检查看到的是 `finally` 排空后的计数，而逐帧检查可能被单次 64 KiB 读取中的第 1025 帧误触发。 +- `tests/runtime.spec.ts`——程序调度 1024 个慢 binding（仍未结算）并返回 `"done"` 时以其值正常完成：done 帧结算运行后检查为空操作，且严格阈值允许恰好 1024 个在途调用。已实测失败前置：无条件的事件边界检查恰好在该用例上失败。 +- `tests/runtime.spec.ts`——单次 62 KiB 写入的 1025 个紧凑调用对抗永不结算的 binding，在远早于 `maxWallMs` 时以 `worker-exit` 结算，即使之后不再有帧到达：每宏任务检查在该批之后触发。已实测失败前置：按事件刷新的接纳快照在没有后续帧时永不复查，运行等到墙钟。 +- `tests/runtime.spec.ts`——单次写入的 1025 个紧凑调用**外加同批 done 帧**以 `worker-exit` 结算：done 处理器在接受帧之前复查计数，而批后检查会在 done 结算运行后空操作。已实测失败前置：没有 done 复查时运行成功完成并留下未结算闭包。 +- `tests/runtime.spec.ts`——1300 个即时调用的突发（帧跨管道读取拆分）全部完成：检查在该宏任务的所有 `finally` 之后运行，而按事件快照在 flowing 模式同宏任务内多个事件、微任务未排空时会看到过期的在途计数。 +- `tests/runtime.spec.ts`——字典同时含 `"\ud83d\ude00"` 与 `"\U0001f600"` 两个键的完成值与 binding 参数按非 lossless 拒绝（invalid-output / lossless-JSON 调用拒绝）：两种拼写折叠为一个 JSON 成员，宿主的 JSON.parse 会静默折叠。已实测失败前置：没有碰撞检查时两者都以丢键 round-trip。 +- `tests/protocol.spec.ts`——`hostFrameParseCeiling` 从模拟堆推导有效解析上限：默认堆上协议上限约束，约 304 MiB 宿主上限得出 15 MiB 上限，极小堆几乎不留下解析空间。 +- `tests/runtime.spec.ts`——128 MiB old space 的子 node 在加载期拒绝 50 MiB 完成值预算（`maxValueBytes must not exceed`），而地址空间门单独会放行。已实测失败前置：忽略堆上限时该预算正常加载。 +- `tests/runtime.spec.ts`——128 MiB old space 的子 node 构造帧恰在推导上限处的宽唯一键字典并解析，存活。已实测失败前置：解析系数回退到 8 时推导上限翻倍，同一子进程在解析中 OOM。 +- 套件的临时 fixture（`dsh-bad-bin-`、`dsh-fake-bin-`、`dsh-rlimit-*`、`dsh-staging-`、heartbeat 目录、wrapper 脚本）现登记并在每个测试后移除，重复运行不再在共享 tmpdir 累积 `dsh-*` 工件。 +- 两个 namespace 形态测试——`errorClass.name`/`errorClass.memberNameProperty` 与 `namespace.global` 经由第二次读取即抛错或改变的 getter 暴露；运行正常引导并完成，且每个字段恰好读取一次（已断言）。已实测失败前置：没有快照时，errorClass getter 在校验内抛错，global getter 注入不同名字，程序以 `NameError` 失败。 +- `tests/runtime.spec.ts`——子进程洪泛回复超过可写高水位线的调用，阻塞第一次排空写入；恢复的排空在第二波调用仍待发时消费超过压缩上限的积压，子进程直接读取 fd 3（阻塞回复泵）验证全部 1524 条回复送达。无固定睡眠：子进程的读取以排空的投递速率节流，宿主在毫秒内完成一波推送，因此压缩点队列必然已满；换行按块计数（每条回复恰好一个），绝不重扫累计总量——那会是 O(n²)。已实测失败前置：移除待发帧的 splice 会丢掉第二波回复，运行挂到墙钟。 +- `tests/protocol.spec.ts`——2,000,000 元素数组与 100,000 键对象以精确序列化大小计量、少一个字节即拒绝，并仍能发现尾部的 `-0`，钉住游标遍历的广度行为。 + +## Alternatives considered + +**暂停 fd-3 读侧而非计数在途调用。** 拒绝：暂停读取也会让子进程在最后一个调用后可能发送的 `done` 与 `log` 帧处理停滞，改变结算时机；计数上限是确定性的，且与既有帧上限模式一致。 + +**逐帧检查在途计数。** 拒绝：`finally` 在微任务队列上运行，微任务只在宏任务结束时排空，因此单个事件携带超过 `MAX_PENDING_REPLIES` 个合法调用帧时，即使每个 binding 都立即结算，逐帧检查也会误触发。 + +**在调用接纳处对照按事件刷新的快照检查。** 两次拒绝：无条件的事件边界检查会把程序返回未 await 调用时的 `done` 帧改判为 worker-exit；按 `data` 事件刷新的快照在 flowing 模式同一宏任务内多个事件时过期（第二块携带超过上限的在途调用的合法突发会被误杀）。每宏任务在微任务排空后检查真实计数，在两个轴上都不依赖分块；严格阈值让恰好 `MAX_PENDING_REPLIES` 个在途调用正常完成。 + +**只读取一次元数据但保留原始 errorClass 对象。** 拒绝：引导帧的 `JSON.stringify` 会重新调用 getter；只有存入普通副本才能保证两个阶段读到相同的值。 + +**依赖排空的 `finally` 重置来回收队列内存。** 拒绝：重置只在排空结束时运行；永不排空的排空会持续增长。排空进行中的压缩在排空存活期间限制后备存储。 + +**保留完成值计量器的显式成员栈。** 拒绝：字节预算限制遍历本身，但不限制栈的引用数——那是 O(width)——接近帧上限的宽值会复制数百万引用，在解析成功后 OOM 宿主；每层游标形态保持 O(depth) 状态。 + +## Consequences + +在途 binding 闭包与回复积压一样受限，向永不结算的 binding 洪泛调用的子进程会让运行提前失败，而不是把闭包累积到墙钟；合法的并发大 gather 不受影响（上限在微任务队列排空后检查）。引导帧序列化校验批准的元数据，与 getter 状态无关。回复队列的后备存储在持续的部分排空期间保持有界；压缩是内部内存卫生，无可观察的行为变化。完成值计量器以 O(depth) 辅助状态保持精确的字节核算，宽完成值不再因计量本身 OOM 宿主。 diff --git a/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-load-and-dispatch-hardening.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-load-and-dispatch-hardening.i18n.yaml new file mode 100644 index 0000000000..cc95841077 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-load-and-dispatch-hardening.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-load-and-dispatch-hardening.md +2026-08-29-code-runtime-python-load-and-dispatch-hardening.md: 3d64f96420cd337fc8c7bb44e02f912e1868deed +2026-08-29-code-runtime-python-load-and-dispatch-hardening.zh.md: 64772eb14a09585b1ee0ab10ffcf1298b77b0a35 diff --git a/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-load-and-dispatch-hardening.md b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-load-and-dispatch-hardening.md new file mode 100644 index 0000000000..3d64f96420 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-load-and-dispatch-hardening.md @@ -0,0 +1,41 @@ +# Agent Note: Load-time pythonBin validation, binding snapshot, and reply-drain settle in the CPython backend + +Status: implemented + +English | [中文](2026-08-29-code-runtime-python-load-and-dispatch-hardening.zh.md) + +## Problem + +Review of the CPython subprocess backend (packages/experimental/code-runtime-python) surfaced four non-blocking findings that a long-running host could still misbehave under: an explicit `pythonBin` path bypassed the load-time configuration checks, a throwing binding member accessor could escape the fd-3 data callback and terminate the host, the reply drain could hang forever waiting for a `drain` event that a destroyed pipe never emits, and two leak assertions diffed a global tmpdir in a way a parallel vitest worker could false-positive on. + +## Decision + +### An explicit pythonBin must be an executable regular file at load + +`resolvePythonBin` returned an absolute or slash-containing `pythonBin` verbatim, so a missing, non-executable, or directory path passed the constructor's load checks (which only rejected empty/NUL values and unresolvable basenames) and surfaced only at the first `run()` as a misleading `worker-exit`. The explicit-path branch now validates with the same `accessSync(X_OK)` + `statSync().isFile()` checks the PATH branch uses (a directory passes `X_OK`, so the regular-file requirement is the deciding half), resolving relative explicit paths against the host CWD first — the same place `spawn` would have looked. A failing explicit path makes `resolvePythonBin` return `undefined`, and the load check now distinguishes the two failure classes in its message: `is not an executable regular file` for an explicit path, `does not resolve on PATH` for a basename. + +### Binding callables are snapshotted during validation + +`namespace.functions` is caller-supplied, so its members may be exposed through getters or a Proxy. Reading one of them inside the fd-3 `data` callback — `record[message.name]` — threw OUTSIDE the dispatcher's try and terminated the host (an `uncaughtException` handler, if installed, would only let the run degrade to the wall clock). `validateBindings` now reads every member into a plain own-property record during run()'s synchronous validation segment, so a throwing accessor becomes the seam-misuse rejection run() already reserves for malformed bindings. The snapshot is also the single key set the boot frame advertises AND dispatch reads, so a getter whose keys differ between reads cannot desynchronize the child's allowed names from what the host will actually call. The record is null-prototype (`Object.create(null)`): the seam contract treats member names like `__proto__` or `constructor` as ordinary own properties, and a plain `{}` assignment of `__proto__` hits the prototype setter instead of creating the own property, dropping the name from the boot frame and making a call to it fail with `KeyError`. + +### The reply drain settles on a destroyed pipe + +`drainReplies` awaited `once(proto, 'drain')` after a full-buffer write; a pipe destroyed under the wait (child exited, close-deadline teardown) never emits `drain` again, and `events.once` rejects only on `error`, not on `close` — the await could hang forever, leaving `draining` true and the unconsumed queue (and any wide payloads it still holds) pinned with the closure. The wait now listens for `drain`, `close`, and `error` together, removing all three listeners whichever wins, and the drain loop short-circuits on `proto.destroyed` before the next write, so the `finally` clears the queue and resets `draining`. + +## Testing + +- `tests/runtime.spec.ts` — the load-rejection cases cover a missing absolute path, a non-executable regular file, a directory, and a slash-containing relative path, each asserting the `is not an executable regular file` message; a positive case keeps an absolute interpreter path loading and running. A case with a getter that throws on read asserts `run()` rejects as seam misuse; a companion with a counting getter asserts the accessor is read exactly once (the snapshot), proving dispatch and the boot frame share the snapshot. The spawn-failure case now stages an executable wrapper, loads the runtime, deletes the wrapper, and asserts the run still resolves `worker-exit` (a load-time-valid path can still fail at run time; the old fixture used a path that is now rejected at load). +- `tests/boot-write-failure.spec.ts` — a fake child backpressures every fd-3 write and destroys the pipe while the host waits for `drain`; the run settles on the wall clock instead of hanging on the drain wait. +- The two staging-leak cases assert the exact paths this test file staged (recorded by the mocked `mkdtempSync`) are gone, instead of diffing a global tmpdir that a sibling worker could perturb. + +## Alternatives considered + +**Leave the explicit-path branch unvalidated and let the first run() report it.** Rejected: a missing, non-executable, or directory interpreter path is a self-contained configuration error that the caller can fix without running a program, and the empty/NUL and basename checks already set the precedent that these fail at load. The run-time `worker-exit` it produced was also indistinguishable from a substrate failure, so the caller could not tell a configuration mistake from an environment problem. + +**Guard the member access inside the dispatch path instead of snapshotting.** Rejected: a try around `record[message.name]` would still read the getter on EVERY call, repeating its side effects and allowing its key set to differ between the boot frame's advertisement and dispatch. Snapshotting once, during validation, converts the throw into the seam-misuse rejection run() already reserves and fixes the key set to one record. + +**Extend the drain wait with a timeout.** Rejected: a timeout would settle the wait while the pipe might still be alive, dropping a queued reply that a still-open pipe could have taken. Listening for `close`/`error` settles exactly when the pipe is gone, which is the only case where `drain` can never arrive. + +## Consequences + +Load now rejects a self-contained configuration error earlier (an explicit interpreter path that is not an executable regular file), matching the basename treatment. Binding member accessors are read once, at validation, so a getter's side effects cannot repeat per call. A destroyed fd-3 pipe no longer strands the reply drain. The leak assertions are immune to concurrent staging by sibling workers. diff --git a/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-load-and-dispatch-hardening.zh.md b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-load-and-dispatch-hardening.zh.md new file mode 100644 index 0000000000..64772eb14a --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-load-and-dispatch-hardening.zh.md @@ -0,0 +1,41 @@ +# Agent Note: CPython 后端的加载期 pythonBin 校验、binding 快照与回复排空结算 + +Status: implemented + +[English](2026-08-29-code-runtime-python-load-and-dispatch-hardening.md) | 中文 + +## Problem + +对 CPython 子进程后端（packages/experimental/code-runtime-python）的评审浮出四项非阻断发现，在长驻宿主上仍可能表现异常：显式 `pythonBin` 路径绕过加载期配置校验；抛错的 binding 成员访问器可能逃出 fd-3 data 回调并终止宿主；回复排空可能永远等待一个已销毁管道不会再发出的 `drain` 事件；两处泄漏断言对全局 tmpdir 做差集，并行 vitest worker 可能误报。 + +## Decision + +### 显式 pythonBin 在加载期必须是可执行的普通文件 + +`resolvePythonBin` 对绝对路径或含斜杠的 `pythonBin` 原样返回，因此不存在、不可执行或指向目录的路径能通过构造器的加载期检查（只拒绝空串/NUL 值与无法解析的裸名），直到首次 `run()` 才以误导性的 `worker-exit` 暴露。显式路径分支现在复用 PATH 分支所用的 `accessSync(X_OK)` + `statSync().isFile()` 检查（目录也能通过 `X_OK`，因此普通文件要求是起决定作用的一半），先把相对显式路径解析到宿主 CWD——与 `spawn` 会查找的位置相同。失败的显式路径使 `resolvePythonBin` 返回 `undefined`，加载检查现在在消息中区分两类失败：显式路径报 `is not an executable regular file`，裸名报 `does not resolve on PATH`。 + +### binding 可调用对象在校验期被快照 + +`namespace.functions` 由调用方提供，其成员可能通过 getter 或 Proxy 暴露。在 fd-3 `data` 回调中读取其中一个成员——`record[message.name]`——会在分发器 try 之外抛出并终止宿主（即使安装了 `uncaughtException` 处理器，运行也只会退化到墙钟超时）。`validateBindings` 现在在 run() 的同步校验段把每个成员读入一个普通自有属性记录，因此抛错的访问器变成 run() 为畸形 binding 预留的 seam-misuse 拒绝。该快照同时是 boot 帧宣告与分发读取的同一份键集，因此键随读取变化的 getter 无法让子进程被允许的名字与宿主实际调用的名字失步。记录采用无原型构造（`Object.create(null)`）：seam 契约把 `__proto__`、`constructor` 之类的成员名当作普通自有属性，普通 `{}` 对 `__proto__` 的赋值会命中原型 setter 而非创建自有属性，使该名字从 boot 帧消失、对其的调用以 `KeyError` 失败。 + +### 回复排空在管道已销毁时结算 + +`drainReplies` 在缓冲区满写入后 `await once(proto, 'drain')`；在等待期间被销毁的管道（子进程退出、close 截止时间拆卸）永远不会再发出 `drain`，而 `events.once` 只在 `error` 时拒绝、不在 `close` 时结算——该 await 可能永远挂起，使 `draining` 保持 true，未消费的队列（及其仍持有的宽 payload）随闭包滞留。等待现在同时监听 `drain`、`close` 与 `error`，任一事件胜出即移除全部三个监听器；排空循环在下一次写入前用 `proto.destroyed` 短路，因此 `finally` 会清空队列并复位 `draining`。 + +## Testing + +- `tests/runtime.spec.ts`——加载拒绝用例覆盖不存在的绝对路径、不可执行的普通文件、目录与含斜杠的相对路径，各自断言 `is not an executable regular file` 消息；一个正向用例让绝对解释器路径通过加载并运行。一个 getter 在读取时抛错的用例断言 `run()` 以 seam misuse 拒绝；一个配套用例用计数 getter 断言访问器恰好被读取一次（快照），证明分发与 boot 帧共享快照。spawn 失败用例现在先暂存一个可执行 wrapper、加载 runtime、删除 wrapper，再断言运行仍 resolve 为 `worker-exit`（加载期合法的路径仍可能在运行期失败；旧 fixture 用的路径现在在加载期就被拒绝）。 +- `tests/boot-write-failure.spec.ts`——一个 fake child 让每次 fd-3 写入都背压，并在宿主等待 `drain` 时销毁管道；运行在墙钟上结算，而不是挂在排空等待上。 +- 两处暂存泄漏用例断言本测试文件暂存的确切路径（由被 mock 的 `mkdtempSync` 记录）已消失，而不是对可能被同级 worker 扰动的全局 tmpdir 做差集。 + +## Alternatives considered + +**让显式路径分支不做校验，由首次 run() 报告。** 已拒绝：不存在、不可执行或指向目录的解释器路径是调用方无需运行程序即可修复的自包含配置错误，且空串/NUL 与裸名检查已确立这些应在加载期失败的先例。它产生的运行期 `worker-exit` 也与子进程故障无法区分，调用方无法分辨配置错误与环境问题。 + +**在分发路径内守卫成员访问，而非快照。** 已拒绝：在 `record[message.name]` 周围加 try 仍会在每次调用时读取 getter，重复其副作用，并允许其键集在 boot 帧宣告与分发之间不一致。在校验期快照一次，把抛错转化为 run() 已预留的 seam-misuse 拒绝，并把键集固定为同一份记录。 + +**给排空等待加超时。** 已拒绝：超时会在管道可能仍存活时结算等待，丢弃一个仍可被存活的管道接收的排队回复。监听 `close`/`error` 恰好在管道消失时结算，这是 `drain` 永远不会到达的唯一情形。 + +## Consequences + +加载期现在更早地拒绝一个自包含配置错误（非可执行普通文件的显式解释器路径），与裸名的处理一致。binding 成员访问器在校验期被读取一次，getter 的副作用不会逐次调用重复。已销毁的 fd-3 管道不再搁浅回复排空。泄漏断言对同级 worker 的并发暂存免疫。 diff --git a/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-reply-backlog-and-surrogate-count.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-reply-backlog-and-surrogate-count.i18n.yaml new file mode 100644 index 0000000000..0e084349e7 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-reply-backlog-and-surrogate-count.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-reply-backlog-and-surrogate-count.md +2026-08-29-code-runtime-python-reply-backlog-and-surrogate-count.md: 5ae31f669e2e207bc2f496d11ca3464f032783f1 +2026-08-29-code-runtime-python-reply-backlog-and-surrogate-count.zh.md: 799178dd54ceddd9b80b11e94d723282a037d398 diff --git a/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-reply-backlog-and-surrogate-count.md b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-reply-backlog-and-surrogate-count.md new file mode 100644 index 0000000000..5ae31f669e --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-reply-backlog-and-surrogate-count.md @@ -0,0 +1,33 @@ +# Agent Note: Bound the reply backlog and count lone surrogates without a match list in the CPython backend + +Status: implemented + +English | [中文](2026-08-29-code-runtime-python-reply-backlog-and-surrogate-count.zh.md) + +## Problem + +A further review round on the CPython subprocess backend (packages/experimental/code-runtime-python) surfaced two unbounded-allocation findings. First, `replyQueue` had no bound: a child that never reads fd 3 keeps the reply pipe full forever, so the drain loop waits on `drain` while every call frame it keeps sending resolves a binding and queues another reply — the backlog (and the binding results it pins) grows until the wall clock. Second, `_json_str_cost` counted lone surrogates with `_SURROGATE.findall(folded)`, which materializes one single-character string per surrogate: a surrogate-dense completion value near the budget (each surrogate serializes to six bytes, so a budget-sized value holds millions of them) allocates millions of objects before the meter returns, defeating the meter's own contract of counting without building. + +## Decision + +### The reply backlog is capped at 1024 pending frames + +`sendReply` now counts pending replies separately from the consumed slots the drain loop clears, and settles the run as a `worker-exit` with a reply-queue message before pushing when the backlog reaches `MAX_PENDING_REPLIES`. The counter is decremented as the drain writes each frame and reset when the drain finishes, so it measures only replies the host still holds. This mirrors the frame cap's treatment of an oversized inbound frame: a child that stops participating in the protocol fails the run early instead of growing host memory until the wall clock. It is a count bound, not a byte bound — binding results carry no seam-level byte cap, so the bound limits how many are retained, not how large any one is. + +### Lone surrogates are counted by length difference, not by a match list + +`_json_str_cost` computed `lone = len(_SURROGATE.findall(folded))`, building a list of one single-character string per lone surrogate. The count is now the length difference between `folded` and `without = _SURROGATE.sub("", folded)`: after pair-combining, every remaining surrogate is lone and exactly one code point, so the number removed is the count, and the `without` string is needed by the meter anyway. The meter returns the identical byte cost with no per-surrogate objects. + +## Testing + +- `tests/runtime.spec.ts` — a hostile child floods 5000 sequential valid call frames and never reads fd 3; the run settles as `worker-exit` with the reply-queue message long before `maxWallMs`, proving the backlog cap fires instead of a wall-clock timeout. A surrogate-dense completion of 3,000,000 lone surrogates pins the boundary at scale: 18,000,002 serialized bytes succeed at an 18,000,002 budget and report `output-limit` one byte under, proving the meter counts every surrogate exactly (the len-diff is verified equal to the old findall count across lone-high, lone-low, paired, astral, and mixed cases). + +## Alternatives considered + +**Pause the fd-3 read side while waiting for drain instead of capping the queue.** Rejected: pausing reads would also stall processing of `done` and `log` frames the child may send after its last call, changing settlement timing; a count cap is deterministic and matches the existing frame-cap pattern. + +**Keep findall and rely on the character-count lower bound.** Rejected: the lower bound admits a string by CHARACTER count while each surrogate serializes to six bytes, so a budget-sized surrogate-dense string passes it and reaches the meter; the match list is exactly the allocation the meter exists to avoid. + +## Consequences + +A child that stops consuming its replies now fails the run as a `worker-exit` once 1024 replies are retained, bounding host memory without a wall-clock wait. The completion-value meter counts lone surrogates with no per-surrogate allocation, keeping its documented counting-without-building contract for surrogate-dense values. diff --git a/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-reply-backlog-and-surrogate-count.zh.md b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-reply-backlog-and-surrogate-count.zh.md new file mode 100644 index 0000000000..799178dd54 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-29-code-runtime-python-reply-backlog-and-surrogate-count.zh.md @@ -0,0 +1,33 @@ +# Agent Note: 在 CPython 后端限制回复积压并改用长度差计数孤立代理项 + +Status: implemented + +[English](2026-08-29-code-runtime-python-reply-backlog-and-surrogate-count.md) | 中文 + +## Problem + +对 CPython 子进程后端（packages/experimental/code-runtime-python）的又一轮评审浮出两项无界分配发现。其一，`replyQueue` 没有上限：从不读取 fd 3 的子进程让回复管道永远占满，排空循环只能等待 `drain`，而它持续发送的每个调用帧都会解析一个 binding 并入队一条回复——积压（连同其钉住的 binding 结果）一直增长到墙钟。其二，`_json_str_cost` 用 `_SURROGATE.findall(folded)` 计数孤立代理项，每个代理项物化一个单字符字符串：接近预算的代理项密集完成值（每个代理项序列化为六个字节，预算大小的值可容纳数百万个）会在计量返回前分配数百万个对象，违背计量器自身「计数而不构建」的契约。 + +## Decision + +### 回复积压限制为 1024 个待发帧 + +`sendReply` 现在把待发回复数与排空循环已清空的槽位分开计数，当积压达到 `MAX_PENDING_REPLIES` 时，在入队前以带回复队列消息的 `worker-exit` 结算运行。计数器在排空写入每帧时递减、排空结束时重置，因此只度量宿主仍持有的回复。这与帧上限对超大入站帧的处理一致：停止参与协议的子进程让运行提前失败，而不是让宿主内存增长到墙钟。这是计数上限而非字节上限——binding 结果在 seam 层没有字节上限，因此该上限限制保留的数量，而非单个结果的大小。 + +### 孤立代理项改用长度差计数，而非匹配列表 + +`_json_str_cost` 原先计算 `lone = len(_SURROGATE.findall(folded))`，为每个孤立代理项构建一个单字符字符串的列表。现在计数改为 `folded` 与 `without = _SURROGATE.sub("", folded)` 的长度差：配对合并后，剩余的每个代理项都是孤立且恰好一个码点，因此被移除的数量即计数，而 `without` 字符串本就是计量需要的。计量器返回完全相同的字节成本，且不产生任何按代理项计的对象。 + +## Testing + +- `tests/runtime.spec.ts`——敌意子进程洪泛 5000 个连续合法调用帧且从不读取 fd 3；运行在远早于 `maxWallMs` 时以带回复队列消息的 `worker-exit` 结算，证明积压上限先于墙钟超时触发。3,000,000 个孤立代理项的代理项密集完成值在规模上钉住边界：18,000,002 个序列化字节在 18,000,002 预算下成功、少一个字节时报 `output-limit`，证明计量器精确计数每个代理项（长度差在孤立高、孤立低、配对、星面和混合用例下与旧 findall 计数逐一相等，已实测验证）。 + +## Alternatives considered + +**在等待 drain 时暂停 fd-3 读侧而非限制队列。** 拒绝：暂停读取也会让子进程在最后一个调用后可能发送的 `done` 与 `log` 帧处理停滞，改变结算时机；计数上限是确定性的，且与既有帧上限模式一致。 + +**保留 findall 并依赖字符计数下界。** 拒绝：下界按字符数放行字符串，而每个代理项序列化为六个字节，因此预算大小的代理项密集字符串能通过下界并进入计量器；匹配列表正是计量器要避免的分配。 + +## Consequences + +停止消费回复的子进程现在会在保留 1024 条回复时以 `worker-exit` 结算运行，无需等待墙钟即可限制宿主内存。完成值计量器对孤立代理项的计数不再产生按代理项计的分，保持其对代理项密集值「计数而不构建」的既有契约。 diff --git a/.agents/notes/implemented/bug-fix/2026-09-01-host-goal-pause-aborts-turn.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-09-01-host-goal-pause-aborts-turn.i18n.yaml new file mode 100644 index 0000000000..f113c39c96 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-09-01-host-goal-pause-aborts-turn.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-09-01-host-goal-pause-aborts-turn.md +2026-09-01-host-goal-pause-aborts-turn.md: 8a6acc1403ea9d8f1dab241e7888bb98aadcde31 +2026-09-01-host-goal-pause-aborts-turn.zh.md: 60170cd1029ce210fa1002ab407b4083af36d72e diff --git a/.agents/notes/implemented/bug-fix/2026-09-01-host-goal-pause-aborts-turn.md b/.agents/notes/implemented/bug-fix/2026-09-01-host-goal-pause-aborts-turn.md new file mode 100644 index 0000000000..8a6acc1403 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-09-01-host-goal-pause-aborts-turn.md @@ -0,0 +1,29 @@ +# Agent Note: Host-initiated goal pause aborts the live turn + +Status: implemented + +English | [中文](2026-09-01-host-goal-pause-aborts-turn.zh.md) + +## Problem + +Clicking "pause goal" in the Web UI moved the goal to `paused` and disarmed automatic continuation, but the model turn already running kept going. The model could keep acting and call `update_goal resume` inside that same turn, immediately undoing the pause, so a manual pause had no real control over goal execution. + +## Decision + +The goal round driver now reads the `change` on every `goal/changed` event. When `operation === 'pause'` and the pause was not initiated by the agent's own turn, the driver aborts the live turn with `agent.cancel({ kind: 'user' }, { keepInbox: true })`. The Web button runs outside any agent initiator boundary, while a model's `update_goal pause` runs with the agent as the current initiator; the driver distinguishes them with `ctx.agents.currentInitiator() !== agent`. The abort is intentionally broad — it stops any live turn, not just a goal round — because a manual pause is a strong "stop now" signal and disarming alone stops future rounds but not the execution already under way. + +`keepInbox` preserves pending work. A queued goal round already fails the existing pre-step reservation check once the goal is disarmed, so it cannot run after the pause. + +The idle handler that pauses a cancelled goal is fenced to the dropped attempt's exact `{ goalId, revision }`. A resume bumps the revision, so a pause followed by an immediate resume — before the aborted turn converges to idle — is preserved instead of being re-paused by the stale cancelled attempt. + +## Alternatives considered + +**Cancel on every pause, including the model's own.** Rejected: a model that pauses in response to a direct human request should finish its turn and report; aborting mid-tool-call cuts off that acknowledgment without adding control. + +**Put the cancellation in the goal service's `pause`.** Rejected: `pause` is one shared entry point for host and model callers, so the service would still need the same initiator test. Keeping control handling in the round driver leaves the goal service a durable state and event owner. + +**Scope the abort to a turn actually running a goal round.** Rejected: the live turn is the execution the user asked to stop, and the extra attempt-state check adds a subtle path without changing the outcome the issue asks for. + +## Consequences + +A Web "pause goal" now aborts the running turn, so the model cannot keep acting or resume the just-paused goal in that turn. A pause followed by an immediate resume keeps the resumed goal running. Model-initiated pauses are unchanged. The change is confined to the round driver and its tests; the goal domain, tool authority, and durable formats are unchanged. diff --git a/.agents/notes/implemented/bug-fix/2026-09-01-host-goal-pause-aborts-turn.zh.md b/.agents/notes/implemented/bug-fix/2026-09-01-host-goal-pause-aborts-turn.zh.md new file mode 100644 index 0000000000..60170cd102 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-09-01-host-goal-pause-aborts-turn.zh.md @@ -0,0 +1,29 @@ +# Agent Note: 宿主发起的 goal 暂停中止当前轮次 + +Status: implemented + +[English](2026-09-01-host-goal-pause-aborts-turn.md) | 中文 + +## 问题 + +在 Web UI 点击「暂停目标」会把 goal 改成 `paused` 并解除自动续跑的武装（disarmed），但已经在跑的模型轮次不会停止。模型还能继续行动，并在同一个轮次里调用 `update_goal resume`，立刻撤销这次暂停，因此人工暂停对 goal 执行没有真正的控制力。 + +## 决策 + +goal round driver 现在会读取每个 `goal/changed` 事件里的 `change`。当 `operation === 'pause'` 且暂停不是由 agent 自己的轮次发起时，driver 用 `agent.cancel({ kind: 'user' }, { keepInbox: true })` 中止当前轮次。Web 按钮运行在任何 agent initiator 边界之外，而模型调用 `update_goal pause` 时当前 initiator 就是该 agent；driver 用 `ctx.agents.currentInitiator() !== agent` 来区分两者。中止是有意放宽的——它会停掉任何正在运行的轮次，而不只是 goal round——因为人工暂停是强烈的「现在停止」信号，仅 disarmed 只能阻止后续轮次，停不掉正在进行的执行。 + +`keepInbox` 会保留待处理工作。一旦 goal 被 disarmed，已排队的 goal round 就会在既有的 pre-step reservation 校验里失败，因此暂停后不会再运行。 + +暂停被取消 goal 的 idle 处理器被栅栏限定到被丢弃 attempt 的精确 `{ goalId, revision }`。resume 会推进 revision，因此在被中止轮次收敛到 idle 之前「暂停后立即 resume」会被保留，而不会被过期的 cancelled attempt 再次暂停。 + +## 考虑过的替代方案 + +**对每次暂停都中止轮次，包括模型自己发起的。** 否决：响应人类直接请求而暂停的模型应当完成本轮并给出回复；在工具调用中途中止只会截断这层确认，却换不来更多控制力。 + +**把中止逻辑放进 goal 服务的 `pause`。** 否决：`pause` 是宿主与模型共用的唯一入口，服务里同样需要这个 initiator 判断。把控制处理留在 round driver，可以让 goal 服务保持为持久状态与事件的拥有者。 + +**把中止限定到真正在跑 goal round 的轮次。** 否决：正在运行的轮次正是用户要求停止的执行，额外的 attempt 状态检查只会增加一条微妙路径，却不改变本 issue 要求的结果。 + +## 后果 + +现在 Web 的「暂停目标」会中止正在运行的轮次，模型无法继续行动或在同一轮次里恢复刚被暂停的 goal。暂停后立即 resume 会保留被恢复的 goal 继续运行。模型发起的暂停行为不变。改动局限于 round driver 及其测试；goal 领域、工具授权与持久化格式都不变。 diff --git a/.agents/notes/implemented/feature/2026-06-15-ptc.i18n.yaml b/.agents/notes/implemented/feature/2026-06-15-ptc.i18n.yaml index ed42409d93..255067c0f0 100644 --- a/.agents/notes/implemented/feature/2026-06-15-ptc.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-06-15-ptc.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-06-15-ptc.md -2026-06-15-ptc.md: 5b3971c33df4be46c8a466e564ac86ba6454663a -2026-06-15-ptc.zh.md: fe44d8a0e97913998fc001c92f632f493888c720 +2026-06-15-ptc.md: c0a0b3ad54716761a74eed8b54c11480e9ca41bc +2026-06-15-ptc.zh.md: b774fc001509f30d41d1c519f344f67d797c2cbf diff --git a/.agents/notes/implemented/feature/2026-06-15-ptc.md b/.agents/notes/implemented/feature/2026-06-15-ptc.md index 5b3971c33d..c0a0b3ad54 100644 --- a/.agents/notes/implemented/feature/2026-06-15-ptc.md +++ b/.agents/notes/implemented/feature/2026-06-15-ptc.md @@ -10,7 +10,7 @@ In the registry's native presentation, the agent loop advertises every visible c For multi-step tool work this is token-heavy and serial. The model cannot compose tools — loop over a result set, branch on an intermediate value, fan out, post-process — without a full model round-trip per call, and each round-trip drags the entire intermediate result back into context whether the model needs it or not. -Cloudflare's [PTC mode](https://blog.cloudflare.com/ptc/) proposes an alternative grounded in a simple observation: LLMs are better at writing code than at emitting tool calls, because they have seen millions of lines of real code and comparatively few contrived tool-calling traces. Instead of one tool call per step, the model writes a TypeScript program against a generated API over the tools, the program executes in a sandboxed runtime, and the model curates what comes back — only what it prints or returns — instead of every intermediate result. +Cloudflare's [Code Mode](https://blog.cloudflare.com/code-mode/) proposes an alternative grounded in a simple observation: LLMs are better at writing code than at emitting tool calls, because they have seen millions of lines of real code and comparatively few contrived tool-calling traces. Instead of one tool call per step, the model writes a TypeScript program against a generated API over the tools, the program executes in a sandboxed runtime, and the model curates what comes back — only what it prints or returns — instead of every intermediate result. Tool presentation belongs to the registry that owns tool visibility: implementing a second presentation as an after-the-fact waterfall transform would make correctness depend on listener order and fight [reconstructable requests](../architecture/2026-07-05-reconstructable-requests.md). The execution substrate is also part of the foundation rather than a placeholder: Node `worker_threads` provides a separate isolate, an empty environment, heap caps, and termination of a hot synchronous loop, while fitting the harness's existing trust model (§Trust posture). diff --git a/.agents/notes/implemented/feature/2026-06-15-ptc.zh.md b/.agents/notes/implemented/feature/2026-06-15-ptc.zh.md index fe44d8a0e9..b774fc0015 100644 --- a/.agents/notes/implemented/feature/2026-06-15-ptc.zh.md +++ b/.agents/notes/implemented/feature/2026-06-15-ptc.zh.md @@ -10,7 +10,7 @@ Status: implemented 对于多步工具操作，这种方式 token 开销大且串行。模型无法组合工具——遍历结果集、根据中间值分支、扇出、后处理——每次调用都需要一次完整的模型往返，而每次往返都会把完整的中间结果拖回上下文，不管模型是否需要。 -Cloudflare 的 [PTC mode](https://blog.cloudflare.com/ptc/) 提出了一种替代方案，基于一个简单的观察：LLM（大语言模型）编写代码的能力优于发出工具调用，因为它们见过数百万行真实代码，而人为构造的工具调用 trace 相对很少。模型不再每步发出一次工具调用，而是针对工具生成的 API 编写一段 TypeScript 程序，程序在沙箱运行时中执行，模型只筛选返回的内容——仅限它 print 或 return 的部分——而非所有中间结果。 +Cloudflare 的 [Code Mode](https://blog.cloudflare.com/code-mode/) 提出了一种替代方案，基于一个简单的观察：LLM（大语言模型）编写代码的能力优于发出工具调用，因为它们见过数百万行真实代码，而人为构造的工具调用 trace 相对很少。模型不再每步发出一次工具调用，而是针对工具生成的 API 编写一段 TypeScript 程序，程序在沙箱运行时中执行，模型只筛选返回的内容——仅限它 print 或 return 的部分——而非所有中间结果。 工具呈现属于掌管工具可见性的注册表：如果把第二种呈现方式实现为事后的 waterfall（瀑布式事件）变换，正确性将依赖监听器顺序，并与[可重建请求](../architecture/2026-07-05-reconstructable-requests.zh.md)冲突。执行基底同样属于基础设施而非占位实现：Node `worker_threads` 提供独立 isolate、空环境、堆上限以及对热同步循环的终止能力，同时契合 harness 既有的信任模型（§信任姿态）。 diff --git a/.agents/notes/implemented/feature/2026-06-30-session-store-fork-api.i18n.yaml b/.agents/notes/implemented/feature/2026-06-30-session-store-fork-api.i18n.yaml index d121a5db15..fc258a1609 100644 --- a/.agents/notes/implemented/feature/2026-06-30-session-store-fork-api.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-06-30-session-store-fork-api.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-06-30-session-store-fork-api.md -2026-06-30-session-store-fork-api.md: ff7617f4bb306926a7b0782751ae82f6ef0c6371 -2026-06-30-session-store-fork-api.zh.md: ecef7ba2985321677b29d3fc7692a8dcb7afb2b6 +2026-06-30-session-store-fork-api.md: a2d169a36a8a0b624abef98377d644d713e5d64c +2026-06-30-session-store-fork-api.zh.md: 5ebf05e677f2d3676f860220b1cf9ff960565402 diff --git a/.agents/notes/implemented/feature/2026-06-30-session-store-fork-api.md b/.agents/notes/implemented/feature/2026-06-30-session-store-fork-api.md index ff7617f4bb..a2d169a36a 100644 --- a/.agents/notes/implemented/feature/2026-06-30-session-store-fork-api.md +++ b/.agents/notes/implemented/feature/2026-06-30-session-store-fork-api.md @@ -20,11 +20,11 @@ The store exposes one operation: type SessionForkSource = Session | SessionId class SessionStore extends Service { - fork(source: SessionForkSource, boundary?: number, childSessionId?: SessionId): Session + fork(source: SessionForkSource, boundary?: SessionSeq, childSessionId?: SessionId): Session } ``` -`boundary` is the inclusive source event `seq` to copy through. When omitted, it defaults to the source session's current last event; on an empty source, omitted `boundary` creates an empty child. Fork-specific validation checks that the requested boundary exists and that the selected prefix's latest turn boundary is not an unmatched `turn/start`. The selected prefix may therefore end at `turn/end` or at a later standalone event, then is deep-cloned into the child seed. The child inherits the source session's `cwd`, stamps `parentSession` to the source id, and sets `seedLength` to the copied prefix length. When `childSessionId` is omitted, `SessionStore` generates one using its existing id policy. +`boundary` is the branded inclusive source event `seq` to copy through. When omitted, it defaults to the source session's current last event; on an empty source, omitted `boundary` creates an empty child. Fork-specific validation checks that the requested boundary exists and that the selected prefix's latest turn boundary is not an unmatched `turn/start`. The selected prefix may therefore end at `turn/end` or at a later standalone event, then is deep-cloned into the child seed. The child inherits the source session's `cwd`, stamps `parentSession` to the source id, sets logical `isSeeded: true`, and supplies the copied prefix length separately as `inheritedEventCount`. When `childSessionId` is omitted, `SessionStore` generates one using its existing id policy. An empty prefix is forkable; any non-empty boundary must be a safe existing sequence outside an open turn. Typed errors distinguish missing sources, stale objects, duplicate child ids, invalid boundaries, and prefixes ending during execution. Broader log validation and crash repair remain with their existing owners. @@ -44,6 +44,6 @@ The Host creates the child through the agent registry with the selected seed and ## Consequences -The public API stays small and discoverable: live session branching is part of `ctx.sessions`, next to `create({ seed })`, rather than a standalone service or a two-step helper pair. Persistence continues to work through existing `session/created` and `session/flush` behavior: a forked child starts life with seeded events, so existing backends persist that seed once and preserve `parentSession` / `seedLength` in the header. +The public API stays small and discoverable: live session branching is part of `ctx.sessions`, next to `create({ seed })`, rather than a standalone service or a two-step helper pair. Persistence continues to work through existing `session/created` and `session/flush` behavior: a forked child starts life with seeded events, so the JSONL backend persists that seed once and preserves logical `parentSession` / `isSeeded` plus the separate exact cut (encoded as v0 physical `seedLength`). This decision excludes ACP `session/fork`, unloaded persisted-session forking, model-facing tools, and subagent refactors. If a future ACP method is added, it should advertise the capability only after it has protocol and snapshot coverage; this Agent Note adds no ACP wire behavior, so no ACP snapshot is required. Fork-child replay remains covered by the existing [seed-boundary testing Agent Note](../testing/2026-06-22-fork-child-replay-seed-boundary.md); focused store, Host, carrier, and client tests pin the boundary and reconciliation contracts, while the real Chromium scenario pins the assembled message action and lineage tree. diff --git a/.agents/notes/implemented/feature/2026-06-30-session-store-fork-api.zh.md b/.agents/notes/implemented/feature/2026-06-30-session-store-fork-api.zh.md index ecef7ba298..5ebf05e677 100644 --- a/.agents/notes/implemented/feature/2026-06-30-session-store-fork-api.zh.md +++ b/.agents/notes/implemented/feature/2026-06-30-session-store-fork-api.zh.md @@ -20,11 +20,11 @@ store 暴露一个操作： type SessionForkSource = Session | SessionId class SessionStore extends Service { - fork(source: SessionForkSource, boundary?: number, childSessionId?: SessionId): Session + fork(source: SessionForkSource, boundary?: SessionSeq, childSessionId?: SessionId): Session } ``` -`boundary` 是要复制到的源事件 `seq`（含该序号）。省略时默认为源会话当前的最后一个事件；对空源会话省略 `boundary` 则创建一个空的子会话。fork 特有的校验会检查请求的边界存在，并确认所选前缀最近的轮次边界不是未匹配的 `turn/start`。因此，所选前缀可以结束于 `turn/end` 或更晚的独立事件，随后被深拷贝到子会话的种子中。子会话继承源会话的 `cwd`，将 `parentSession` 设为源会话 id，并将 `seedLength` 设为已复制前缀的长度。省略 `childSessionId` 时，`SessionStore` 使用其现有的 id 策略生成一个。 +`boundary` 是要复制到的品牌化源事件 `seq`（含该序号）。省略时默认为源会话当前的最后一个事件；对空源会话省略 `boundary` 则创建一个空的子会话。fork 特有的校验会检查请求的边界存在，并确认所选前缀最近的轮次边界不是未匹配的 `turn/start`。因此，所选前缀可以结束于 `turn/end` 或更晚的独立事件，随后被深拷贝到子会话的种子中。子会话继承源会话的 `cwd`，将 `parentSession` 设为源会话 id，设置 logical `isSeeded: true`，并把复制前缀的长度单独作为 `inheritedEventCount` 传入。省略 `childSessionId` 时，`SessionStore` 使用其现有的 id 策略生成一个。 空前缀可以被 fork；任何非空边界都必须是位于开放轮次之外且安全、已存在的序号。类型化的错误区分源缺失、对象陈旧、子 id 重复、边界无效和前缀结束于执行过程中等情况。更广泛的日志校验与崩溃恢复仍由其现有的负责方处理。 @@ -44,6 +44,6 @@ Host 通过 agent（智能体）注册表，以选定的种子和谱系创建子 ## 后果 -公开 API 保持精简且易于发现：活跃会话分支是 `ctx.sessions` 的一部分，紧邻 `create({ seed })`，而非一个独立服务或一对两步辅助函数。持久化继续通过现有的 `session/created` 和 `session/flush` 行为运作：fork 出的子会话创建时便带有种子事件，因此现有后端只需持久化该种子一次，并在 header 中保存 `parentSession`／`seedLength`。 +公开 API 保持精简且易于发现：活跃会话分支是 `ctx.sessions` 的一部分，紧邻 `create({ seed })`，而非一个独立服务或一对两步辅助函数。持久化继续通过现有的 `session/created` 和 `session/flush` 行为运作：fork 出的子会话创建时便带有种子事件，因此 JSONL 后端只需持久化该种子一次，并保留 logical `parentSession`／`isSeeded` 与单独的精确 cut（编码为 v0 物理 `seedLength`）。 本决策排除 ACP（Agent Client Protocol）`session/fork`、对未加载的已持久化会话执行 fork、面向模型的工具，以及 subagent 重构。如果未来添加 ACP 方法，应在具备协议与快照覆盖后才声明支持该能力；本 Agent Note 不添加任何 ACP 协议行为，因此不需要 ACP 快照。fork 子会话的回放仍由现有的[种子边界测试 Agent Note](../testing/2026-06-22-fork-child-replay-seed-boundary.zh.md)覆盖；store、Host、载体与客户端的专项测试固定边界和对账约定，真实 Chromium 场景则固定组装后的消息操作与谱系树。 diff --git a/.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.i18n.yaml b/.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.i18n.yaml index 50b5c20bf4..9f4c455c9c 100644 --- a/.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.md -2026-07-10-agent-session-identity-and-log-location.md: 1bd16fa4123aa8a44719aa0e8c40c4e662f7cb3b -2026-07-10-agent-session-identity-and-log-location.zh.md: 1b54949fb34a0593eaa255e8ff548c203c8023c8 +2026-07-10-agent-session-identity-and-log-location.md: c849ffb882a3334c380b6736a8c8fd8d07a9da40 +2026-07-10-agent-session-identity-and-log-location.zh.md: 07826993839257dd893ef2a56059170ef9d0db31 diff --git a/.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.md b/.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.md index 1bd16fa412..c849ffb882 100644 --- a/.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.md +++ b/.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.md @@ -1,4 +1,4 @@ -# Agent Note: Expose agent session identity and JSONL location to tools and hooks +# Agent Note: Expose agent session identity to tools and hooks Status: implemented @@ -6,29 +6,12 @@ English | [中文](2026-07-10-agent-session-identity-and-log-location.zh.md) ## Problem -An agent can identify its workspace through `session.header.cwd`, but a model using bash cannot reliably identify the session that owns the call or the durable transcript that records it. Searching `./.sessions` guesses deployment config and JSONL layout; custom roots, alternate persistence backends, resume, forks, and concurrent parent/child agents make that guess unreliable. Hooks have the same need for transcript location, while future plugins may need to expose other harness-owned environment facts to shell commands. +An agent can identify its workspace through `session.header.cwd`, but a model using bash cannot reliably identify the session that owns the call. Resume, forks, and concurrent parent/child agents make any ambient guess unreliable, while future plugins may need to expose other harness-owned environment facts to shell commands. The boundary must preserve two properties: the owner of a fact decides how to resolve it, and every child receives a per-execution snapshot rather than process-global mutable state. In particular, a nested harness must not leak its ambient `DSH_*` values into a child whose current agent, persistence backend, or configuration differs. ## Decision -Extend the [`SessionPersistence`](../architecture/2026-06-14-session-persistence.md) seam with a synchronous, side-effect-free location query: - -```ts -import type { SessionHeader } from '@deepseek-ai/dsh-session' - -interface SessionLocation { - readonly kind: string - readonly path: string -} - -interface SessionPersistence { - locate(meta: SessionHeader): SessionLocation | undefined -} -``` - -`path` is an absolute local path to the provider's dedicated log for `meta`; `kind` identifies the representation. JSONL returns `{ kind: 'jsonl', path }` using its resolved root and path helpers. An out-of-tree provider without an honest local per-Session artifact returns `undefined`. The query creates and flushes nothing, so it can report a lazy target path before that file exists. - The model-facing bash package owns a `ctx.shellEnv` registry. A contributor declares its stable name, every `DSH_*` key it may return, a description for each key, and `resolve(execution: ToolExecution)`. Duplicate contributor names, duplicate key ownership, reserved keys, malformed declarations, undeclared runtime output, and non-string output fail loudly. Registration is a Cordis effect and is removed with the contributing plugin fiber. `list()` exposes declarations without running resolvers, keeping the environment API enumerable for diagnostics and future prompt/UI consumers. The registry rebuilds a trusted overlay for every foreground and background bash `ToolExecution`: @@ -36,15 +19,16 @@ The registry rebuilds a trusted overlay for every foreground and background bash - `DSH_HOME` is always the absolute configured Harness home. The standalone [`@deepseek-ai/dsh-home-paths`](../../../../packages/util/home-paths/README.md) utility owns its precedence: explicit `dshHome`, then ambient `$DSH_HOME`, then `~/.dsh`. - `DSH_SHELL=1` is always present and identifies a model bash child managed by DeepSeek Harness. - `DSH_SESSION_ID` is present when the execution has an agent and equals `agent.session.header.id`. -- The built-in persistence translator contributes `DSH_SESSION_JSONL` only when `ctx.sessionPersistence.locate(header)` returns `kind: 'jsonl'`. -Session persistence remains the fact owner: JSONL does not depend on tool-bash or register shell variables itself, and hooks continue to consume `locate()` directly. Tool-bash is the translation layer from the persistence fact into a shell convention. Other plugins that need shell-visible facts depend on the registry and register their own keys; they do not modify `process.env`. +A transcript-location fact is deliberately absent. An earlier form of this decision also extended the persistence seam with a `locate()` path query feeding a `DSH_SESSION_JSONL` variable and the hook bridges' `transcript_path`; the [persistence export and pre-release trims](../simplification/2026-08-27-persistence-export-and-pre-release-trims.md) note owns removing that half — the paths were only readable with compression disabled, and the seam no longer exposes artifact locations. + +Plugins that need shell-visible facts depend on the registry and register their own keys; they do not modify `process.env`. The bash seam exports `DSH_ENV_PREFIX` as the single namespace source and derives `DshEnvironmentKey` from its `typeof`. Tool-bash derives built-in names and model guidance from that constant, while executors use it for ambient filtering. The seam carries the managed overlay separately as `ShellExecRequest.dshEnv` / `ShellExecSpec.dshEnv`: ordinary `env` remains the general in-process plugin surface used by hooks, while `dshEnv` is typed to managed keys. The local executor removes every inherited ambient managed key, applies its ordinary scrub/terminal environment/explicit `env`, and finally merges the trusted `dshEnv` snapshot, so an `env` entry can never displace a managed value. This guarantees that a missing value means absent now rather than inherited from an outer or previous harness. The model-facing tool still ignores model-supplied `env`/`stdin` arguments. The bash tool description teaches only the durable convention: current harness environment facts are available through managed `$DSH_*` variables and may be inspected when needed. It does not enumerate persistence-specific keys or add a permanent system-prompt section. Tool schemas are already logged in request headers and tool output is logged as `tool/result`, so no new session event is required. -The [Claude Code and Codex hook bridges](2026-06-30-hook-bridges.md) resolve transcript location from the same persistence seam when constructing payloads. Codex uses `transcript_path: string | null`; Claude Code preserves its string field and falls back to `''`. Hook lookup neither materializes nor flushes a session. +The [Claude Code and Codex hook bridges](2026-06-30-hook-bridges.md) keep `transcript_path` in their wire payloads for protocol shape but always send `''` (Claude Code) / `null` (Codex); the [persistence export and pre-release trims](../simplification/2026-08-27-persistence-export-and-pre-release-trims.md) note owns that degradation. ## Peer product findings @@ -52,36 +36,30 @@ Peer products separate stable identity from physical storage. Codex injects stab ## Lifecycle and persistence semantics -A fresh session receives its id before the first turn, so its first bash call can read `DSH_SESSION_ID` and a JSONL target. The JSONL file may still be absent until the first successful turn-end checkpoint, and during an open turn it contains only the last flushed prefix. `DSH_SESSION_JSONL` is a location hint, not an authorization credential or freshness guarantee. - -Resume reuses the loaded header and therefore the same id and location. Fork and spawn create new session ids and locations. Parent and child calls resolve from their own `ToolExecution.agent`; each command receives an immutable snapshot even when calls overlap. A persistence service replacement affects later collections because the translator queries `ctx.get('sessionPersistence')` at execution time; the registry itself is effect-scoped and HMR-safe. +A fresh session receives its id before the first turn, so its first bash call can read `DSH_SESSION_ID`. Resume reuses the loaded header and therefore the same id. Fork and spawn create new session ids. Parent and child calls resolve from their own `ToolExecution.agent`; each command receives an immutable snapshot even when calls overlap. The registry is effect-scoped and HMR-safe. `dshHome` is session-independent deployment context. Agent-core resolves one value through `@deepseek-ai/dsh-home-paths` and routes it to both tool-bash and local skill discovery; standalone consumers call the same resolver. If top-level `dshHome` and `skills.local.dshHome` are both supplied and resolve differently, composition fails instead of exposing contradictory homes. Persistence may change independently without freezing its facts into the session prefix. ## Testing -Unit coverage pins registry declaration validation, effect disposal, per-execution collection, the `dshHome` precedence, and the local executor's `DSH_*` scrub/rebuild order. Request-recording tests cover foreground/background snapshots, no-agent calls, absent/JSONL persistence, ignored model `env`, and parent/child isolation. JSONL and no-artifact locator contract tests plus both hook bridge suites pin available and unavailable transcript dialects. +Unit coverage pins registry declaration validation, effect disposal, per-execution collection, the `dshHome` precedence, and the local executor's `DSH_*` scrub/rebuild order. Request-recording tests cover foreground/background snapshots, no-agent calls, ignored model `env`, and parent/child isolation. Both hook bridge suites pin the constant degraded transcript dialects. -A keyless full-loop integration drives the real agent loop, JSONL persistence, tool-bash, and bash-local on the first turn. The child prints `DSH_HOME`, `DSH_SHELL`, session id, JSONL target, and an inherited stale sentinel; the test verifies current values, absence of the stale variable, pre-flush file absence, and the eventual persisted header. Snapshot coverage pins the generic bash description in the recorded request header. No with-key test is required because the contract is deterministic local execution rather than model choice. +A keyless full-loop integration drives the real agent loop, JSONL persistence, tool-bash, and bash-local on the first turn. The child prints `DSH_HOME`, `DSH_SHELL`, session id, and an inherited stale sentinel; the test verifies current values, absence of the stale variable, and the eventual persisted header. Snapshot coverage pins the generic bash description in the recorded request header. No with-key test is required because the contract is deterministic local execution rather than model choice. ## Alternatives considered **Only an id plus `find`.** Search cannot know a custom root or backend layout and races under multiple sessions. -**Only an absolute path.** A path can be unavailable, lazy, or representation-specific and is not stable session identity. - **Global `process.env`.** Concurrent agents would overwrite one another and nested harnesses would inherit stale current-session values. -**Put persistence instructions in the session prefix.** A session prefix is frozen while the active service can change across HMR or future backend switching; persistence-specific guidance would become stale. - **A typed waterfall event.** Listeners cannot declare ownership without running, and later listeners can silently overwrite keys. A registry detects key conflicts at registration and remains enumerable. -**Have each persistence backend register bash env directly.** That reverses the dependency from storage into one consumer and forces bash into deployments that do not use it. `locate()` is also still required by hooks. +**Have each persistence backend register bash env directly.** That reverses the dependency from storage into one consumer and forces bash into deployments that do not use it. **A model-facing `session_info` tool.** It adds schema and another call while bash already supplies the query API; the registry generalizes to future environment facts without one tool per fact. ## Consequences -Every model bash child receives current Harness home and shell identity, and agent calls additionally receive stable session identity. JSONL-backed calls get an optional target path; non-file persistence omits it honestly. The managed `DSH_*` facts inside these children come from the harness: ambient values are removed, current trusted values are re-added last, and an ordinary caller's `env` entry cannot displace them. +Every model bash child receives current Harness home and shell identity, and agent calls additionally receive stable session identity. The managed `DSH_*` facts inside these children come from the harness: ambient values are removed, current trusted values are re-added last, and an ordinary caller's `env` entry cannot displace them. -The namespace is discoverable but not secret. Paths can reveal configured roots, lazy targets can be absent or stale, and a command can override variables inside its own shell syntax. Consumers treat them as correlation and environment facts, verify transcript metadata when attribution matters, and rely on sandbox/filesystem policy rather than variable secrecy for authorization. +The namespace is discoverable but not secret. `DSH_HOME` can reveal a configured root, and a command can override variables inside its own shell syntax. Consumers treat them as correlation and environment facts and rely on sandbox/filesystem policy rather than variable secrecy for authorization. diff --git a/.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.zh.md b/.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.zh.md index 1b54949fb3..0782699383 100644 --- a/.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.zh.md +++ b/.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.zh.md @@ -1,4 +1,4 @@ -# Agent Note: 向工具与钩子公开 agent 会话标识和 JSONL 位置 +# Agent Note: 向工具与钩子公开 agent 会话标识 Status: implemented @@ -6,29 +6,12 @@ Status: implemented ## 问题 -agent（智能体）可以通过 `session.header.cwd` 识别其工作区，但使用 bash 的模型无法可靠识别当前调用所属的会话，也无法找到记录该调用的持久 transcript（文本记录）。搜索 `./.sessions` 等同于猜测部署配置和 JSONL 布局；自定义根目录、替代持久化后端、恢复、fork，以及并发运行的父子 agent，都会让这种猜测失效。钩子同样需要 transcript 位置，而未来的插件也可能需要向 shell 命令公开其他由 harness 所有的环境事实。 +agent（智能体）可以通过 `session.header.cwd` 识别其工作区，但使用 bash 的模型无法可靠识别当前调用所属的会话。恢复、fork 以及并发运行的父子 agent 会让任何来自环境的猜测都不可靠，而未来的插件也可能需要向 shell 命令公开其他由 harness 所有的环境事实。 这项边界必须维持两个属性：事实的所有者决定如何解析该事实；每个子进程接收每次执行的快照，而不是进程级可变全局状态。尤其是嵌套 harness 不能把环境中的 `DSH_*` 值泄漏给当前 agent、持久化后端或配置均可能不同的子进程。 ## 决策 -在 [`SessionPersistence`](../architecture/2026-06-14-session-persistence.zh.md) seam 上增加同步、无副作用的位置查询： - -```ts -import type { SessionHeader } from '@deepseek-ai/dsh-session' - -interface SessionLocation { - readonly kind: string - readonly path: string -} - -interface SessionPersistence { - locate(meta: SessionHeader): SessionLocation | undefined -} -``` - -`path` 是 provider 为 `meta` 保留的专用日志本地绝对路径；`kind` 标识其表示。JSONL 使用解析后的 root 与路径 helper 返回 `{ kind: 'jsonl', path }`。无法诚实提供逐 Session 本地产物的仓库外 provider 返回 `undefined`。该查询不会创建或刷写任何内容，因此即使文件尚不存在，也可以报告按需创建的目标路径。 - 面向模型的 bash 包拥有一个 `ctx.shellEnv` 注册表。贡献方声明稳定名称、它可能返回的每个 `DSH_*` 键、每个键的说明，以及 `resolve(execution: ToolExecution)`。贡献方名称重复、键所有权重复、使用保留键、声明格式错误、运行时输出未声明或输出不是字符串时，系统都会明确失败。注册属于 Cordis effect，并随贡献插件的 fiber 一同移除。`list()` 无需运行解析器即可公开声明，从而让环境 API 可供诊断工具和未来的提示词／UI 消费方枚举。 注册表会为每次前台和后台 bash `ToolExecution` 重新构建受信任的覆盖层： @@ -36,15 +19,16 @@ interface SessionPersistence { - `DSH_HOME` 始终是配置的 Harness home 绝对路径。独立的 [`@deepseek-ai/dsh-home-paths`](../../../../packages/util/home-paths/README.zh.md) 工具库规定其优先级：显式 `dshHome`，其次是环境中的 `$DSH_HOME`，最后是 `~/.dsh`。 - `DSH_SHELL=1` 始终存在，用于标识由 DeepSeek Harness 管理、面向模型的 bash 子进程。 - 执行具有关联 agent 时，`DSH_SESSION_ID` 存在并等于 `agent.session.header.id`。 -- 内置的持久化转换层提供 `DSH_SESSION_JSONL` 的条件是 `ctx.sessionPersistence.locate(header)` 返回 `kind: 'jsonl'`。 -会话持久化仍然是事实所有者：JSONL 不依赖 tool-bash，也不会自行注册 shell 变量；钩子继续直接使用 `locate()`。tool-bash 是把持久化事实转换为 shell 约定的转换层。其他需要向 shell 公开事实的插件依赖该注册表，并注册各自的键；它们不修改 `process.env`。 +transcript（文本记录）位置事实被有意省略。本决策的早期形式还在持久化 seam 上增加了 `locate()` 路径查询，为 `DSH_SESSION_JSONL` 变量和钩子桥接层的 `transcript_path` 提供来源；[持久化导出与预发布精简](../simplification/2026-08-27-persistence-export-and-pre-release-trims.zh.md) Note 负责移除这一半——这些路径只有在禁用压缩时才可读，该 seam 也不再公开产物位置。 + +需要向 shell 公开事实的插件依赖该注册表，并注册各自的键；它们不修改 `process.env`。 bash seam 导出 `DSH_ENV_PREFIX` 作为唯一的命名空间来源，并派生 `DshEnvironmentKey`，其来源是该常量的 `typeof`。tool-bash 从该常量派生内置名称与模型指引，执行器则使用该常量过滤环境中已有的值。seam 通过 `ShellExecRequest.dshEnv`／`ShellExecSpec.dshEnv` 单独传递受管理的覆盖层：普通 `env` 仍是钩子所用的通用进程内插件接口，`dshEnv` 则以类型约束为受管理键。本地执行器移除环境中继承的全部受管理键，依次应用普通清理、终端环境和显式 `env`，最后合并受信任的 `dshEnv` 快照，因此 `env` 条目永远无法顶掉受管理的值。这保证了值缺失表示它当前确实不存在，而不是从外层或先前的 harness 继承而来。面向模型的工具仍忽略模型提供的 `env`／`stdin` 参数。 bash 工具说明只讲解持久约定：当前 harness 环境事实通过受管理的 `$DSH_*` 变量提供，可以在需要时查看。它不会枚举持久化专用键，也不会添加永久的系统提示词章节。工具 schema 已记录在请求 header 中，工具输出则记录为 `tool/result`，因此无需新增会话事件。 -[Claude Code 和 Codex 钩子桥接层](2026-06-30-hook-bridges.zh.md)在构造 payload 时，从同一持久化 seam 解析 transcript 位置。Codex 使用 `transcript_path: string | null`；Claude Code 保留其字符串字段，并回退为 `''`。钩子查询不会物化或刷写会话。 +[Claude Code 和 Codex 钩子桥接层](2026-06-30-hook-bridges.zh.md)为保持协议格式，仍在线上 payload 中保留 `transcript_path` 字段，但始终发送 `''`（Claude Code）／`null`（Codex）；这项降级由[持久化导出与预发布精简](../simplification/2026-08-27-persistence-export-and-pre-release-trims.zh.md) Note 负责。 ## 同类产品调研 @@ -52,36 +36,30 @@ bash 工具说明只讲解持久约定：当前 harness 环境事实通过受管 ## 生命周期与持久化语义 -新会话在第一个轮次之前获得 id，因此它的首次 bash 调用即可读取 `DSH_SESSION_ID` 和 JSONL 目标。JSONL 文件可能要等到第一次成功的轮次结束检查点后才存在，而且在一个轮次仍未结束时，它只包含上次刷写的前缀。`DSH_SESSION_JSONL` 是位置提示，不是授权凭据或新鲜度保证。 - -恢复操作复用已加载的 header，因此 id 和位置不变。fork 和 spawn 会创建新的会话 id 与位置。父子调用分别从自己的 `ToolExecution.agent` 解析事实；即使调用重叠，每条命令也会收到不可变快照。替换持久化服务会影响后续收集，因为转换层在执行时查询 `ctx.get('sessionPersistence')`；注册表本身受 effect 作用域约束，并且可安全用于 HMR（热模块替换）。 +新会话在第一个轮次之前获得 id，因此它的首次 bash 调用即可读取 `DSH_SESSION_ID`。恢复操作复用已加载的 header，因此 id 不变。fork 和 spawn 会创建新的会话 id。父子调用分别从自己的 `ToolExecution.agent` 解析事实；即使调用重叠，每条命令也会收到不可变快照。注册表受 effect 作用域约束，并且可安全用于 HMR（热模块替换）。 `dshHome` 是与会话无关的部署上下文。agent-core 通过 `@deepseek-ai/dsh-home-paths` 解析出一个值，并将其同时传给 tool-bash 和本地 skill（技能）发现；独立消费方调用同一解析器。如果顶层 `dshHome` 与 `skills.local.dshHome` 均已提供但解析结果不同，组合会失败，而不会公开互相矛盾的 home。持久化可以独立变更，无需把其事实冻结到会话前缀中。 ## 测试 -单元测试覆盖注册表声明校验、effect 释放、逐次执行收集、`dshHome` 优先级，以及本地执行器清理并重建 `DSH_*` 的顺序。请求录制测试覆盖前台／后台快照、无 agent 调用、持久化不存在或为 JSONL、忽略模型 `env`，以及父子隔离。JSONL 与无产物定位器约定测试、两套钩子桥接测试均固定 transcript 可用和不可用两种方言。 +单元测试覆盖注册表声明校验、effect 释放、逐次执行收集、`dshHome` 优先级，以及本地执行器清理并重建 `DSH_*` 的顺序。请求录制测试覆盖前台／后台快照、无 agent 调用、忽略模型 `env`，以及父子隔离。两套钩子桥接测试均锁定恒定的降级 transcript 方言。 -一项无密钥的完整循环集成测试会在第一个轮次驱动真实的 agent loop、JSONL 持久化、tool-bash 与 bash-local。子进程打印 `DSH_HOME`、`DSH_SHELL`、会话 id、JSONL 目标和继承的陈旧哨兵值；测试校验当前值、陈旧变量不存在、刷写前文件不存在，并最终检查持久化 header。快照测试会固定录制请求 header 中的通用 bash 说明。该约定属于确定性的本地执行，不涉及模型选择，因此无需带密钥测试。 +一项无密钥的完整循环集成测试会在第一个轮次驱动真实的 agent loop、JSONL 持久化、tool-bash 与 bash-local。子进程打印 `DSH_HOME`、`DSH_SHELL`、会话 id 和继承的陈旧哨兵值；测试校验当前值、陈旧变量不存在，并最终检查持久化 header。快照测试会固定录制请求 header 中的通用 bash 说明。该约定属于确定性的本地执行，不涉及模型选择，因此无需带密钥测试。 ## 考虑过的替代方案 **只提供 id，再用 `find`。** 搜索无法得知自定义根目录或后端布局，并且在多会话环境下存在竞态。 -**只提供绝对路径。** 路径可能不可用、延迟创建或取决于表示形式，不能作为稳定的会话标识。 - **使用全局 `process.env`。** 并发 agent 会互相覆盖，嵌套 harness 也会继承陈旧的当前会话值。 -**把持久化说明放入会话前缀。** 活动服务可以在 HMR 或未来的后端切换中改变，而会话前缀保持冻结；持久化专用指引会因此变得陈旧。 - **使用类型化 waterfall 事件。** 监听器不运行就无法声明所有权，而后续监听器可以无提示地覆盖键。注册表能在注册时检测键冲突，并且保持可枚举。 -**让每个持久化后端直接注册 bash 环境。** 这会反转依赖方向，让存储层依赖某一个消费方，并迫使未使用 bash 的部署也引入它。钩子仍然需要 `locate()`。 +**让每个持久化后端直接注册 bash 环境。** 这会反转依赖方向，让存储层依赖某一个消费方，并迫使未使用 bash 的部署也引入它。 **增加面向模型的 `session_info` 工具。** bash 已经提供查询 API，新增工具只会多出 schema 和一次调用；注册表可以扩展至未来的环境事实，无需为每项事实增加一个工具。 ## 影响 -每个面向模型的 bash 子进程都会收到当前 Harness home 和 shell 标识，关联 agent 的调用还会收到稳定的会话标识。使用 JSONL 后端的调用可以获得可选的目标路径；非文件持久化会如实省略该值。这些子进程中受管理的 `DSH_*` 事实来自 harness：系统移除环境中已有的受管理值、在最后重新加入当前受信任的值，普通调用方的 `env` 条目无法顶掉它们。 +每个面向模型的 bash 子进程都会收到当前 Harness home 和 shell 标识，关联 agent 的调用还会收到稳定的会话标识。这些子进程中受管理的 `DSH_*` 事实来自 harness：系统移除环境中已有的受管理值、在最后重新加入当前受信任的值，普通调用方的 `env` 条目无法顶掉它们。 -该命名空间可被发现，但并非秘密。路径可能泄露配置的根目录，延迟创建的目标也可能不存在或处于陈旧状态，而且命令可以在自己的 shell 语法中覆盖变量。消费方应把这些值视为关联信息和环境事实，在归属关系重要时校验 transcript 元数据，并依靠沙箱／文件系统策略而不是变量保密性来完成授权。 +该命名空间可被发现，但并非秘密。`DSH_HOME` 可能泄露配置的根目录，而且命令可以在自己的 shell 语法中覆盖变量。消费方应把这些值视为关联信息和环境事实，并依靠沙箱／文件系统策略而不是变量保密性来完成授权。 diff --git a/.agents/notes/implemented/feature/2026-07-21-continuable-background-subagents.i18n.yaml b/.agents/notes/implemented/feature/2026-07-21-continuable-background-subagents.i18n.yaml index 8ec6ca23e0..2bd3567191 100644 --- a/.agents/notes/implemented/feature/2026-07-21-continuable-background-subagents.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-21-continuable-background-subagents.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-21-continuable-background-subagents.md -2026-07-21-continuable-background-subagents.md: b2a3a8c53db5ae2860ed5cc6edccadfcd417e7fa -2026-07-21-continuable-background-subagents.zh.md: 24cc09e621731cbb54f4d081d232f0598220d418 +2026-07-21-continuable-background-subagents.md: 24ab64879fa18f5de9952af2dba321e2cae17477 +2026-07-21-continuable-background-subagents.zh.md: 0cf06f7fe30d0953bb6e7f1a4c96527fa181805a diff --git a/.agents/notes/implemented/feature/2026-07-21-continuable-background-subagents.md b/.agents/notes/implemented/feature/2026-07-21-continuable-background-subagents.md index b2a3a8c53d..24ab64879f 100644 --- a/.agents/notes/implemented/feature/2026-07-21-continuable-background-subagents.md +++ b/.agents/notes/implemented/feature/2026-07-21-continuable-background-subagents.md @@ -57,7 +57,7 @@ The continuation manager does not serialize two callers that race a stopped chil ### Model-facing `send_message` -The model receives one `send_message(subagent_id, message)` tool backed by `SubagentRuntime.followup()`, matching the intent verb on `Agent`. The service operation owns steer-or-resume orchestration and is distinct from the run's `SubagentRun.steer?()`, which only delivers to an already active run. The tool performs no lifecycle routing of its own. It attributes the follow-up as `{ kind: 'coordinator', senderSessionId: parent.id }` and forwards `{ source, signal }`; the service requires both facts in one options object. The source crosses both live steering and cold resume, while cancellation owns only a pending live-delivery wait because a cold-resume Task returns immediately and owns its later cancellation. The child model still receives ordinary user-role content, while the durable source prevents model-generated follow-ups from being classified as direct human input. A human adapter instead supplies `{ kind: 'user' }` and its interaction signal. The tool lives in the separately loaded `@deepseek-ai/dsh-tool-subagent-control` package so provider-bound `@deepseek-ai/dsh-tool-subagent` instances can continue registering distinct delegation tools for spawn, fork, or ACP without registering duplicate global control tools. +The model receives one global `send_message(agent_id, message)` tool backed by `SubagentRuntime.sendMessage()`. The exact live sender may name only its direct parent or direct continuable child; the service owns adjacency checks, cold resume, fixed Steer scheduling, and durable `{ kind: 'agent-message', senderSessionId }` attribution. Cancellation owns work only until inbox acceptance. A human adapter remains separate because browser-authored input carries user provenance and request identity rather than Agent authority. The tool lives in the separately loaded `@deepseek-ai/dsh-tool-subagent-control` package so provider-bound `@deepseek-ai/dsh-tool-subagent` instances can continue registering distinct delegation tools for spawn, fork, or ACP without registering duplicate global control tools. - If the child has a running Task and live-steering capability, the service calls `run.steer(message, source)` and returns the existing Job id; it creates no Task of its own. - If the child has no running Task, `send_message` creates a fresh Task, cold-resumes the durable session with the message, and returns the new Job id. @@ -71,9 +71,9 @@ Human input uses the same `followup` operation. The UI may display the child tra ### Durable child handle and cold resume -The continuation manager snapshots every descriptor input with the seam's `snapshotSubagentDescriptor()` (built on [`snapshotJsonValue`](../../../../packages/util/values/src/index.ts)) before Task creation, matching the detached lossless-JSON boundary already used by Agent messages. A child-scoped setup contribution — a prepended one-shot `agent/prompt-submit` listener installed by the in-process driver — appends one model-hidden `subagent/descriptor` event before downstream prompt admission can block or throw. Allowed admission opens the initial child turn afterward; rejected admission leaves the descriptor as a pre-turn log-only fact, and the activation's final required checkpoint persists it. The event carries no `surfaceOp`, remains outside model history, and survives when compaction replaces surface history. A known child id is resumable only when loading that child session yields a supported descriptor in the child's own suffix (after `seedLength`, so a fork seed cannot leak an ancestor's descriptor) and its header identifies the caller as the direct parent. +The continuation manager snapshots every descriptor input with the seam's `snapshotSubagentDescriptor()` (built on [`snapshotJsonValue`](../../../../packages/util/values/src/index.ts)) before Task creation, matching the detached lossless-JSON boundary already used by Agent messages. A child-scoped setup contribution — a prepended one-shot `agent/prompt-submit` listener installed by the in-process driver — appends one model-hidden `subagent/descriptor` event before downstream prompt admission can block or throw. Allowed admission opens the initial child turn afterward; rejected admission leaves the descriptor as a pre-turn log-only fact, and the activation's final required checkpoint persists it. The event carries no `surfaceOp`, remains outside model history, and survives when compaction replaces surface history. A known child id is resumable only when loading that child session yields a supported descriptor at or after its exact `inheritedEventCount`, so a fork seed cannot leak an ancestor's descriptor, and its header identifies the caller as the direct parent. -The continuable arm of the versioned descriptor (`SUBAGENT_DESCRIPTOR_VERSION` in [descriptor.ts](../../../../packages/subagent/subagent/src/descriptor.ts)) carries `mode: 'continuable'`, the subagent provider name, resolved child `agentOptions.provider` and `agentOptions.model`, and optional `persona` and `toolFilter`. It does not snapshot the merge-extensible `AgentOptions` object: unrelated extension values cannot make continuation fail merely because they are not JSON. It deliberately omits `subagentDepth`; cold resume relies on the persisted header's `delegationDepth` rather than reconstructing depth from the descriptor. `outputSchema` belongs to one activation's result contract rather than durable child composition. The child header remains authoritative for the child id, `cwd`, `parentSession`, `seedLength`, and `delegationDepth`, while the persisted child transcript owns the fork seed and subsequent history. [`delegationDepthOf()`](../../../../packages/subagent/subagent/src/index.ts) takes the maximum of header and runtime values, so reconstructed runtime options may deepen the persisted value but never lower it and a resumed child cannot regain a top-level delegation budget. +The continuable arm of the versioned descriptor (`SUBAGENT_DESCRIPTOR_VERSION` in [descriptor.ts](../../../../packages/subagent/subagent/src/descriptor.ts)) carries `mode: 'continuable'`, the subagent provider name, resolved child `agentOptions.provider` and `agentOptions.model`, and optional `persona` and `toolFilter`. It does not snapshot the merge-extensible `AgentOptions` object: unrelated extension values cannot make continuation fail merely because they are not JSON. It deliberately omits `subagentDepth`; cold resume relies on the persisted header's `delegationDepth` rather than reconstructing depth from the descriptor. `outputSchema` belongs to one activation's result contract rather than durable child composition. The child header remains authoritative for the child id, `cwd`, `parentSession`, `isSeeded`, and `delegationDepth`; body-bearing persistence metadata owns the exact `inheritedEventCount`, while the child transcript owns the fork seed and subsequent history. [`delegationDepthOf()`](../../../../packages/subagent/subagent/src/index.ts) takes the maximum of header and runtime values, so reconstructed runtime options may deepen the persisted value but never lower it and a resumed child cannot regain a top-level delegation budget. Cold resume cannot depend on an optional method of `SubagentRun`, because that run has been disposed and is not retained across process restart. A run represents one disposable activation and exposes only activation-scoped operations. `SubagentRun.steer?()` names the confirmed live-only capability so it cannot be confused with service orchestration or the model-facing tool. diff --git a/.agents/notes/implemented/feature/2026-07-21-continuable-background-subagents.zh.md b/.agents/notes/implemented/feature/2026-07-21-continuable-background-subagents.zh.md index 24cc09e621..0cf06f7fe3 100644 --- a/.agents/notes/implemented/feature/2026-07-21-continuable-background-subagents.zh.md +++ b/.agents/notes/implemented/feature/2026-07-21-continuable-background-subagents.zh.md @@ -57,7 +57,7 @@ durable child Session ### 面向模型的 `send_message` -模型获得一个由 `SubagentRuntime.followup()` 支撑的 `send_message(subagent_id, message)` 工具，与 `Agent` 上的意图动词一致。该服务操作负责在 steering 与恢复之间编排；它不同于 run 的 `SubagentRun.steer?()`，后者只能向已活跃的 run 发送消息。工具本身不执行生命周期路由。该工具将后续消息的来源标记为 `{ kind: 'coordinator', senderSessionId: parent.id }`，并转发 `{ source, signal }`；服务要求在一个选项对象中同时提供这两项信息。来源会贯穿在线 steering 和 cold resume 两条路径，而取消只控制尚未完成的在线投递等待，因为 cold resume Task 会立即返回，并自行负责后续取消。child 模型收到的仍是普通的 user role 内容，而持久化的来源信息可防止模型生成的后续消息被归类为直接用户输入。用户适配器则提供 `{ kind: 'user' }` 及其交互信号。该工具位于单独加载的 `@deepseek-ai/dsh-tool-subagent-control` 包中，因此按提供方绑定的 `@deepseek-ai/dsh-tool-subagent` 实例可以继续为 spawn、fork 或 ACP 注册不同的委派工具，而不会重复注册全局控制工具。 +模型获得一个由 `SubagentRuntime.sendMessage()` 支撑的全局 `send_message(agent_id, message)` 工具。确切在线 sender 只能指定其直接 parent 或直接可继续 child；服务负责相邻关系检查、冷恢复、固定 Steer 调度，以及持久化 `{ kind: 'agent-message', senderSessionId }` 来源信息。取消只负责 inbox 接受前的工作。用户适配器保持分离，因为浏览器编写的输入携带用户来源信息与请求身份，而不是 Agent 权限。该工具位于单独加载的 `@deepseek-ai/dsh-tool-subagent-control` 包中，因此按提供方绑定的 `@deepseek-ai/dsh-tool-subagent` 实例可以继续为 spawn、fork 或 ACP 注册不同的委派工具，而不会重复注册全局控制工具。 - 如果 child 存在运行中的 Task 并支持在线消息，服务会调用 `run.steer(message, source)` 并返回现有 job id；它不会创建新 Task。 - 如果 child 没有运行中的 Task，`send_message` 会创建新 Task，使用该消息从持久化存储恢复会话，并返回新的 job id。 @@ -71,9 +71,9 @@ durable child Session ### 持久化 child handle 与从持久化存储恢复 -继续执行管理器在创建 Task 前，通过 seam 的 `snapshotSubagentDescriptor()`（基于 [`snapshotJsonValue`](../../../../packages/util/values/src/index.ts) 构建）对每项描述符输入建立快照；这一边界与 Agent 消息现有的分离式无损 JSON 边界一致。作用于 child 作用域的 setup contribution——由进程内驱动前置安装的一次性 `agent/prompt-submit` 监听器——会在下游 prompt admission 能够阻止请求或抛出异常之前追加一个对模型隐藏的 `subagent/descriptor` 事件。admission 获准后才会开启 child 的初始轮次；admission 被拒绝时，描述符会作为轮次前的仅日志事实保留，并由该 activation 最终的必需检查点持久化。该事件不携带 `surfaceOp`，不进入模型历史，并在压缩替换 surface 历史时继续保留。只有在加载已知 child id 对应的 child 会话后，能在该 child 自身的后缀中（`seedLength` 之后，因此 fork seed 不会泄露祖先的描述符）得到受支持的描述符，且会话 header 将调用方标识为直接 parent 时，该 id 才可恢复。 +继续执行管理器在创建 Task 前，通过 seam 的 `snapshotSubagentDescriptor()`（基于 [`snapshotJsonValue`](../../../../packages/util/values/src/index.ts) 构建）对每项描述符输入建立快照；这一边界与 Agent 消息现有的分离式无损 JSON 边界一致。作用于 child 作用域的 setup contribution——由进程内驱动前置安装的一次性 `agent/prompt-submit` 监听器——会在下游 prompt admission 能够阻止请求或抛出异常之前追加一个对模型隐藏的 `subagent/descriptor` 事件。admission 获准后才会开启 child 的初始轮次；admission 被拒绝时，描述符会作为轮次前的仅日志事实保留，并由该 activation 最终的必需检查点持久化。该事件不携带 `surfaceOp`，不进入模型历史，并在压缩替换 surface 历史时继续保留。只有在加载已知 child id 对应的 child 会话后，能在其精确 `inheritedEventCount` 位置或之后得到受支持的描述符，从而阻止 fork seed 泄露祖先描述符，且会话 header 将调用方标识为直接 parent 时，该 id 才可恢复。 -版本化描述符的可继续分支（[descriptor.ts](../../../../packages/subagent/subagent/src/descriptor.ts) 中的 `SUBAGENT_DESCRIPTOR_VERSION`）携带 `mode: 'continuable'`、subagent 提供方名称、已解析的 child `agentOptions.provider` 和 `agentOptions.model`，以及可选的 `persona` 与 `toolFilter`。它不会对可通过声明合并扩展的 `AgentOptions` 对象建立快照：与此无关的扩展值不会仅因无法表示为 JSON 而导致继续执行失败。描述符会特意省略 `subagentDepth`；从持久化存储恢复时，系统依赖持久化 header 中的 `delegationDepth`，而不根据描述符重建深度。`outputSchema` 属于单次激活的结果约定，不属于持久化 child 组合配置。child header 仍是 child id、`cwd`、`parentSession`、`seedLength` 和 `delegationDepth` 的权威信息，持久化 child transcript 则负责保存 fork seed 和后续历史。[`delegationDepthOf()`](../../../../packages/subagent/subagent/src/index.ts) 会在 header 值和运行时值中取最大值，因此重建后的运行时选项可以加深持久化值，但绝不能降低它，恢复后的 child 无法重新获得顶层委派预算。 +版本化描述符的可继续分支（[descriptor.ts](../../../../packages/subagent/subagent/src/descriptor.ts) 中的 `SUBAGENT_DESCRIPTOR_VERSION`）携带 `mode: 'continuable'`、subagent 提供方名称、已解析的 child `agentOptions.provider` 和 `agentOptions.model`，以及可选的 `persona` 与 `toolFilter`。它不会对可通过声明合并扩展的 `AgentOptions` 对象建立快照：与此无关的扩展值不会仅因无法表示为 JSON 而导致继续执行失败。描述符会特意省略 `subagentDepth`；从持久化存储恢复时，系统依赖持久化 header 中的 `delegationDepth`，而不根据描述符重建深度。`outputSchema` 属于单次激活的结果约定，不属于持久化 child 组合配置。child header 仍是 child id、`cwd`、`parentSession`、`isSeeded` 和 `delegationDepth` 的权威信息；含正文的持久化 metadata 拥有精确 `inheritedEventCount`，child transcript 则负责保存 fork seed 和后续历史。[`delegationDepthOf()`](../../../../packages/subagent/subagent/src/index.ts) 会在 header 值和运行时值中取最大值，因此重建后的运行时选项可以加深持久化值，但绝不能降低它，恢复后的 child 无法重新获得顶层委派预算。 从持久化存储恢复不能依赖 `SubagentRun` 的可选方法，因为该 run 已被 dispose，并且进程重启后不会保留。run 表示一次可 dispose 的激活，只暴露作用于当前激活的操作。`SubagentRun.steer?()` 这一名称明确指代提供确认语义且仅适用于在线消息的功能，以免该功能与服务编排或面向模型的工具混淆。 diff --git a/.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.i18n.yaml b/.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.i18n.yaml index 05e4629152..67b922e203 100644 --- a/.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.md -2026-07-23-session-telemetry-otel-revival.md: 836531f605eee5c0dcdf108e9ee2b2744d6f72aa -2026-07-23-session-telemetry-otel-revival.zh.md: 9874598a58ff91bd2cc5ef58c771744d3f3d9610 +2026-07-23-session-telemetry-otel-revival.md: 1110db4b0dfcfd68bde97f9964383ddd041f328b +2026-07-23-session-telemetry-otel-revival.zh.md: 6c046c5f963760638deacacda51f8d94a13e0efd diff --git a/.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.md b/.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.md index 836531f605..1110db4b0d 100644 --- a/.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.md +++ b/.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.md @@ -28,7 +28,7 @@ The boundary axiom holds: the harness's aspect ends at `emit()`. Batching, retry **Map onto OTel spans (GenAI semantic conventions) instead of logs.** Rejected for this revival: the branch implementation's log mapping is reviewed and shipped-shaped; the span model is lossy for forkable, interruptible sessions and belongs to a future consumer with real span queries to serve. -**Full-log replay when no handoff cursor survived (re-export constructor seeds).** Shipped in the first revival round, then narrowed: adoption now replays from the session's construction boundary (`Session.firstLiveSeq`, the constructor-seed length — a fact the session already validated but did not expose; `header.seedLength` cannot serve, it is the durable fork-lineage value and a resumed session's constructor seed is its full stored log). A resumed session's history already shipped from the previous process under the same id, and a fork's inherited prefix already shipped in the parent's stream — re-exporting either re-billed every resume for its full history and doubled query-time counts on OTLP backends with no native ingest dedupe. Receivers stitch fork lineage via `session.parent_id` + `session.seed_length`. What the narrowing gives up, consistently with the at-most-once stance: a resume no longer backfills records the previous process failed to deliver (telemetry unmounted then, or queued at crash) — the full replay's only real benefit, bought at the common case's expense. A deployment that states a backfill requirement needs the deferred outbox above, not replay. The boundary also swallows the synthetic turn closers `SessionPersistence.load()` writes when repairing a crash-interrupted log (they sit below `firstLiveSeq` despite never existing in the previous process) — deliberate, not incidental: exporting a synthetic closer cannot complete the remote turn whose real tail records died in the crashed process's queue, it can only make an incomplete turn look closed. The wire stream stays faithful to what the crashed process actually shipped; receivers read a never-closed turn on a resumed stream as "the previous process died inside it" (the OTel README states the rule), and a later clean `shutdown` marker attests only to the resumed process's exit. Threading the pre-repair boundary through load/prepare so repairs export as live events would couple three packages to un-ship that signal. +**Full-log replay when no handoff cursor survived (re-export constructor seeds).** Shipped in the first revival round, then narrowed: adoption now replays from the session's construction boundary (`Session.firstLiveSeq`, the constructor-seed length; `Session.inheritedEventCount` cannot serve because it is the durable fork-lineage cut and a resumed session's constructor seed is its full stored log). A resumed session's history already shipped from the previous process under the same id, and a fork's inherited prefix already shipped in the parent's stream — re-exporting either re-billed every resume for its full history and doubled query-time counts on OTLP backends with no native ingest dedupe. Receivers stitch fork lineage via `session.parent_id` + `session.seed_length`. What the narrowing gives up, consistently with the at-most-once stance: a resume no longer backfills records the previous process failed to deliver (telemetry unmounted then, or queued at crash) — the full replay's only real benefit, bought at the common case's expense. A deployment that states a backfill requirement needs the deferred outbox above, not replay. The boundary also swallows the synthetic turn closers `SessionPersistence.load()` writes when repairing a crash-interrupted log (they sit below `firstLiveSeq` despite never existing in the previous process) — deliberate, not incidental: exporting a synthetic closer cannot complete the remote turn whose real tail records died in the crashed process's queue, it can only make an incomplete turn look closed. The wire stream stays faithful to what the crashed process actually shipped; receivers read a never-closed turn on a resumed stream as "the previous process died inside it" (the OTel README states the rule), and a later clean `shutdown` marker attests only to the resumed process's exit. Threading the pre-repair boundary through load/prepare so repairs export as live events would couple three packages to un-ship that signal. **Forwarding the seam's turn-boundary `flush()` hint to the OTel provider's `forceFlush()`.** Shipped in the first revival round, then removed: three distinct silent-loss paths shared the wrapper state — a dispose racing an in-flight flush (the SDK's concurrent-flush guard makes shutdown's internal drain skip), overlapping hints displacing the retained promise, and the provider's fixed 30-second flush timeout rejecting while the processor still drains. Every path exists only because the forwarding made this backend the process's second flusher against undocumented SDK internals from the upstream experimental tree; with no `flush()` implemented, the batch processor is the only flusher, its `scheduledDelayMillis` (already deployment-tunable through the `processor` passthrough) governs export cadence, and `shutdown()`'s drain is complete by construction. Reinstate only if a deployment states a turn-boundary latency requirement `scheduledDelayMillis` cannot meet — and then by calling the retained `BatchLogRecordProcessor`'s own `forceFlush()`, never the provider's timeout-wrapped one. diff --git a/.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.zh.md b/.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.zh.md index 9874598a58..6c046c5f96 100644 --- a/.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.zh.md +++ b/.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.zh.md @@ -29,7 +29,7 @@ Status: implemented **映射到 OTel span（GenAI 语义约定）而非日志。** 本次复活否决：分支实现的日志映射已经过评审、形态可交付；span 模型对可 fork、可中断的会话有损，留给将来真正有 span 查询需求的消费方。 -**handoff 游标未存活时全量回放日志（重新导出构造函数种子）。** 首轮复活曾交付此方案，其后收窄：接管操作现在从会话的构造边界起回放（`Session.firstLiveSeq`，即构造函数种子长度，这一事实会话早已校验过却未曾暴露；`header.seedLength` 不能胜任：它是持久保存的 fork 谱系（lineage）值，而恢复会话的构造函数种子是其完整的已存储日志）。恢复会话的历史已由上一个进程以同一 id 发出，fork 继承的前缀也已在父会话的流中发出；再次导出任何一者，都会让每次恢复为其完整历史重复付费，并在没有原生摄取去重的 OTLP 后端上使查询时的计数翻倍。接收端基于 `session.parent_id` + `session.seed_length` 拼接 fork 谱系。此次收窄放弃的内容与至多一次立场一致：恢复不再回填上一个进程未能投递的记录（彼时遥测未挂载，或崩溃时仍在队列中）——这本是全量回放唯一的真实收益，代价却由常见情形承担。提出回填要求的部署需要的是上文已推迟的 outbox，而不是回放。该边界同样吞掉 `SessionPersistence.load()` 修复被崩溃打断的日志时写入的合成轮次关闭事件（它们落在 `firstLiveSeq` 之前，尽管在上一个进程中从未存在过）。这是有意为之，而非附带效果：远端轮次的真实尾部记录已随崩溃进程的队列一同消亡，导出合成关闭事件无法补全该轮次，只会让一个未完成的轮次看起来已经关闭。导出的流忠实于崩溃进程实际发出的内容；接收端会把恢复后的流中一个从未关闭的轮次读作「上一个进程死在了该轮次之内」（OTel README 陈述了这条规则），其后干净的 `shutdown` 标记也只证明恢复后进程自身的退出。若为让修复以实时事件的身份导出而将修复前边界贯穿 load/prepare 传递，将使三个包相互耦合，只为抹除这一信号。 +**handoff 游标未存活时全量回放日志（重新导出构造函数种子）。** 首轮复活曾交付此方案，其后收窄：接管操作现在从会话的构造边界起回放（`Session.firstLiveSeq`，即构造函数种子长度；`Session.inheritedEventCount` 不能胜任，因为它是持久保存的 fork 谱系 cut，而恢复会话的构造函数种子是其完整的已存储日志）。恢复会话的历史已由上一个进程以同一 id 发出，fork 继承的前缀也已在父会话的流中发出；再次导出任何一者，都会让每次恢复为其完整历史重复付费，并在没有原生摄取去重的 OTLP 后端上使查询时的计数翻倍。接收端基于 `session.parent_id` + `session.seed_length` 拼接 fork 谱系。此次收窄放弃的内容与至多一次立场一致：恢复不再回填上一个进程未能投递的记录（彼时遥测未挂载，或崩溃时仍在队列中）——这本是全量回放唯一的真实收益，代价却由常见情形承担。提出回填要求的部署需要的是上文已推迟的 outbox，而不是回放。该边界同样吞掉 `SessionPersistence.load()` 修复被崩溃打断的日志时写入的合成轮次关闭事件（它们落在 `firstLiveSeq` 之前，尽管在上一个进程中从未存在过）。这是有意为之，而非附带效果：远端轮次的真实尾部记录已随崩溃进程的队列一同消亡，导出合成关闭事件无法补全该轮次，只会让一个未完成的轮次看起来已经关闭。导出的流忠实于崩溃进程实际发出的内容；接收端会把恢复后的流中一个从未关闭的轮次读作「上一个进程死在了该轮次之内」（OTel README 陈述了这条规则），其后干净的 `shutdown` 标记也只证明恢复后进程自身的退出。若为让修复以实时事件的身份导出而将修复前边界贯穿 load/prepare 传递，将使三个包相互耦合，只为抹除这一信号。 **将 seam 的轮次边界 `flush()` 提示转发到 OTel 提供方的 `forceFlush()`。** 首轮复活曾交付此转发，其后移除：三条不同的静默丢失路径共用同一份包装层状态——dispose 与进行中的 flush 之间的竞态（SDK 的并发 flush 防护会令 shutdown 的内部排空被跳过）、相互重叠的提示顶掉留存的 promise、以及提供方固定的 30 秒 flush 超时在批处理器仍在排空时便 reject。这些路径存在的唯一原因，是该转发让这个后端成为进程内第二个执行 flush 的组件，面对的还是上游实验性（experimental）源码树中未见诸文档的 SDK 内部行为；不实现 `flush()` 时，批处理器就是唯一执行 flush 的组件，其 `scheduledDelayMillis`（已可由部署方经 `processor` passthrough 调优）决定导出节奏，`shutdown()` 的排空从构造上就是完整的。仅当某个部署提出 `scheduledDelayMillis` 无法满足的轮次边界延迟要求时才恢复此转发——且届时应调用留存的 `BatchLogRecordProcessor` 自身的 `forceFlush()`，绝不调用提供方那个带超时包装的版本。 diff --git a/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.i18n.yaml b/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.i18n.yaml index 8a4c76f227..ac6cdbcd01 100644 --- a/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.md -2026-07-25-subagent-policy-inheritance.md: 34751a4e29e48c84d37425857b8b1b56c8d866eb -2026-07-25-subagent-policy-inheritance.zh.md: 3d5a6c7a88579583a0ab027c52dd3063eed0a8b2 +2026-07-25-subagent-policy-inheritance.md: 6df9ad53f8588018fca53ebae3a9dfba0d4382de +2026-07-25-subagent-policy-inheritance.zh.md: f3f7328d558ea55379cb95e53bb010f9f2d2c792 diff --git a/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.md b/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.md index 34751a4e29..6df9ad53f8 100644 --- a/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.md +++ b/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.md @@ -12,7 +12,7 @@ Sandbox and approval overrides are per-session log folds. An in-process subagent The delegation boundary snapshots `sandboxPolicy.overrideOf(parent.session)` before its first await, through the shared child-agent helpers (`captureDelegatedPolicyOverrides`/`appendDelegatedPolicyOverrides` in `dsh-subagent`), which the one-shot driver and the [continuable start](2026-08-10-continuable-subagent-policy-inheritance.md) both call. A later parent switch belongs to the parent's future; cancel-and-redelegate takes a new snapshot. The sandbox-policy service is optional, and only the explicit session override is copied, never deployment defaults or one-shot grants. The approval policy is not inherited: the same capture pins every child to `'never'` — the [approvals-pinned decision](2026-08-10-subagent-approval-pinned-never.md) supersedes this note's original approval-override inheritance. -Each captured value becomes a source-tagged `sandbox/mode` or `approval/policy` event appended during the child factory's unpublished setup. The session constructor has already fixed `Session.firstLiveSeq` at the fork-prefix length, so the inherited facts follow fork history, reach telemetry when the child is announced, and leave `SessionHeader.seedLength` at the prefix length. Existing last-event-wins folds therefore make the delegation snapshot beat stale fork history and let a later child switch beat the snapshot. A grandchild folds its parent's logged state, so the rule composes without another inheritance mechanism. +Each captured value becomes a source-tagged `sandbox/mode` or `approval/policy` event appended during the child factory's unpublished setup. The session constructor has already fixed `Session.firstLiveSeq` after the constructor seed, while `Session.inheritedEventCount` keeps the exact fork-prefix length, so the inherited facts follow fork history and reach telemetry when the child is announced without changing its lineage cut. Existing last-event-wins folds therefore make the delegation snapshot beat stale fork history and let a later child switch beat the snapshot. A grandchild folds its parent's logged state, so the rule composes without another inheritance mechanism. Ordinary session appends validate the inherited events before publication, and persistence captures the complete unpublished log when the session is announced. Any materialized child log therefore stores the inherited events with its first batch; there is no second policy store, schema field, or query index. The `source: 'delegation'` marker lets approval narration distinguish inheritance from a child-side user switch. diff --git a/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.zh.md b/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.zh.md index 3d5a6c7a88..f3f7328d55 100644 --- a/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.zh.md +++ b/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.zh.md @@ -12,7 +12,7 @@ Status: implemented 委派边界在第一次 await 之前，经由共享的子 agent 辅助函数（`dsh-subagent` 中的 `captureDelegatedPolicyOverrides`／`appendDelegatedPolicyOverrides`）对 `sandboxPolicy.overrideOf(parent.session)` 获取快照；一次性驱动器与[可继续启动](2026-08-10-continuable-subagent-policy-inheritance.zh.md)都会调用这些辅助函数。父级后续的切换属于父级的未来；取消后重新委派会取得新快照。沙箱策略服务为可选，仅复制显式会话覆盖项，绝不复制部署默认值或一次性授权。审批策略不继承：同一次捕获会把每个子 agent 钉定为 `'never'`——[审批钉定决策](2026-08-10-subagent-approval-pinned-never.zh.md)取代了本 note 原先的审批覆盖项继承。 -每个捕获值都会成为子 agent 工厂在未发布设置阶段追加的一条带来源标记的 `sandbox/mode` 或 `approval/policy` 事件。会话构造函数已将 `Session.firstLiveSeq` 固定为 fork 前缀的长度，因此继承事实会排在 fork 历史之后，在子 agent 公布时进入遥测，同时让 `SessionHeader.seedLength` 保持为此前缀的长度。因此，既有的末事件胜出折叠会让委派快照压过陈旧的 fork 历史，并让子 agent 后续的切换压过该快照。孙代 agent 会折叠其父级已记录的状态，因此无需另一套继承机制即可组合此规则。 +每个捕获值都会成为子 agent 工厂在未发布设置阶段追加的一条带来源标记的 `sandbox/mode` 或 `approval/policy` 事件。会话构造函数已把 `Session.firstLiveSeq` 固定在 constructor seed 之后，而 `Session.inheritedEventCount` 保留精确的 fork 前缀长度，因此继承事实会排在 fork 历史之后，并在子 agent 公布时进入遥测，却不改变其谱系 cut。因此，既有的末事件胜出折叠会让委派快照压过陈旧的 fork 历史，并让子 agent 后续的切换压过该快照。孙代 agent 会折叠其父级已记录的状态，因此无需另一套继承机制即可组合此规则。 普通的会话追加会在发布前校验继承事件，持久化层则在会话公布时捕获完整的未发布日志。因此，任何已物化的子 agent 日志都会在首批数据中存下继承事件；不存在第二套策略存储、schema 字段或查询索引。`source: 'delegation'` 标记让审批叙述能够区分继承与子 agent 侧的用户切换。 diff --git a/.agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.i18n.yaml b/.agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.i18n.yaml index 55a7d77f40..dbc7dcafd1 100644 --- a/.agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.md -2026-07-28-continuable-subagent-conversations.md: f456bacbf775bf914b47051e19639811e2385f65 -2026-07-28-continuable-subagent-conversations.zh.md: ea6df53026b6a76d8e100fa4fef0aebcd8fcd7cf +2026-07-28-continuable-subagent-conversations.md: 1efb18ab69cfe779a7de85f288a6e3aa7caaca83 +2026-07-28-continuable-subagent-conversations.zh.md: 7cb1b4a4cff1fc7f6fc41662d05693917084b63a diff --git a/.agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.md b/.agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.md index f456bacbf7..1efb18ab69 100644 --- a/.agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.md +++ b/.agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.md @@ -44,7 +44,7 @@ Cold resume does not dispatch through a subagent provider. The continuation mana `SubagentProvider.start()` and `SubagentRun` remain exclusively on the unchanged one-shot path. A continuable Activation directly owns its `AgentHandle` and never creates, wraps, or retains a `SubagentRun`; `SubagentRun.steer?()` is therefore absent. -`ctx.subagents.followup(parent, childId, content, { source, signal })` remains the sole parent-to-child continuation-message operation. The exact live parent Agent authorizes delivery; cold resume checks that authority before reconstruction and every path checks it again in the final no-await inbox-admission span, so a parent unregistered or replaced during materialization cannot authorize delivery. `source` records who supplied the admitted message and grants no authority. The model-facing `send_message` tool keeps only its stable `subagent_id` and `message` fields and always submits a follow-up turn. Both start and follow-up return the accepted `MessageId`, and neither reports how the manager materialized the Activation. +`ctx.subagents.sendMessage(sender, targetId, content, { signal })` is the sole model-authored continuation-message operation. The exact live sender authorizes delivery to its direct parent or direct continuable child; cold resume checks direct-child authority before reconstruction and every path checks again in the final no-await inbox-admission span, so an Agent unregistered or replaced during materialization cannot authorize delivery. The service derives durable `agent-message` provenance from that sender. The model-facing `send_message` tool keeps only `agent_id` and `message` and uses fixed Steer scheduling. Both start and send return the accepted `MessageId`, and neither reports how the manager materialized the Activation. For start and follow-up, the caller signal owns lookup, materialization, and admission only until inbox acceptance. After the operation returns its `MessageId`, the manager owns the Activation independently; later caller cancellation does not cancel the accepted turn or dispose the child. @@ -111,15 +111,13 @@ Top-level teardown is host-owned rather than represented as another Activation. The activation-owner scope exists because ordinary Cordis owner effects unwind in reverse registration order, which cannot express the dynamic child graph. Manager initialization registers the private scope's structural disposer first and its drain disposer afterward, so reverse unwind invokes the drain before releasing that scope; merely registering a cleanup effect on the same scope as later Agent handles would allow structural handle disposal to bypass child-first ordering. Each materialization registers its barrier participant and snapshots its exact live ancestry before starting the inner transaction, then remains tracked until it installs an Activation or fully rolls back. The Activation retains weak membership of that ancestry, so an intermediate Agent may leave the registry without hiding a still-live descendant from its host root. Each Activation installs one memoized disposal promise before cancellation or recursive callbacks, allowing scoped host shutdown, global manager unload, child release, and normal settlement to converge without double release. Cancellation propagates top-down before slow descendant cleanup; handle release remains child-first. Sibling branches drain independently; one disposal failure is recorded but does not prevent the manager from attempting the remaining selected handles, and the aggregate drain reports failure after all selected branches settle. Durable child Sessions survive this process-local teardown. -### Report delivery extension +### Adjacent-Agent messaging -The optional child-scoped `report(output)` tool was added later without changing Activation residency or adding another queue. It can be called zero or multiple times per turn, derives the live direct parent rather than accepting a recipient, and selects quiet injection or a waking parent follow-up through deployment config. The [report-tool Agent Note](2026-07-30-continuable-subagent-report-tool.md) owns its authority, acknowledgement, setup-contribution, and delivery contracts. +The shared `sendMessage(sender, targetId, content, options)` service operation adds no second queue. It accepts an exact live sender, permits only its direct parent or direct continuable child, and uses fixed Steer scheduling through the Agent inbox. The global `send_message({ agent_id, message })` tool exposes that same operation in both directions; the child's initial task identifies its direct parent when the tool is visible. The [adjacent-Agent messaging Agent Note](../architecture/2026-08-27-adjacent-agent-steer-messaging.md) owns its schema, authority, attribution, and prompt placement. -### Deferred steering +### Fixed Steer scheduling -This version exposes no subagent steering operation. Parent continuation messages always open later FIFO turns, so the continuation layer stores no current-turn controller and adds no controller-aware Agent admission contract. - -A later host UI may expose separate **Steer** and **Follow up** actions. Host steering would be strict and live-only: it may call the existing Agent steering path only while the Activation accepts a next step, must reject otherwise, and must never fall back to queueing or cold resume. Exposing parent steering to a model-facing tool remains a separate design. +Every accepted Agent message uses `Agent.steer()`. A running target claims it at the nearest step boundary; an idle or cold-resumed target starts a turn. The continuation layer does not expose a caller-selectable quiet, next-turn, or follow-up mode. ### Authority and recorded sender identity diff --git a/.agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.zh.md b/.agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.zh.md index ea6df53026..7cb1b4a4cf 100644 --- a/.agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.zh.md +++ b/.agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.zh.md @@ -44,7 +44,7 @@ inbox 接受消息前发生任何失败，操作都会在不返回任何 id 的 `SubagentProvider.start()` 和 `SubagentRun` 只保留在不变的 one-shot 路径上。可继续激活直接持有自身的 `AgentHandle`，绝不创建、包装或保留 `SubagentRun`；因此，`SubagentRun.steer?()` 不存在。 -`ctx.subagents.followup(parent, childId, content, { source, signal })` 仍是唯一的从 parent 到 child 的继续执行消息操作。确切的在线 parent Agent 授权投递；冷恢复会在重建前检查该权限，每条路径还会在最终无 await 的 inbox 准入区间再次检查，因此在物化期间被注销或替换的 parent 无法授权投递。`source` 记录谁提供了获准消息，不赋予任何权限。面向模型的 `send_message` 工具只保留稳定的 `subagent_id` 和 `message` 字段，并始终提交一个 follow-up 轮次。start 和 follow-up 都返回已接受的 `MessageId`，两者都不报告管理器如何物化激活。 +`ctx.subagents.sendMessage(sender, targetId, content, { signal })` 是唯一由模型编写的继续执行消息操作。确切在线 sender 授权向其直接 parent 或直接可继续 child 投递；冷恢复会在重建前检查直接 child 权限，每条路径还会在最终无 await 的 inbox 准入区间再次检查，因此在物化期间被注销或替换的 Agent 无法授权投递。服务从该 sender 推导持久化 `agent-message` 来源信息。面向模型的 `send_message` 工具只保留 `agent_id` 和 `message`，并使用固定 Steer 调度。start 与 send 都返回已接受的 `MessageId`，两者都不报告管理器如何物化 Activation。 对于 start 和 follow-up，调用方 signal 只在 inbox 接受消息前持有查找、物化和准入。操作返回 `MessageId` 后，管理器会独立持有该激活；调用方之后的取消不会取消已接受的轮次，也不会 dispose child。 @@ -111,15 +111,13 @@ Agent inbox 是唯一队列。每条继续执行消息都使用 `Agent.followup( activation-owner 作用域之所以存在，是因为普通 Cordis owner effect 按注册逆序撤销，无法表达动态 child 图。管理器初始化时先注册私有作用域的结构化 disposer，再注册自身的 drain disposer，使逆序撤销先执行 drain、再释放该作用域；如果只在与后续 Agent handle 相同的作用域上注册 cleanup effect，结构化 handle dispose 就可能绕过 child-first 顺序。每个物化过程都会在启动内部事务前注册其屏障参与项，并对其确切的在线祖先建立快照，然后保持跟踪，直到安装 Activation 或完全回滚。Activation 会保留其在这组祖先中的弱成员关系，因此中间 Agent 即使离开注册表，也不会让仍在线的后代脱离宿主根节点的可见范围。每个 Activation 都会在取消或递归回调前安装一个记忆化的 dispose promise，使限定作用域的宿主关闭、全局管理器卸载、child 释放和正常结算能够汇合，而不会重复释放。取消会在等待缓慢的后代清理之前自顶向下传播；handle 释放仍是 child-first。同级分支独立 drain；系统会记录单次 dispose 失败，但仍会尝试其余选中 handle，聚合 drain 则在所有选中分支结算后报告失败。这次进程内拆卸不会销毁持久化 child 会话。 -### 报告投递扩展 +### 相邻 Agent 消息 -后来添加的可选 child 作用域 `report(output)` 工具不会改变 Activation 驻留状态，也不会增加另一条队列。它每轮可调用零次或多次，不允许指定接收方，而是推导在线的直接 parent；投递采用静默注入还是唤醒 parent follow-up，由部署配置选择。[report 工具 Agent Note](2026-07-30-continuable-subagent-report-tool.zh.md)规定其权限、确认、设置贡献和投递约定。 +共享的 `sendMessage(sender, targetId, content, options)` 服务操作不会增加第二条队列。它接收确切在线 sender，只允许其直接 parent 或直接可继续 child，并通过 Agent inbox 使用固定 Steer 调度。全局 `send_message({ agent_id, message })` 工具在两个方向暴露同一个操作；当 child 可以看到该工具时，其初始任务会标明直接 parent。[相邻 Agent 消息 Agent Note](../architecture/2026-08-27-adjacent-agent-steer-messaging.zh.md)规定其 schema、权限、来源信息与提示词位置。 -### 延后的 steering（中途引导） +### 固定 Steer 调度 -本版本不暴露 subagent steering 操作。parent 的继续执行消息始终开启后续 FIFO 轮次，因此继续执行层不存储当前轮次控制方，也不新增能够感知控制方的 Agent 准入约定。 - -后续宿主 UI 可以分别暴露 **Steer** 和 **Follow up** 操作。宿主 steering 必须严格且仅限在线使用：只有当激活接受下一步骤时，它才能调用现有的 Agent steering 路径；其他情况必须拒绝，而且绝不能转为排队或冷恢复。是否通过面向模型的工具暴露 parent steering 仍需单独设计。 +每条已接受的 Agent 消息都使用 `Agent.steer()`。运行中的目标会在最近的 step 边界领取消息；空闲或冷恢复的目标会启动一个轮次。继续执行层不暴露由调用方选择的 quiet、next-turn 或 follow-up 模式。 ### 权限与已记录的发送方身份 diff --git a/.agents/notes/implemented/feature/2026-07-30-continuable-subagent-report-tool.md b/.agents/notes/implemented/feature/2026-07-30-continuable-subagent-report-tool.md deleted file mode 100644 index 07d17f18f3..0000000000 --- a/.agents/notes/implemented/feature/2026-07-30-continuable-subagent-report-tool.md +++ /dev/null @@ -1,118 +0,0 @@ -# Agent Note: Continuable subagent report tool - -Status: implemented - -English | [中文](2026-07-30-continuable-subagent-report-tool.zh.md) - -## Problem - -Continuable in-process subagents can receive later parent messages, retain descendants, settle, and cold-resume, but the base lifecycle gives them no way to send selected content back to their direct parent. Their complete output already remains reconstructable from the durable child Session, so the missing capability is explicit delivery rather than result storage. - -Treating every final assistant message as an implicit result would conflate turn completion with reporting. A long-lived child may have nothing useful to report in one turn, may report progress several times in another, and must remain available after reporting. Recipient authority, quiet versus next-step delivery, acknowledgement, durability, and retry behavior therefore need one explicit contract. - -## Decision - -Add the independently installed `@deepseek-ai/dsh-tool-subagent-report` package. It contributes an ordinary model-facing `report` tool to each continuable in-process child Activation. The mechanism accepts zero or multiple calls in a turn; the child is separately instructed to call it once before finishing ([the report obligation](2026-08-06-continuable-child-report-obligation.md)). Success neither concludes the turn, settles the Activation, nor prevents later parent follow-ups, and finishing a turn never reports automatically. - -The feature is a collaboration control, not a result-bearing execution wrapper. It adds no Task, `SubagentRun`, result promise, Activation state, delivery queue, or replay path. - -### Model-facing contract - -`report` accepts exactly `{ output: string }` and returns exactly `{ messageId: string }`. It accepts no child id, recipient id, or delivery mode. `exec.agent` binds the tool call to the reporting child, the service derives the sole recipient from durable `parentSession`, and deployment config owns scheduling. - -`messageId` is the stable `MessageId` of the user-role message accepted into the parent's inbox. It is not a read receipt, parent-log acknowledgement, turn-completion receipt, or persistence flush. - -The description states that reporting is required before finishing, repeatable, direct-parent-only, and non-terminal. It warns that a failed tool result may still follow an accepted send because a later `tools/post-execute` failure can replace the result. Without an idempotency key, stronger wording would encourage duplicate retries after ambiguous failure. - -The tool uses generic rendering with no locations. Its acknowledgement includes `messageId`. Scope-local registration keeps presentation and execution aligned: roots, one-shot children, remote providers, sibling scopes, and agentless execution neither see nor execute `report`. It installs after the child's global `toolFilter`, so a delegation allow-list cannot accidentally remove the structural return channel; deployments that require no return channel omit the package. - -### Service authority - -The subagent seam exposes `ctx.subagents.reportFrom(child, content, { delivery, signal }): Promise`. The exact live child Agent is the sender credential. The continuation manager accepts only an Activation whose `handle.agent === child`, derives its direct parent from the child's durable header, and requires that id to resolve to a live parent Agent in the final synchronous authorization-and-send span. The API accepts no caller-selected recipient, ancestor, or sender fields. - -Roots, one-shot children, forged objects, stale Agents, and same-id replacements fail with `UNAUTHORIZED`. A closing child Activation fails with `ACTIVATION_CLOSING`; manager drain and pre-acceptance cancellation retain their existing lifecycle errors. A missing or send-rejecting direct parent fails with `PARENT_UNAVAILABLE` and `direct parent is not live; report was not delivered`. Failure returns no id, cold-resumes no parent, writes no offline mailbox, and mutates no absent-parent Session. - -Nested reporting crosses exactly one edge. A grandchild reports to its direct child parent, never to the top-level coordinator. That intermediate child may explicitly report a derived update later. - -### Delivery policy - -The package validates `reportDelivery: 'quiet' | 'next-step'`; the default is `next-step` ([ordering decision](../bug-fix/2026-08-17-subagent-report-settlement-ordering.md)). - -Quiet delivery calls `parent.inject()`. It adds model-visible next-step context without waking an idle parent; a running parent stages the report for the next safe log position. - -Next-step delivery calls `parent.steer()`. It wakes a parked parent and joins a running parent's nearest step boundary. When that parent is itself a continuable Activation, the send uses the manager's existing admission accounting so the parent cannot settle between synchronous inbox insertion and the admission microtask. Reports share the next-step FIFO with a later settlement notice, preserving their accepted causal order. - -Both modes frame one user-role message as `Background subagent reported:` followed by the exact `output`. The durable message source is `{ kind: 'subagent-report', senderSessionId: child.id }`. Normal Agent ordering governs concurrent sends; the subagent layer creates no second queue. - -### Acknowledgement and recovery - -Success means the exact live parent synchronously accepted the message. The context becomes reconstructable only when it reaches its normal log boundary; a next-step delivery has woken the parent, while quiet delivery may remain pending. The inbox message id remains separate from the returned stable message id. - -The first version provides no durable mailbox, idempotency key, delivery receipt, retry protocol, or exactly-once claim. A process failure can leave the caller uncertain, and retry after an unknown outcome may duplicate a report. The durable child transcript remains the recovery source when the parent is unavailable. - -### Composition and lifecycle - -The subagent seam adds `registerContinuableSetup(contribution): () => void`, backed by `SubagentActivationSetupRegistry`. Each synchronous contribution receives the unpublished child context and returns the disposer for its installation. The continuation manager first applies base child composition, then current contributions in registration order through the same setup closure used for fresh creation and cold resume. - -The registry owns registration, per-child installation records, setup rollback, child-scope cleanup, and immediate revocation. Applying a batch returns the Agent setup commit that revalidates provisioning after every setup await and immediately before Agent publication. A throwing or concurrently revoked contribution therefore rejects before either Agent or Session publication and rolls back the batch. New registrations affect a resident child only on its next Activation; removing a registration first closes it to new setup and then revokes every provisioning or resident installation immediately. Registration disposal and child-context disposal are idempotent and attempt every release before aggregating failures. - -This seam keeps the continuation manager unaware of tool names. The report package installs only `report` and its child-scoped guidance section; `@deepseek-ai/dsh-tool-subagent-control` independently installs parent-side `send_message` and `list_agents`. A deployment can install either direction, both, or neither. Providers remain data-only, durable descriptors do not snapshot report availability or delivery mode, and cold resume uses the deployment's current contributions and policy. - -### Snapshot coverage - -The ACP snapshot harness adds `waitForSubagentTurnEnd`, selecting the Nth harvested child by the same order as `session.N.jsonl`. It waits for a closed child turn containing a request header so a continuable child's earlier descriptor-seed turn cannot satisfy the boundary. This lets the assembled scenario wait for the child-side report without inventing a parent-visible signal. - -The authored snapshot starts a continuable child, executes the real scope-local `report` tool, and observes default next-step delivery before the manager's later settlement notice. A snapshot-only maintenance fence holds the parent until both messages are pending, proving next-step input is claimed before queued next-turn input when the parent resumes. It declares child pins `1`, so the otherwise non-global `report` schema and the child's own prompt are checked against `tool-schemas.1.expected.json` and `system-prompt.1.expected.md` while the root keeps the class pins. The generated tool catalog separately mints a child scope to include the same scope-local schema. - -## Alternatives considered - -### Automatically deliver every final answer - -Automatic delivery cannot represent zero reports, progress reports, or several selected updates. It also couples reporting to settlement and can duplicate content already reported explicitly. - -### Always wake the parent - -Waking on every report creates unsolicited turns and can cascade through nested subagents. Quiet delivery was chosen as the default on the assumption that the parent had another reason to read its context. [The report obligation](2026-08-06-continuable-child-report-obligation.md) supersedes that choice: a parked background coordinator has no such reason, so waking is the default and this paragraph now records why `quiet` still exists. - -### Let the child choose the delivery mode - -Giving the model a mode argument grants it control over scheduler pressure and makes behavior deployment-dependent. The child chooses content and timing; deployment config chooses whether that content wakes the parent. - -### Register a global tool - -A global `report` would advertise an unusable capability to roots, one-shot children, remote children, and agentless callers. Execution-time rejection would make schema visibility disagree with authority. - -### Combine both directions in the control package - -`send_message` and `report` have different audiences, scopes, configuration, and lifecycle. Independent packages let deployments grant either direction without implying the other. - -### Persist an offline parent mailbox - -Mutating or cold-resuming an absent parent requires a new durable addressing, authorization, conflict, acknowledgement, and replay protocol. Requiring a live direct parent keeps the first version on the existing Agent send path. - -### Reintroduce a Task or result promise - -A result-bearing wrapper makes one report or one turn appear terminal and recreates the lifetime mismatch that continuable Activations removed. Explicit repeatable sends need no intermediate execution object. - -### Validate setup after Agent creation - -A post-creation revocation check can reject the Activation only after the Agent and Session have been published. Disposing the returned handle removes the live objects but cannot delete persistence through the current seam, leaving a resumable child that the continuation manager said was never established. Returning an `AgentSetupCommit` instead lets the Agent factory perform the same mutable-state check synchronously at its publication boundary. - -## Consequences - -- A continuable in-process child exposes exactly one scope-local `report` schema only while the report package's contribution is installed; unrelated Agents never expose it. -- The tool returns the parent message's stable `MessageId`; its inbox occurrence is not a separate public identity. -- Only the exact resident child may report, and only to the exact live direct parent derived from durable lineage. The service has no recipient parameter or offline fallback. -- Next-step delivery is the validated default: it wakes an idle parent or extends a running parent's turn at the nearest step boundary. Quiet delivery never wakes an idle parent. -- Child cancellation or disposal after parent acceptance does not retract the report. Before acceptance, child disposal, drain, parent loss, or caller cancellation rejects the operation. -- Fresh and resumed Activations compose current setup contributions before publication. Grants wait for the next Activation; revocation is immediate for resident children. -- Unit coverage pins visibility, allow-list behavior, both delivery modes, stable message and sender identities, nested routing, invalid senders, absent parents, cancellation, drain, revocation races, and the absence of Jobs or implicit final reporting. -- The keyless assembled snapshot proves the real child tool, default next-step ordering before settlement, and durable parent framing. - -### Accepted risks - -The acceptance boundary is weaker than durable end-to-end delivery. A crash can leave the result ambiguous, and retries may duplicate reports. - -Next-step delivery can amplify model work when nested children report frequently. Reports waiting together share one step, and deployment ownership through `reportDelivery` bounds but does not remove that risk. - -Registry presence is the parent liveness signal. A host-owned parent whose `AgentHandle.dispose()` has started but has not yet unwound its scope can still accept and append a report that it will not act on in this process. Closing that gap requires an Agent-level disposal-start signal rather than subagent-layer inference. diff --git a/.agents/notes/implemented/feature/2026-07-30-continuable-subagent-report-tool.zh.md b/.agents/notes/implemented/feature/2026-07-30-continuable-subagent-report-tool.zh.md deleted file mode 100644 index 9cbf76a9f3..0000000000 --- a/.agents/notes/implemented/feature/2026-07-30-continuable-subagent-report-tool.zh.md +++ /dev/null @@ -1,118 +0,0 @@ -# Agent Note: 可继续 subagent 报告工具 - -Status: implemented - -[English](2026-07-30-continuable-subagent-report-tool.md) | 中文 - -## 问题 - -可继续的进程内 subagent 能够接收 parent 后续发来的消息、保留后代、结算并冷恢复，但基础生命周期无法让它们将选中内容发送给直接 parent。child 的完整输出已可从持久化会话中重建，因此缺失的能力是显式投递，而非结果存储。 - -如果将每条 assistant 最终消息都视为隐式结果，就会混淆轮次完成与报告。长期运行的 child 可能在某个轮次中无内容可报告，也可能在另一个轮次多次报告进展，而且报告后必须仍可继续工作。因此，接收方权限、静默投递与 next-step 投递、确认、持久性和重试行为都需要一份显式约定。 - -## 决策 - -新增可独立安装的 `@deepseek-ai/dsh-tool-subagent-report` 包。它会向每个可继续进程内 child Activation 贡献一个普通的面向模型 `report` 工具。机制本身接受一个轮次中调用零次或多次；child 会另行被要求在结束前调用一次（见[报告义务](2026-08-06-continuable-child-report-obligation.zh.md)）。调用成功既不会结束该轮次或结算 Activation，也不会阻止 parent 之后继续 follow-up；完成轮次也绝不会自动报告。 - -该功能是协作控制，不是承载结果的执行包装层。它不新增 Task、`SubagentRun`、结果 promise、Activation 状态、投递队列或回放路径。 - -### 面向模型的约定 - -`report` 只接受 `{ output: string }`，也只返回 `{ messageId: string }`。它不接受 child id、接收方 id 或投递模式。`exec.agent` 将工具调用绑定到发送报告的 child；服务从持久化 `parentSession` 中推导唯一接收方，调度则由部署配置决定。 - -`messageId` 是已接受进入 parent inbox 的用户角色消息所对应的稳定 `MessageId`。它不是已读回执、parent 日志确认、轮次完成回执或持久化 flush。 - -工具描述会明确报告操作在结束前必须执行、可重复、仅限直接 parent 且不会结束轮次。它还会警告：发送被接受后，后续 `tools/post-execute` 失败可能替换工具结果，因此工具结果失败时内容仍可能已经送达。没有幂等键时，更强的表述会诱导调用方在结果不明确的失败后重复重试。 - -该工具使用不带 location 的通用渲染，其确认中包含 `messageId`。作用域局部注册使呈现与执行保持一致：root、one-shot child、远程提供方、同级作用域和无 agent（智能体）执行既不能看到，也不能执行 `report`。它会在 child 的全局 `toolFilter` 之后安装，因此委派 allow-list 不会意外移除这条结构性返回通道；不需要返回通道的部署不安装该包。 - -### 服务权限 - -subagent seam 暴露 `ctx.subagents.reportFrom(child, content, { delivery, signal }): Promise`。确切的在线 child Agent 是发送方凭据。继续执行管理器只接受 `handle.agent === child` 的 Activation，从 child 的持久化 header 中推导其直接 parent，并要求该 id 在最终的同步授权与发送区间解析为一个在线 parent Agent。该 API 不接受由调用方选择的接收方、祖先或发送方字段。 - -root、one-shot child、伪造对象、陈旧 Agent 和同 id 替换对象都以 `UNAUTHORIZED` 失败。正在关闭的 child Activation 以 `ACTIVATION_CLOSING` 失败；管理器 drain 和接受前取消保留既有的生命周期错误。直接 parent 不存在或拒绝接受时，以 `PARENT_UNAVAILABLE` 和 `direct parent is not live; report was not delivered` 失败。失败不返回 id，不冷恢复 parent，不写入离线邮箱，也不会修改缺失 parent 的会话。 - -嵌套报告恰好跨越一条边。grandchild 会向其直接 child parent 报告，绝不会直接向顶层 coordinator 报告。中间 child 可以稍后显式报告自己归纳的更新。 - -### 投递策略 - -该包会校验 `reportDelivery: 'quiet' | 'next-step'`，默认值为 `next-step`（见[顺序决策](../bug-fix/2026-08-17-subagent-report-settlement-ordering.zh.md)）。 - -静默投递调用 `parent.inject()`。它会添加模型可见的 next-step 上下文，但不唤醒空闲 parent；运行中的 parent 会把报告暂存到下一个安全日志位置。 - -Next-step 投递调用 `parent.steer()`。它会唤醒停驻的 parent，并加入运行中 parent 最近的 step 边界。当该 parent 本身也是可继续 Activation 时，发送会使用管理器现有的准入记账，防止 parent 在同步插入 inbox 与准入微任务之间结算。报告与稍后的结算通知共享 next-step FIFO，从而保持其被接受时的因果顺序。 - -两种模式都会将一条用户角色消息封装为 `Background subagent reported:`，后面跟随完全原样的 `output`。持久化消息来源为 `{ kind: 'subagent-report', senderSessionId: child.id }`。并发发送的顺序由 Agent 的常规规则决定；subagent 层不会创建第二条队列。 - -### 确认与恢复 - -成功表示确切的在线 parent 已同步接受该消息。上下文只有到达正常日志边界后才可重建；next-step 投递已经唤醒 parent，而静默投递可能继续等待。inbox 消息 id 不会成为另一个公开身份。 - -首个版本不提供持久化邮箱、幂等键、投递回执、重试协议或恰好一次保证。进程故障可能让调用方无法确定结果，在结果未知时重试则可能重复报告。parent 不可用时，持久化 child transcript（文本记录）仍是恢复来源。 - -### 组合与生命周期 - -subagent seam 新增 `registerContinuableSetup(contribution): () => void`，由 `SubagentActivationSetupRegistry` 支撑。每个同步贡献都会接收尚未发布的 child 上下文，并返回其安装的 disposer。继续执行管理器首先应用基础 child 组合，然后通过同一个用于首次创建与冷恢复的设置闭包，按注册顺序应用当前贡献。 - -注册表负责注册、每个 child 的安装记录、设置回滚、child 作用域清理和立即撤销。应用一个批次会返回 Agent setup 提交对象，用于在每次 setup 的 await 结算后以及紧邻 Agent 发布前重新校验配置状态。因此，某项贡献抛出异常或被并发撤销时，会在 Agent 与会话发布前拒绝操作并回滚该批次。新注册项只会在驻留 child 的下一个 Activation 生效；移除注册项时，会先将它对新设置关闭，再立即撤销为正在预配置或驻留的每个 child 安装的实例。注册 dispose（资源释放）与 child 上下文 dispose 都是幂等的，两者都会先尝试每项释放，再聚合失败。 - -该 seam 使继续执行管理器无需知道工具名。report 包只安装 `report` 及其 child 作用域指引 section；`@deepseek-ai/dsh-tool-subagent-control` 则独立安装 parent 侧的 `send_message` 和 `list_agents`。部署时可安装任一方向、同时安装两者或两者均不安装。提供方仍只负责数据，持久化描述符不会对 report 可用性或投递模式建立快照，冷恢复则使用部署当前的贡献与策略。 - -### 快照覆盖 - -ACP（Agent Client Protocol）快照 harness 新增 `waitForSubagentTurnEnd`，按与 `session.N.jsonl` 相同的顺序选择第 N 个已收集 child。它会等待一个包含请求 header 的已闭合 child 轮次，以防可继续 child 早期播种描述符的轮次错误满足该边界。这样，整体组装的场景无需伪造 parent 可见信号，就能等待 child 侧报告。 - -手写快照会启动一个可继续 child，执行真实的作用域局部 `report` 工具，并观察默认 next-step 投递先于管理器稍后的结算通知。一个仅用于快照的 maintenance 围栏会保持 parent，直至两条消息都处于待领取状态，从而证明 parent 恢复时先领取 next-step 输入、再领取排队的 next-turn 输入。它声明 child pin `1`，因此本不属于全局的 `report` schema 与该 child 自身的提示词会分别与 `tool-schemas.1.expected.json` 和 `system-prompt.1.expected.md` 比对，root 则继续使用类别 pin。生成的工具目录会另外铸造一个 child 作用域，以收录同一个作用域局部 schema。 - -## 曾考虑的替代方案 - -### 自动投递每个最终回答 - -自动投递无法表示零次报告、进展报告或多次精选更新。它还会将报告与结算耦合，并可能重复投递已显式报告的内容。 - -### 始终唤醒 parent - -每次报告都唤醒 parent 会产生未经请求的轮次，还可能沿嵌套 subagent 级联扩散。当初选择静默投递作为默认值，前提是 parent 还有别的理由去读自己的上下文。[报告义务](2026-08-06-continuable-child-report-obligation.zh.md)取代了该选择：已经停驻的后台协调者并没有这样的理由，因此唤醒成为默认值，而本段现在记录的是 `quiet` 为何仍然保留。 - -### 允许 child 选择投递模式 - -向模型提供 mode 参数会赋予其控制调度器压力的能力，并使行为依赖部署。child 只决定内容和时机；该内容是否唤醒 parent，由部署配置决定。 - -### 注册全局工具 - -全局 `report` 会向 root、one-shot child、远程 child 和无 agent 调用方公布一项无法使用的能力。到执行时才拒绝，会使 schema 可见性与权限不一致。 - -### 将两个方向合并到 control 包 - -`send_message` 与 `report` 的受众、作用域、配置和生命周期各不相同。独立的包可让部署授予任意一个方向，而不暗示也授予另一个方向。 - -### 持久化离线 parent 邮箱 - -修改或冷恢复不在线的 parent，需要一套新的持久化寻址、权限、冲突、确认和回放协议。要求直接 parent 在线，可以让首个版本继续使用现有 Agent 发送路径。 - -### 重新引入 Task 或结果 promise - -承载结果的包装层会让一次报告或一个轮次看似具有终止性，并重新引入可继续 Activation 已经移除的生命周期不匹配。显式、可重复的发送无需中间执行对象。 - -### 在 Agent 创建后校验 setup - -创建完成后的撤销检查只能在 Agent 与会话均已发布后拒绝 Activation。对返回的 handle 执行 dispose 会移除实时对象，但当前 seam 无法删除持久化内容，因此会留下一个仍可恢复的 child，而继续执行管理器却判定它从未建立。改为返回 `AgentSetupCommit`，Agent 工厂便可在自身的发布边界同步执行同一项可变状态检查。 - -## 影响 - -- 只有安装 report 包贡献时，可继续进程内 child 才会恰好暴露一个作用域局部 `report` schema；无关 Agent 永远不会暴露该 schema。 -- 工具返回 parent 消息的稳定 `MessageId`；其 inbox 中的出现不会成为另一个公开身份。 -- 只有确切的驻留 child 才能报告，且只能报告给根据持久化谱系推导的确切在线直接 parent。服务不接受接收方参数，也不提供离线 fallback。 -- Next-step 投递是校验后的默认模式：它会唤醒空闲 parent，或在最近的 step 边界延长运行中 parent 的轮次。静默投递绝不会唤醒空闲 parent。 -- parent 接受后取消或 dispose child 不会撤回报告。接受前，child dispose、drain、parent 丢失或调用方取消都会拒绝操作。 -- 新建和恢复的 Activation 都会在发布前组合当前设置贡献。新授权等待下一个 Activation 才生效，而已驻留 child 的授权撤销立即生效。 -- 单元覆盖固定可见性、allow-list 行为、两种投递模式、稳定的消息与发送方身份、嵌套路由、无效发送方、缺失的 parent、取消、drain、撤销竞争，以及不存在 Task 或隐式最终报告。 -- 无密钥整体组装快照证明真实 child 工具、默认 next-step 顺序先于结算，以及持久化 parent 封装。 - -### 已接受的风险 - -该接受边界弱于持久化端到端投递。崩溃可能导致结果不明，重试则可能重复报告。 - -嵌套 child 频繁报告时，next-step 投递可能放大模型工作量。一起等待的报告会共享一个 step，通过 `reportDelivery` 交由部署所有者控制也会限制该风险，但无法完全消除。 - -注册表中的存在性就是 parent 在线信号。宿主拥有的 parent 如果已开始 `AgentHandle.dispose()` 但尚未完成其作用域清理，仍可能接受并追加一条本进程不会再处理的报告。要弥合这个缺口，需要 Agent 层面的 dispose 开始信号，不能由 subagent 层推断。 diff --git a/.agents/notes/implemented/feature/2026-07-31-web-default-search.i18n.yaml b/.agents/notes/implemented/feature/2026-07-31-web-default-search.i18n.yaml index bc6f790d7c..70c7102ed5 100644 --- a/.agents/notes/implemented/feature/2026-07-31-web-default-search.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-31-web-default-search.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-31-web-default-search.md -2026-07-31-web-default-search.md: eb0de16b5bf6133bbdb5275106f42eba75ddf607 -2026-07-31-web-default-search.zh.md: e1cc622e70d8af8e71a8c60aa7e7ceb9a2b91a5e +2026-07-31-web-default-search.md: f196bfcd0c42bcfd6aacaac46971b4b9948732d7 +2026-07-31-web-default-search.zh.md: cd313c714acc22ca470ece681624bae19c1e8b4f diff --git a/.agents/notes/implemented/feature/2026-07-31-web-default-search.md b/.agents/notes/implemented/feature/2026-07-31-web-default-search.md index eb0de16b5b..f196bfcd0c 100644 --- a/.agents/notes/implemented/feature/2026-07-31-web-default-search.md +++ b/.agents/notes/implemented/feature/2026-07-31-web-default-search.md @@ -4,13 +4,15 @@ Status: implemented English | [中文](2026-07-31-web-default-search.zh.md) +The [shared-base Web fetch default](2026-09-01-shared-base-web-fetch-default.md) supersedes this record's fetch opt-in decision. This record remains authoritative for the default search provider, credential resolution, endpoint, timeout, and the separation between provider availability and model-tool registration. + ## Problem The harness had a complete Web capability family—provider registry, DeepSeek/Exa/Perplexity search providers, local fetch, stable model tools, and structured result presentation—but the shipped `dsh web` composition mounted none of it. The model could not discover current information unless a deployment supplied a custom overlay. Merely mounting the existing DeepSeek provider would not complete the WebUI path: the Models page stores `DEEPSEEK_API_KEY` through `ctx.credentials`, while the search provider froze only the process environment at plugin load, so a key entered or rotated in the running UI would not reach search. ## Decision -`apps/cli/config/base.cordis.yml` explicitly mounts `dsh-web` with `searchProvider: deepseek-official` and `fetchProvider: http`, `dsh-web-search-deepseek`, `dsh-web-fetch-http`, and `dsh-tool-web` with `fetch: false` and `searchTimeoutMs: 60000`. The shared base therefore keeps only `web_search` visible unless a product preset enables fetch; the shipped Web `cordis`, `code`, and `standard` presets do so. Explicit provider ids keep selection independent of registration order and leave personal or `--config` overlays able to replace or disable the rows. The one-minute shipped budget covers an auxiliary DeepSeek Messages request plus server-side retrieval while leaving `dsh-tool-web`'s provider-neutral 30-second default unchanged for custom compositions. The [Web capability seam decision](../architecture/2026-06-24-web-capability-seam.md) owns the public-fetch security policy and Web preset default. +`packages/bundle/base/cordis.patch.yml` explicitly mounts `dsh-web` with `searchProvider: deepseek-official` and `fetchProvider: http`, `dsh-web-search-deepseek`, `dsh-web-fetch-http`, and `dsh-tool-web` with `searchTimeoutMs: 60000`. The [shared-base Web fetch default](2026-09-01-shared-base-web-fetch-default.md) owns the current `fetch: true`; this record continues to own provider selection, search credentials, and timeout. Explicit provider ids keep selection independent of registration order and leave personal or `--patch` overlays able to replace or disable the rows. The one-minute shipped budget covers an auxiliary DeepSeek Messages request plus server-side retrieval while leaving `dsh-tool-web`'s provider-neutral 30-second default unchanged for custom compositions. The [Web capability seam decision](../architecture/2026-06-24-web-capability-seam.md) owns the public-fetch security policy. DeepSeek search uses the same `DEEPSEEK_API_KEY` credential reference as the official conversation adapter. The provider resolves that reference inside every search through the optional `ctx.credentials` service; only a composition without the seam falls back to the launching process environment, and a non-empty literal `apiKey` remains the programmatic last resort. A stored or rotated Web Models key therefore reaches the next search without restarting or retaining the value on the provider. Because `WebSearchProvider.available()` is synchronous, it treats an installed resolver as locally usable and missing dynamic credentials fail the operation with the provider-specific `WEB_PROVIDER_CREDENTIAL_MISSING` code while the stable tool schema stays registered. @@ -30,8 +32,8 @@ The default mount does not create a Web-specific permission policy. `web_search` **Raise `dsh-tool-web`'s provider-neutral timeout.** Rejected because custom providers and deployments own different latency expectations; the shipped DeepSeek composition owns this deployment budget. -**Enable fetch on every shared-base surface.** Rejected because the shared base serves products with different network postures. It mounts the public-only provider but keeps the tool opt-in; the shipped Web presets deliberately enable it, while another product can leave it hidden or add stricter network policy. +**Enable fetch on every shared-base surface.** This record rejected the alternative because shared-base products could require different network policies. The [shared-base Web fetch default](2026-09-01-shared-base-web-fetch-default.md) supersedes that rejection after the shipped products converged on one full tool roster; its public-destination and no-approval constraints remain current. ## Consequences -Native model requests on every shared-base surface carry the `web_search` schema and search guidance; Web/headless PTC mode exposes the same search capability beneath `run_code`. Search adds a complete auxiliary model call and may use the native server tool multiple times; its exact secret-free request remains reconstructable from the initiating session log. The shipped Web `cordis`, `ptc`, and `standard` presets additionally expose `web_fetch` with public-address enforcement and no per-call approval. The Web snapshot lane boots the shipped tree, drives a replayed `web_search` call through the real DeepSeek provider against a local Messages fixture, asserts the durable auxiliary request and structured result, and pins the settled browser presentation. Composition smokes pin the shared search roster and per-preset fetch choices; the built composition dump pins the one-minute shipped search budget; provider tests pin missing, stored, and rotated credential behavior plus literal and ambient compatibility. +Native model requests on headless, full SDK, ACP, and custom base-only profiles carry the `web_search` and `web_fetch` schemas and guidance; Web presets expose the same pair, including beneath `run_code` in PTC mode. Search adds a complete auxiliary model call and may use the native server tool multiple times; its exact secret-free request remains reconstructable from the initiating session log. Fetch enforces public addresses and requires no per-call approval. The Web snapshot lane boots the shipped tree, drives a replayed `web_search` call through the real DeepSeek provider against a local Messages fixture, asserts the durable auxiliary request and structured result, and pins the settled browser presentation. Shared snapshot headers pin the common fetch schema and prompt guidance. Composition smokes pin the tool roster; the built composition dump pins the one-minute shipped search budget; provider tests pin missing, stored, and rotated credential behavior plus literal and ambient compatibility. diff --git a/.agents/notes/implemented/feature/2026-07-31-web-default-search.zh.md b/.agents/notes/implemented/feature/2026-07-31-web-default-search.zh.md index e1cc622e70..cd313c714a 100644 --- a/.agents/notes/implemented/feature/2026-07-31-web-default-search.zh.md +++ b/.agents/notes/implemented/feature/2026-07-31-web-default-search.zh.md @@ -4,13 +4,15 @@ Status: implemented [English](2026-07-31-web-default-search.md) | 中文 +[共享 base 的 Web 抓取默认值](2026-09-01-shared-base-web-fetch-default.zh.md)取代本文关于抓取按需启用的决策。本文继续负责默认搜索提供方、凭据解析、端点、超时，以及提供方可用性与模型工具注册之间的区分。 + ## 问题 该 harness 已具备完整的 Web 能力体系：提供方注册表、DeepSeek、Exa 和 Perplexity 搜索提供方、本地抓取、稳定的面向模型工具，以及结构化结果呈现，但已交付的 `dsh web` 组合没有挂载其中任何一项。除非部署提供自定义覆盖层，否则模型无法发现最新信息。仅挂载现有 DeepSeek 提供方仍无法打通 WebUI 链路：Models 页面通过 `ctx.credentials` 存储 `DEEPSEEK_API_KEY`，而搜索提供方只会在插件加载时固定读取进程环境，因此在运行中的 UI 输入或轮换的密钥无法用于搜索。 ## 决策 -`apps/cli/config/base.cordis.yml` 明确挂载 `dsh-web`，配置 `searchProvider: deepseek-official` 与 `fetchProvider: http`，同时挂载 `dsh-web-search-deepseek`、`dsh-web-fetch-http`，并以 `fetch: false` 和 `searchTimeoutMs: 60000` 挂载 `dsh-tool-web`。因此，共享 base 只会暴露 `web_search`，除非产品 preset 启用抓取；已交付的 Web `cordis`、`ptc` 与 `standard` preset 会启用抓取。显式提供方 id 使选择不受注册顺序影响，同时个人覆盖层或 `--config` 覆盖层仍可替换或禁用这些配置项。已交付的一分钟预算用于覆盖一次辅助 DeepSeek Messages 请求及服务端检索，同时保持 `dsh-tool-web` 提供方无关的 30 秒默认值不变，以供自定义组合使用。[Web 能力 seam 决策](../architecture/2026-06-24-web-capability-seam.zh.md)负责公开抓取安全策略与 Web preset 默认值。 +`packages/bundle/base/cordis.patch.yml` 明确挂载 `dsh-web`，配置 `searchProvider: deepseek-official` 与 `fetchProvider: http`，同时挂载 `dsh-web-search-deepseek`、`dsh-web-fetch-http`，并以 `searchTimeoutMs: 60000` 挂载 `dsh-tool-web`。[共享 base 的 Web 抓取默认值](2026-09-01-shared-base-web-fetch-default.zh.md)负责当前的 `fetch: true`；本文继续负责提供方选择、搜索凭据与超时。显式提供方 id 使选择不受注册顺序影响，同时个人覆盖层或 `--patch` 覆盖层仍可替换或禁用这些配置项。已交付的一分钟预算用于覆盖一次辅助 DeepSeek Messages 请求及服务端检索，同时保持 `dsh-tool-web` 提供方无关的 30 秒默认值不变，以供自定义组合使用。[Web 能力 seam 决策](../architecture/2026-06-24-web-capability-seam.zh.md)负责公开抓取安全策略。 DeepSeek 搜索使用与官方会话适配器相同的 `DEEPSEEK_API_KEY` 凭据引用。提供方在每次搜索内部通过可选的 `ctx.credentials` 服务解析该引用；只有未挂载该 seam 的组合才会回退到启动进程的环境变量，非空的 `apiKey` 字面值仍作为程序化配置的最后兜底。因此，由 Web 的 Models 页存储或轮换的密钥无需重启即可用于下一次搜索，提供方也无需保留该值。由于 `WebSearchProvider.available()` 是同步方法，它会将已安装解析器视为本地可用；若动态凭据缺失，操作会以提供方专属错误码 `WEB_PROVIDER_CREDENTIAL_MISSING` 失败，而稳定的工具 schema 仍保持注册。 @@ -30,8 +32,8 @@ DeepSeek 搜索使用与官方会话适配器相同的 `DEEPSEEK_API_KEY` 凭据 **提高 `dsh-tool-web` 的提供方无关超时。** 不予采纳：自定义提供方和部署有各自不同的延迟预期；这一部署预算应归已交付的 DeepSeek 组合所有。 -**在每个共享 base surface 上启用抓取。** 不予采纳：共享 base 服务于网络策略不同的产品。它会挂载仅限公网的提供方，但保持工具按需启用；已交付的 Web preset 会有意启用该工具，其他产品则可以继续隐藏它或添加更严格的网络策略。 +**在每个共享 base surface 上启用抓取。** 本文曾因各产品可能需要不同网络策略而否决该方案。已交付产品采用同一个完整工具集合后，[共享 base 的 Web 抓取默认值](2026-09-01-shared-base-web-fetch-default.zh.md)取代了该否决；仅限公开目的地址与无需逐次审批的约束仍然有效。 ## 后果 -每个共享 base surface 的原生模型请求都会携带 `web_search` schema 与搜索指引；Web／无头 PTC 模式 通过 `run_code` 公开相同的搜索能力。搜索会增加一次完整的辅助模型调用，并可能多次使用原生服务器工具；发起会话的日志仍可精确重建其不含密钥的请求。已交付的 Web `cordis`、`ptc` 与 `standard` preset 还会暴露 `web_fetch`，实施公开地址强制校验且无需逐次审批。Web 快照通道会启动已交付配置树，使用本地 Messages fixture（测试前置数据），经由真实 DeepSeek 提供方驱动一次回放的 `web_search` 调用，断言持久化的辅助请求与结构化结果，并固定最终浏览器呈现。组合冒烟测试会固定共享搜索清单与各 preset 的抓取选择；构建后组合配置的转储固定已交付的一分钟搜索预算；提供方测试固定缺失、已存储及已轮换凭据的行为，以及字面值与环境变量的兼容性。 +headless、完整 SDK、ACP 与仅使用 base 的自定义 profile 的原生模型请求都会携带 `web_search` 和 `web_fetch` schema 与指引；Web preset 会暴露同一对工具，PTC mode 还会通过 `run_code` 暴露它们。搜索会增加一次完整的辅助模型调用，并可能多次使用原生服务器工具；发起会话的日志仍可精确重建其不含密钥的请求。抓取会强制使用公开地址，并且无需逐次审批。Web 快照通道会启动已交付配置树，使用本地 Messages fixture（测试前置数据），经由真实 DeepSeek 提供方驱动一次回放的 `web_search` 调用，断言持久化的辅助请求与结构化结果，并固定最终浏览器呈现。共享 snapshot header 会固定通用的抓取 schema 与提示指引。组合冒烟测试会固定工具集合；构建后组合配置的转储固定已交付的一分钟搜索预算；提供方测试固定缺失、已存储及已轮换凭据的行为，以及字面值与环境变量的兼容性。 diff --git a/.agents/notes/implemented/feature/2026-07-31-web-workspace-file-links.i18n.yaml b/.agents/notes/implemented/feature/2026-07-31-web-workspace-file-links.i18n.yaml index 5a5c329383..e8023fd069 100644 --- a/.agents/notes/implemented/feature/2026-07-31-web-workspace-file-links.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-31-web-workspace-file-links.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-31-web-workspace-file-links.md -2026-07-31-web-workspace-file-links.md: ddc093e1ce5e646f6b96f1517b2a26c9e10fe423 -2026-07-31-web-workspace-file-links.zh.md: f463de969bab7a47145abbbd46363b8807bcc673 +2026-07-31-web-workspace-file-links.md: 944d7bef9b83eb34895c5f74e511600a713d786b +2026-07-31-web-workspace-file-links.zh.md: 71f0051204e0cb8b65995fc88a9edb230db8f5be diff --git a/.agents/notes/implemented/feature/2026-07-31-web-workspace-file-links.md b/.agents/notes/implemented/feature/2026-07-31-web-workspace-file-links.md index ddc093e1ce..944d7bef9b 100644 --- a/.agents/notes/implemented/feature/2026-07-31-web-workspace-file-links.md +++ b/.agents/notes/implemented/feature/2026-07-31-web-workspace-file-links.md @@ -14,7 +14,7 @@ Two distinct defects sat behind that. The transcript never said what a turn had ## Decision -**A finished turn ends with the files it produced.** The row is its own plugin, `@deepseek-ai/dsh-client-ui-deliverables`, registered into the `conversation.chat.turnTail` hole the chat view renders between a closing message's body and its IconActions — ui-conversation owns the hole and the owner currency (nodes, closing seq, `openFile`), the plugin owns every policy. `producedForClosing` reads the paths off the mutation tools' own follow-along `locations` — a diff card, or a generic card whose `kind` is `edit` (the shape `str_replace_editor`'s insert presents) — so a turn's output is listed whether or not the closing message named it, and a new mutation tool joins by declaring what it does rather than by being added to a list. Reads, deletes, and failed calls contribute nothing; a path appears once per turn in first-seen order; accumulation resets on the turn boundary, so a turn that mutates and then ends without content text cannot spill into the next turn's row. The single-line lane measures its chips and localized remainder, then shows the largest prefix that fits (up to six) plus `+ N files`. One cordis.yml line composes the surface in or out; the unregistered hole renders nothing. +**A finished turn ends with the files it produced.** The row is its own plugin, `@deepseek-ai/dsh-client-ui-deliverables`, registered into the `conversation.chat.turnTail` hole the chat view renders between a closing message's body and its IconActions — ui-conversation owns the hole and the owner currency (nodes, closing seq, `openFile`), the plugin owns every policy. `producedForClosing` reads the paths off the mutation tools' own follow-along `locations` — a diff card, or a generic card whose `kind` is `edit` (the shape `str_replace_editor`'s insert presents) — so a turn's output is listed whether or not the closing message named it, and a new mutation tool joins by declaring what it does rather than by being added to a list. Reads, deletes, and failed calls contribute nothing; a path appears once per turn in first-seen order; accumulation resets on the turn boundary, so a turn that mutates and then ends without content text cannot spill into the next turn's row. CSS container-width bands select a prefix of up to six paths and its matching `+ N files` label, while flexbox shrinks and ellipsizes the visible basenames; the component performs no JavaScript layout observation ([decision](../simplification/2026-09-01-css-produced-file-layout.md)). One cordis.yml line composes the surface in or out; the unregistered hole renders nothing. **The path link reads as a link.** Underlined at rest, not only on hover. This is the smaller half of the diff and the larger half of the fix. @@ -28,9 +28,9 @@ Two distinct defects sat behind that. The transcript never said what a turn had - **Same-origin HTTP serving without isolation** — measurably unsafe, and recorded so nobody retries it: a document served beside `/api` drove `settings.describe` to a `200` with full data and `session.list` to 35 KB of every session's transcript, from a page that need not be agent-authored at all (a read row makes every file in a cloned repository openable). - **`Content-Security-Policy: sandbox` over that same-origin serving** — closes the hole by taking the document's origin away, which measurably breaks the pages this feature exists to show: the reported artifact throws `SecurityError` on load, and because an uncaught exception aborts the rest of its `