mward4/deepseek-harness
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 29 Wiki Activity

Merge origin/master into xtr/session-model-selection

Browse Source
This commit is contained in:
_Kerman committed 2026-10-20 20:57:11 +08:00
2 parents 8b79a1ddcc 78aeeae767
commit 639602e63a
339 files changed
+4448 -3734
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

.agents/notes/implemented/architecture/2026-10-17-agent-controller-message-delivery.i18n.yaml
-19
View File
@@ -1,19 +0,0 @@
# Bilingual-pair consistency record for 2026-10-17-agent-controller-message-delivery.md (docs/i18n/README.md): per heading
# section, a hash of its English and Chinese blocks outside code blocks and generated regions.
# After editing either side, bring the other along and re-record with:
# pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-10-17-agent-controller-message-delivery.md
/agent-note-delivering-messages-from-outside-an-agent-through-the-agent-controller:
en: 9105f8148e107f43
zh: a160e87bb0605da4
/agent-note-delivering-messages-from-outside-an-agent-through-the-agent-controller/problem:
en: 5071232d9743096c
zh: b553a626538b3466
/agent-note-delivering-messages-from-outside-an-agent-through-the-agent-controller/decision:
en: 1b61d351c4d2bdf1
zh: d6f8180b30d97b34
/agent-note-delivering-messages-from-outside-an-agent-through-the-agent-controller/alternatives-considered:
en: acc465cfe02b624c
zh: db1d06574f647130
/agent-note-delivering-messages-from-outside-an-agent-through-the-agent-controller/consequences:
en: 9d8486586d044709
zh: 24c7249421fd2541
.agents/notes/implemented/architecture/2026-10-17-agent-controller-message-delivery.md
-51
View File
@@ -1,51 +0,0 @@
# Agent Note: Delivering messages from outside an Agent through the Agent controller
Status: implemented
English | [中文](2026-10-17-agent-controller-message-delivery.zh.md)
## Problem
Plugins need to decide, with Host context, whether a message someone sends may enter a Session now. The experimental forge panel must create a requested worktree before the first message reaches the Agent, and must stop a message after the Session's worktree was removed. The Host had no point for that decision. `session/prompt` appended a prompt to the Agent inbox synchronously, and the only seam every sender passes, `agent/pre-step`, runs after the inbox accepted the message and the client already showed it as sent. A pre-step rejection ends the turn without returning the message to its sender, and a queued message reaches pre-step only when its turn starts. A gate in the Web composer, tried first, had no Host context and was bypassed by every other client.
## Decision
The core package `dsh-agent-controller` owns how a message from outside an Agent enters its inbox. The base bundle mounts it as `ctx.agentController`, so every base-backed profile can deliver through it; `dsh-agent` and `agent-loop` keep their inbox methods unchanged.
| Part | Contract |
|---|---|
| Entry | `ctx.agentController.deliver(agent, { mode, source, signal, prepare, beforeInsert })` prepares the content, builds the message, asks the `agent-controller/admission` waterfall for an `AdmissionDecision`, checks the sender's cancellation and that the Agent is still registered, calls `beforeInsert`, then `followup()` or `steer()`, and resolves with the message id once the message is in the inbox |
| Sender work | `prepare` turns the sender's content into durable message content inside the Agent's delivery order; the optional `beforeInsert` runs synchronously immediately before the append, for bookkeeping that must precede the Agent's wake |
| Listeners | Run in registration order with Agent scope filtering and receive the Agent, the message, and a signal that aborts when the sender cancels or the controller stops; return the decision of `next()` to admit, await before `next()` to hold, and return `{ kind: 'reject', reason, message, presented }` to refuse without asking later listeners; throw only when the check itself fails |
| Delivery mode | Stays with the sender; listeners do not see it, because no admission policy depends on it |
| Order | Messages to one Agent are prepared, decided, and appended one at a time in arrival order; a sender that cancels before the append is not admitted and stops waiting at once, even behind a held message; disposing the controller stops every message that has not entered an inbox and finishes once those deliveries settled |
| Failures | A rejection raises `AgentAdmissionRejected`, a `HarnessError` with code `AGENT_ADMISSION_REJECTED`, with its reason and `presented`; senders recognize it by that code rather than its class and import only types from this package; a thrown listener error raises it with `agent-controller/admission-failed`, and an Agent that left the registry with `agent-controller/agent-gone`; errors from `prepare`, `beforeInsert`, `followup()`, and `steer()` pass through; no listener runs after the message entered the inbox |
Only senders that deliver work on someone's behalf call `deliver()`. Messages the Agent's own machinery produces, such as job results, goal rounds, plan transitions, and question answers, keep `followup()`, `steer()`, and `inject()`: they belong to work that was already admitted. A listener must not wait for a person when the message source is unattended, such as a schedule or a webhook; it passes or refuses such a message.
`presented` tells an interactive client that the refusing plugin explains the refusal in its own interface, so the client adds no generic notice.
The Web Session Remote adopts the controller first. `session/prompt` resolves or resumes the Agent, acknowledges a repeated `requestId` itself, and delivers with a `prepare` that checks image input against the Session model, resolves upload receipts, and stores attachments, and a `beforeInsert` that binds the receipts to the request. It maps a refusal to `session/prompt-rejected` with `reason` and `presented`. The call stays open while a listener holds the message, and the client's cancellation stops it. The SDK server, ACP, headless runs, schedules, webhooks, and subagent prompts move to `deliver()` in later changes, each mapping a refusal to its own protocol. Until they do, a plugin that must cover them keeps an `agent/pre-step` check.
Request deduplication stays in the Session Remote. It reads the Session history synchronously, and the [synchronous-read decision](2026-09-09-deprecate-synchronous-session-event-reads.md) prohibits copying such a read into new production code; the controller can take over deduplication once a projection records accepted request ids.
The Web client keeps its existing echo: Enter clears the composer and shows the message at once, and a refusal reaches the client as an ordinary prompt failure, which retires the echo and restores the draft unless new text was typed meanwhile.
## Alternatives considered
- **An admission stage in `dsh-agent`.** `ctx.agents.admit()` with an `agent/admission` event reaches every sender as well, but it adds a delivery concern and a new extension point to the Agent core, whose contract the loop implements. A package composed in the base bundle reaches the same senders without changing that contract.
- **A Session Remote event, or a layer inside the Session Controller.** It fits the Web path, but the Session Controller is composed only in the Web profile, so ACP, SDK, and headless senders could never reach it.
- **`agent/pre-step`.** It covers every sender, but it runs after the inbox accepted the message and the client showed it as sent, a refusal drops the message instead of returning it, and a queued message is checked only when its turn starts.
- **Gated, asynchronous `followup()` and `steer()`.** Every internal sender would pass listeners meant for outside input, and their callers rely on synchronous enqueueing.
- **A composer gate in the Web client.** It has no Host context, and other clients and `ctx.conversation.send` bypass it.
- **Appending in the innermost step, with refusals thrown.** Listeners could act once `next()` returned with the message already in the inbox, but that capability, which no listener uses, cost three special cases: a listener failure after the append could only be logged, the inbox's own refusal had to be told apart from a listener failure, and a refusal thrown from another module copy of the error class needed a structural check. A returned decision keeps the stage a pure decision, as `agent/pre-step` is; a plugin that reacts to admitted messages listens to `agent/inbox/inserted`.
The liveshare proposal in PR #3854 also introduced an admission hook, for caller identity. [Peer principals](2026-10-12-peer-principal-and-remote-access.md) replaced it; this stage decides whether the Agent can take a message now, independent of who sent it.
## Consequences
- A listener that holds one message keeps later messages to the same Agent waiting; other Agents proceed.
- While a listener holds a Web prompt, the client already shows its echo; a refusal then retires the echo and returns the draft.
- `prepare` stores attachments before the decision, so a refused prompt can leave a stored attachment unreferenced.
- Image prompts deliver inside the Session Remote's image admission chain, so a model switch waits while a listener holds an image prompt.
- Until every outside sender adopts `deliver()`, the stage is not a complete enforcement point.
.agents/notes/implemented/architecture/2026-10-17-agent-controller-message-delivery.zh.md
-51
View File
@@ -1,51 +0,0 @@
# Agent Note：经 Agent 控制器投递来自 Agent 外部的消息
Status: implemented
[English](2026-10-17-agent-controller-message-delivery.md) | 中文
## 问题
插件需要借助 Host 侧的上下文，判断别人发来的一条消息现在能否进入 Session。实验性的 forge 面板必须在第一条消息到达 Agent 之前建好所请求的工作树，并在 Session 的工作树被删除后拦下消息。Host 此前没有做这个判断的位置：`session/prompt` 同步地把提示词追加进 Agent 收件箱，而所有发送方都会经过的唯一扩展点 `agent/pre-step` 运行在收件箱接受消息、客户端已显示为已发送之后。pre-step 拒绝会结束本轮，却不会把消息退回给发送方；排队的消息也要等到它的轮次开始才会到达 pre-step。最初尝试的 Web 输入框闸门拿不到 Host 上下文，其他客户端也都会绕过它。
## 决策
core 包 `dsh-agent-controller` 负责来自 Agent 外部的消息如何进入其收件箱。base 组合包把它挂载为 `ctx.agentController`，因此每个基于 base 的 profile 都能经它投递；`dsh-agent` 与 `agent-loop` 的收件箱方法保持不变。
| 部分 | 约定 |
|---|---|
| 入口 | `ctx.agentController.deliver(agent, { mode, source, signal, prepare, beforeInsert })` 准备内容、构造消息、向 `agent-controller/admission` waterfall 要一个 `AdmissionDecision`，检查发送方是否取消、Agent 是否仍在注册表中，调用 `beforeInsert`，再调用 `followup()` 或 `steer()`，消息进入收件箱后返回消息 id |
| 发送方的工作 | `prepare` 在该 Agent 的投递顺序内把发送方的内容转成持久的消息内容；可选的 `beforeInsert` 紧挨在放入之前同步运行，用于必须先于 Agent 唤醒完成的登记 |
| 监听器 | 按注册顺序运行，按 Agent 作用域过滤，拿到 Agent、消息和一个在发送方取消或控制器停止时中止的信号；返回 `next()` 的决定即放行，在 `next()` 前 await 即暂留，返回 `{ kind: 'reject', reason, message, presented }` 即拒绝且不再询问后面的监听器；只在检查本身执行失败时抛出异常 |
| 投递方式 | 留在发送方，监听器看不到，因为目前没有按投递方式区分的准入策略 |
| 顺序 | 同一个 Agent 的消息按到达顺序逐条准备、决定并放入；在放入之前取消的发送方不会被接纳，并立即结束等待，即使排在被暂留的消息之后；销毁控制器会停止所有尚未进入收件箱的消息，并在这些投递结束后才完成 |
| 失败 | 拒绝时抛出带其 reason 与 `presented` 的 `AgentAdmissionRejected`，它是 code 为 `AGENT_ADMISSION_REJECTED` 的 `HarnessError`；发送方按这个 code 而不是按类识别拒绝，对本包只导入类型；监听器抛出异常时 reason 为 `agent-controller/admission-failed`，Agent 离开注册表时为 `agent-controller/agent-gone`；`prepare`、`beforeInsert`、`followup()` 与 `steer()` 的错误原样抛出；消息进入收件箱之后不再有监听器运行 |
只有代表他人送来工作的发送方调用 `deliver()`。Agent 自身机制产生的消息，例如后台任务结果、goal 轮次、plan 切换和问题答复，继续使用 `followup()`、`steer()` 与 `inject()`：它们属于已经接纳的工作。消息来源无人值守时（例如定时任务或 webhook），监听器不得等待人来处理，只能放行或拒绝。
`presented` 告诉交互式客户端，拒绝它的插件会在自己的界面里给出说明，客户端因此不再追加通用提示。
Web Session Remote 首先接入控制器。`session/prompt` 自己解析或恢复 Agent、确认重复的 `requestId`，再调用投递：`prepare` 按 Session 模型检查图片输入、解析上传凭证并存储附件，`beforeInsert` 把凭证绑定到该请求。它把拒绝映射为带 `reason` 与 `presented` 的 `session/prompt-rejected`。监听器暂留消息期间调用保持未完成，客户端取消会让它停止。SDK 服务端、ACP、headless 运行、定时任务、webhook 与子代理提示词在后续改动中改用 `deliver()`，各自把拒绝映射到自己的协议。在此之前，必须覆盖它们的插件仍保留一个 `agent/pre-step` 检查。
请求去重留在 Session Remote。它同步读取 Session 历史，而[同步读取决策](2026-09-09-deprecate-synchronous-session-event-reads.zh.md)禁止把这类读取复制到新的生产代码；等有投影记录已接收的请求 id 后，控制器才能接管去重。
Web 客户端保留现有的回显方式：Enter 立即清空 composer 并显示消息；拒绝在客户端表现为一次普通的提示词失败，回显撤下，草稿回到 composer，除非期间已经输入了新内容。
## 备选方案
- **在 `dsh-agent` 里加接纳阶段。** `ctx.agents.admit()` 加 `agent/admission` 事件同样能覆盖所有发送方，但它把投递职责和一个新扩展点加进了 Agent 核心，而 Agent 的约定由循环实现。组合在 base 组合包里的独立包能覆盖同样的发送方，又不改变这项约定。
- **Session Remote 自己的事件，或 Session Controller 内部的一层。** 它适合 Web 路径，但 Session Controller 只在 Web profile 中组合，ACP、SDK 与 headless 发送方永远无法使用它。
- **`agent/pre-step`。** 它覆盖所有发送方，但运行在收件箱接受消息、客户端显示为已发送之后；拒绝会丢掉消息而不是退回；排队的消息要到轮次开始才会被检查。
- **让 `followup()` 与 `steer()` 异步并经过闸门。** 每个内部发送方都会经过为外部输入准备的监听器，而它们的调用方依赖同步入队。
- **Web 客户端的输入框闸门。** 它拿不到 Host 上下文，其他客户端和 `ctx.conversation.send` 都会绕过它。
- **在最内层步骤放入消息，并以抛出异常表示拒绝。** 监听器能在 `next()` 返回、消息已进入收件箱之后继续执行，但这项没有监听器使用的能力带来三处特殊处理：放入之后的监听器失败只能记录日志；要把收件箱自身的拒绝与监听器失败区分开；另一个模块副本抛出的错误类需要按结构识别。返回决定让这个阶段和 `agent/pre-step` 一样只做决定；需要对已接纳消息作出反应的插件监听 `agent/inbox/inserted`。
PR #3854 中的 liveshare 方案也引入过一个接纳钩子，用于调用方身份。[Peer principals](2026-10-12-peer-principal-and-remote-access.zh.md) 取代了它；本阶段判断 Agent 现在能否接收一条消息，与发送者是谁无关。
## 影响
- 监听器暂留一条消息时，同一 Agent 之后的消息都要等待；其他 Agent 不受影响。
- 监听器暂留一条 Web 提示词期间，客户端已经显示其回显；随后被拒绝时，回显撤下，草稿回到 composer。
- `prepare` 在决定之前存储附件，因此被拒绝的提示词可能留下一个无人引用的已存附件。
- 图片提示词在 Session Remote 的图片接纳链内投递，因此监听器暂留图片提示词期间，切换模型要等待。
- 在所有外部发送方都改用 `deliver()` 之前，这个阶段还不是完整的强制执行点。
.agents/notes/implemented/architecture/2026-10-19-pi-ai-system-reminder-fallback.i18n.yaml
+19
View File
@@ -0,0 +1,19 @@
# Bilingual-pair consistency record for 2026-10-19-pi-ai-system-reminder-fallback.md (docs/i18n/README.md): per heading
# section, a hash of its English and Chinese blocks outside code blocks and generated regions.
# After editing either side, bring the other along and re-record with:
# pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-10-19-pi-ai-system-reminder-fallback.md
/agent-note-pi-ai-system-reminder-fallback:
en: f43b5c6974289180
zh: abc0a86b1d137189
/agent-note-pi-ai-system-reminder-fallback/problem:
en: 5247278483ca4e7a
zh: ca711671a43fae76
/agent-note-pi-ai-system-reminder-fallback/decision:
en: 604eb72704c73328
zh: 6661d1065f65ecf9
/agent-note-pi-ai-system-reminder-fallback/alternatives-considered:
en: 0e2aaf897587ebb0
zh: 0a7f9a9cb7ddffaf
/agent-note-pi-ai-system-reminder-fallback/consequences:
en: 8a0c53289a04c64a
zh: 591e602eb2036a1a
.agents/notes/implemented/architecture/2026-10-19-pi-ai-system-reminder-fallback.md
+25
View File
@@ -0,0 +1,25 @@
# Agent Note: Pi-ai system reminder fallback
Status: implemented
English | [中文](2026-10-19-pi-ai-system-reminder-fallback.zh.md)
## Problem
Pi-ai concatenates later system messages into the initial prompt for models without native mid-conversation system support. Later system messages supplied to the adapter therefore invalidate the preceding conversation cache prefix.
## Decision
The pi-ai adapter uses the prepared model's existing `supportsMidConvoSystemMessages` capability mapping. It preserves native system messages on capable models and converts each non-empty later system message into a user message wrapped in `<system-reminder>` otherwise, before pi-ai receives the context. Leading system text remains in the initial prompt. Message order, repeated updates, and existing text-block concatenation remain intact. Durable messages retain their original roles.
The fallback belongs to the adapter because it depends on the selected protocol capability. Prompt admission remains owned by the agent loop, which consolidates system-prompt updates on routes without native support; this fallback only handles later system messages present in the supplied request.
## Alternatives considered
**Let pi-ai concatenate updates.** Retains system-role priority but changes the initial prompt whenever later system text changes, defeating prefix reuse.
**Advertise native support or send developer messages.** Misstates model capability or relies on another unsupported instruction role. The fallback uses ordinary user messages and does not change capability metadata.
## Consequences
The wrapper does not confer system-role authority, so reminders have user-role priority. Cache hits remain provider-dependent; tool changes and compaction can still invalidate prefixes. Both text-only and attachment-aware conversion apply the same fallback, and wire tests inspect the real pi-ai serializer output.
.agents/notes/implemented/architecture/2026-10-19-pi-ai-system-reminder-fallback.zh.md
+25
View File
@@ -0,0 +1,25 @@
# Agent Note: Pi-ai 系统提醒回退
Status: implemented
[English](2026-10-19-pi-ai-system-reminder-fallback.md) | 中文
## 问题
对于不原生支持对话中系统消息的模型，pi-ai 会把后续系统消息拼接进初始提示词。因此，传给适配器的后续系统消息会使此前的对话缓存前缀失效。
## 决策
pi-ai 适配器使用已准备模型的现有 `supportsMidConvoSystemMessages` 能力映射。对于具备该能力的模型，适配器保留原生系统消息；否则，在 pi-ai 接收上下文之前，将每条非空的后续系统消息转换为以 `<system-reminder>` 包裹的用户消息。开头的系统文本仍位于初始提示词中。消息顺序、重复更新和现有文本块拼接方式均保持不变。持久消息保留原始角色。
回退由适配器负责，因为它取决于所选协议的能力。提示词纳入仍由 Agent loop 负责，它在不支持原生更新的路由上合并系统提示词更新；此回退仅处理传入请求中已有的后续系统消息。
## 考虑过的替代方案
**让 pi-ai 拼接更新。** 保留系统角色优先级，但后续系统文本每次变化都会改变初始提示词，破坏前缀复用。
**宣称原生支持或发送 developer 消息。** 错报模型能力，或依赖另一种不受支持的指令角色。回退使用普通用户消息，不改变能力元数据。
## 后果
包裹文本不会赋予系统角色权限，因此提醒具有用户角色优先级。缓存命中仍取决于提供方；工具变化和压缩仍可能使前缀失效。纯文本与附件感知转换使用相同的回退，线协议测试检查真实 pi-ai 序列化器的输出。
.agents/notes/implemented/feature/2026-08-28-auto-review.i18n.yaml
+2 -2
View File
@@ -15,8 +15,8 @@
en: db0a2deb9193a00d
zh: dab5a65b88957464
/agent-note-experimental-auto-review-before-each-tool-call/decision/one-complete-reviewer-request:
en: 3a07a59b6c5ec26c
zh: e4f3290dc6d26368
en: 7bd9fef90e8639b9
zh: e2a096885c9c1e83
/agent-note-experimental-auto-review-before-each-tool-call/decision/result-and-cancellation:
en: 647fe68c52b907e0
zh: 267a739bad44ad65
.agents/notes/implemented/feature/2026-08-28-auto-review.md
+10 -9
View File
@@ -30,19 +30,20 @@ The reviewer derives authority from existing Session facts. Text with `source.ki
### One complete reviewer request
The integration uses only the latest `request/header.config` provider/model and the shipped adapter's default reasoning. It neither compares redundant route metadata nor copies the main agent's request. The request has five fixed partitions:
The integration uses only the latest `request/header.config` provider/model and the shipped adapter's default reasoning. It neither compares redundant route metadata nor copies the main agent's request. The request is a fixed system prompt followed by user messages:
| Partition | Retained input |
| Part | Retained input |
| --- | --- |
| `REVIEW_POLICY` | Fixed classification, source authority, and strict result rules; allow executes immediately with Full access and no later confirmation |
| `ENVIRONMENT` | Existing Session header `cwd` only |
| `PROJECT_INSTRUCTIONS` | Visible project instructions with their original sources and constraint role |
| `FILTERED_HISTORY` | Current compaction surface's sourced user/direct-parent messages, checkpoints, image/attachment facts, and historical call names with logged arguments |
| `PENDING_ACTION` | Tool name, description, parameter schema, and parsed arguments |
| System prompt `REVIEW_POLICY` | Fixed classification, source authority, message layout, and strict result rules; allow executes immediately with Full access and no later confirmation |
| First user message `ENVIRONMENT` | Existing Session header `cwd` only |
| One user message per retained item | Current compaction surface's sourced user/direct-parent messages, project instructions with the constraint role, checkpoints, image/attachment facts, historical call names with logged arguments, and a `tool-schema` item before the first call that used each reviewed schema, in surface order |
| Final user message | The pending call with its logged arguments; the latest earlier `tool-schema` item with its mode and name gives the tool description and parameter schema |
The reviewer builds those two action sections from the Session's complete action history: an authorization is the earlier call the `REVIEW_POLICY` source rules make eligible, and a duplicate or conflicting identity has to be visible anywhere in the log rather than inside a recent window, so neither the Session projections nor a bounded read serves the decision. The read is the deprecated synchronous `snapshotEvents()` under a line-scoped `typescript/no-deprecated` waiver.
Each item is its own message and keeps the same text in later reviews: the pending call becomes a historical call, and the integration remembers, for each live Session, the schema each reviewed call showed so that it reappears at the same position. A review's messages therefore begin with all messages of the previous review. Providers that cache only at message ends, such as OpenAI implicit breakpoints on GPT-5.6 and later, or at a final marker with lookback, such as Anthropic `cache_control`, reuse the previous review; one growing user message let them reuse only the policy. The schema memory is process-local, so the first review after a restart omits earlier schemas and misses the cache once.
The main agent's V3 `system/message` nodes, assistant text/reasoning, and tool results are excluded. The current call must belong to the open step recorded by `step/start`; missing step ownership fails closed. It appears only in `PENDING_ACTION`; an unstarted sibling has no historical call fact. Native schema comes from the latest request header. PTC captures a frozen schema at binding construction and passes it through the scheduler into `ToolExecution`; descriptions and parameter schemas never enter start/settle events or the Session/SDK wire. Missing, inconsistent, or ambiguous action facts reject the call without consulting the live registry. An oversized request fails closed without summarization, truncation, another compaction pass, or a small output-token budget.
The reviewer builds the history and the pending call from the Session's complete action history: an authorization is the earlier call the `REVIEW_POLICY` source rules make eligible, and a duplicate or conflicting identity has to be visible anywhere in the log rather than inside a recent window, so neither the Session projections nor a bounded read serves the decision. The read is the deprecated synchronous `snapshotEvents()` under a line-scoped `typescript/no-deprecated` waiver.
The main agent's V3 `system/message` nodes, assistant text/reasoning, and tool results are excluded. The current call must belong to the open step recorded by `step/start`; missing step ownership fails closed. It appears only as the final message; an unstarted sibling has no historical call fact. Native schema comes from the latest request header. PTC captures a frozen schema at binding construction and passes it through the scheduler into `ToolExecution`; descriptions and parameter schemas never enter start/settle events or the Session/SDK wire. Missing, inconsistent, or ambiguous action facts reject the call without consulting the live registry. An oversized request fails closed without summarization, truncation, another compaction pass, or a small output-token budget.
### Result and cancellation
.agents/notes/implemented/feature/2026-08-28-auto-review.zh.md
+10 -9
View File
@@ -30,19 +30,20 @@ Reviewer 从既有 Session 事实派生权威。`source.kind === 'user'` 的文
### 一次完整 reviewer 请求
Integration 只使用最新 `request/header.config` 的 provider／model 与 shipped adapter 默认 reasoning。它既不比较冗余 route metadata，也不复制主 agent 请求。请求固定包含五个分区：
Integration 只使用最新 `request/header.config` 的 provider／model 与 shipped adapter 默认 reasoning。它既不比较冗余 route metadata，也不复制主 agent 请求。请求由固定 system prompt 和其后的 user 消息组成：
| 分区 | 保留输入 |
| 部分 | 保留输入 |
| --- | --- |
| `REVIEW_POLICY` | 固定分类、来源权威与严格结果规则；allow 后立即以 Full access 执行，没有后续确认 |
| `ENVIRONMENT` | 仅既有 Session header `cwd` |
| `PROJECT_INSTRUCTIONS` | 可见项目指令，保留原始来源并赋予约束角色 |
| `FILTERED_HISTORY` | 当前 compaction surface 中带来源的用户／直接父级消息、checkpoint、图片／附件事实，以及历史调用名称与日志参数 |
| `PENDING_ACTION` | 工具名称、描述、参数 schema 与解析后的 arguments |
| System prompt `REVIEW_POLICY` | 固定分类、来源权威、消息布局与严格结果规则；allow 后立即以 Full access 执行，没有后续确认 |
| 第一条 user 消息 `ENVIRONMENT` | 仅既有 Session header `cwd` |
| 每个保留条目一条 user 消息 | 当前 compaction surface 中带来源的用户／直接父级消息、带约束角色的项目指令、checkpoint、图片／附件事实、历史调用名称与日志参数，以及在首次使用某个已审 schema 的调用之前出现的 `tool-schema` 条目，按 surface 顺序排列 |
| 最后一条 user 消息 | 待审调用及其日志参数；其之前同 mode 与名称的最新 `tool-schema` 条目给出工具描述与参数 schema |
reviewer 从 Session 的完整动作历史构建这两个动作分节：授权是符合 `REVIEW_POLICY` 来源规则的那个更早调用，而重复或冲突的身份必须在整份日志的任何位置都可见，而不是只在近期窗口内可见，因此 Session 投影与有界读取都无法支撑该判定。这次读取使用已废弃的同步 `snapshotEvents()`，并带有行级 `typescript/no-deprecated` 豁免。
每个条目各占一条消息，并在后续评审中保持相同文本：待审调用会变成历史调用，integration 为每个存活 Session 记住每次已审调用展示的 schema，使其在同一位置再次出现。因此一次评审的消息以上一次评审的全部消息开头。只在消息结尾缓存的 provider（如 GPT-5.6 及以后的 OpenAI 隐式断点），或在末尾标记处向前回溯的 provider（如 Anthropic `cache_control`），都能复用上一次评审；单条不断增长的 user 消息只能让它们复用策略。Schema 记忆只在进程内，因此重启后的第一次评审会缺少先前的 schema，并有一次缓存未命中。
主 agent 的 V3 `system/message` 节点、assistant 正文／reasoning 与 tool results 全部排除。当前调用必须属于 `step/start` 记录的开放 step；缺少 step 归属时拒绝执行。该调用只在 `PENDING_ACTION` 出现；尚未开始的 sibling 没有历史调用事实。原生 schema 来自最新 request header。PTC 在 binding 构造时捕获冻结 schema，经由调度器传入 `ToolExecution`；描述与参数 schema 不进入开始／结算事件或 Session／SDK wire。动作事实缺失、不一致或有歧义时拒绝调用，不查询 live registry。超窗请求直接拒绝，不做摘要、截断、额外 compaction 或设置小型输出 token 预算。
reviewer 从 Session 的完整动作历史构建历史与待审调用：授权是符合 `REVIEW_POLICY` 来源规则的那个更早调用，而重复或冲突的身份必须在整份日志的任何位置都可见，而不是只在近期窗口内可见，因此 Session 投影与有界读取都无法支撑该判定。这次读取使用已废弃的同步 `snapshotEvents()`，并带有行级 `typescript/no-deprecated` 豁免。
主 agent 的 V3 `system/message` 节点、assistant 正文／reasoning 与 tool results 全部排除。当前调用必须属于 `step/start` 记录的开放 step；缺少 step 归属时拒绝执行。该调用只作为最后一条消息出现；尚未开始的 sibling 没有历史调用事实。原生 schema 来自最新 request header。PTC 在 binding 构造时捕获冻结 schema，经由调度器传入 `ToolExecution`；描述与参数 schema 不进入开始／结算事件或 Session／SDK wire。动作事实缺失、不一致或有歧义时拒绝调用，不查询 live registry。超窗请求直接拒绝，不做摘要、截断、额外 compaction 或设置小型输出 token 预算。
### 结果与取消
.agents/notes/implemented/process/2026-09-29-weighted-approval-decision-check-run.i18n.yaml
-22
View File
@@ -1,22 +0,0 @@
# Bilingual-pair consistency record for 2026-09-29-weighted-approval-decision-check-run.md (docs/i18n/README.md): per heading
# section, a hash of its English and Chinese blocks outside code blocks and generated regions.
# After editing either side, bring the other along and re-record with:
# pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-09-29-weighted-approval-decision-check-run.md
/agent-note-weighted-approval-decisions-are-completed-check-runs:
en: 1b120e96b0f5f340
zh: ff20c20fe0558964
/agent-note-weighted-approval-decisions-are-completed-check-runs/problem:
en: 6ba6f6a15a1562ec
zh: 1cbef3e736c258e6
/agent-note-weighted-approval-decisions-are-completed-check-runs/decision:
en: 7dcebfed221053cc
zh: 84346547d1b95ee3
/agent-note-weighted-approval-decisions-are-completed-check-runs/verification:
en: 3771d95af79c69bb
zh: d4053338d878511f
/agent-note-weighted-approval-decisions-are-completed-check-runs/alternatives-considered:
en: b8daea66b0a90d26
zh: fb7079c815b38281
/agent-note-weighted-approval-decisions-are-completed-check-runs/consequences:
en: 5708f2112ee1e647
zh: 02d29ce5b18db06d
.agents/notes/implemented/process/2026-09-29-weighted-approval-decision-check-run.md
-61
View File
@@ -1,61 +0,0 @@
# Agent Note: Weighted approval decisions are completed check runs
Status: implemented
English | [中文](2026-09-29-weighted-approval-decision-check-run.zh.md)
## Problem
The weighted approval policy must block merging until its conditions hold, distinguish an unmet condition from a failed evaluation, and let tools that wait for pull-request checks to finish terminate. A draft pull request, insufficient approval points, and an effective `CHANGES_REQUESTED` review from a write-capable reviewer are reversible review states rather than evaluation failures.
Commit statuses offer only `pending`, `success`, `failure`, and `error`. A `pending` commit status blocks merging, but `gh pr checks --watch` and other CI watchers treat it as running work and wait until someone approves. A `failure` commit status ends the wait but reports an unmet, reversible condition as a malfunction and conflates the policy outcome with publisher health.
## Decision
The [publisher](../../../../.github/review-ownership/check-approval.mjs) reports the decision as the `weighted approval decision` check run on the pull-request head, written with the job's GitHub Actions token. Branch rules require that check run with GitHub Actions as its expected source.
| Condition | Check run |
|---|---|
| First evaluation of a head | `in_progress` until the decision is written |
| Re-evaluating a head that published `success` | conclusion `action_required` until the decision is written |
| Draft, fewer than the required points, or an effective write-capable `CHANGES_REQUESTED` review | `completed`, conclusion `action_required` |
| Ready, threshold met, and no blocking review | `completed`, conclusion `success` |
| Setup, evaluation, or cancellation failure | `completed`, conclusion `failure` |
A check run that reached `completed` never returns to `in_progress` or `queued`: GitHub answers `200` to that update and leaves the run completed, so only the conclusion is mutable. The publisher therefore revokes a published success by writing the blocking conclusion before dependency setup, which covers the setup and history fetch that dominate a cold evaluation. The head keeps one GitHub Actions run with the required name: the publisher updates the head's existing run and creates one only when none exists. An incomplete list or a run without a valid ID fails the publication. The publish step records the conclusion it wrote in a step output, and the setup-failure step runs only when no conclusion was recorded: a published decision survives a later failure or cancellation, while a run cancelled before that write concludes `failure`. A run cancelled before the publish step records its conclusion, including a cancellation whose file-command outputs the runner drops, therefore concludes `failure`, which still blocks the merge. During a re-evaluation, a previous decision other than a published success keeps its conclusion and title until the new decision is written; a published success is replaced by `action_required` and the preparation title.
A blocking review dominates the point total, so the conclusion remains `action_required` even when counted approvals reach the threshold. GitHub accepts only `success`, `neutral`, and `skipped` for a required check, so `action_required` blocks merging while completing the check.
## Verification
[Approval policy tests](../../../../.github/review-ownership/check-approval.test.mjs) pin the threshold-reaching blocker case, the exact published check-run payloads, revoking a published success through its conclusion, updating an existing run instead of creating another, recording the conclusion that the setup-failure step reads, and invalid list responses. [Workflow tests](../../../../scripts/ci-workflow.spec.ts) pin the `checks: write` permission, the `revoke` and `failure` phase commands, the publish step identity, and the fallback that runs when a run fails or is cancelled before publishing.
The transition matrix and the fallback condition were measured on 2026-09-30 with throwaway tag runs in this repository, using the same GitHub Actions token as the publisher. A run created as `in_progress` and completed as `success` accepted `status: in_progress` and `status: queued` with `200` and stayed `completed`, while conclusion-only updates moved it to `action_required`, back to `success`, and to `failure`. Those runs measured the fallback condition as it stood on 2026-09-30 (`failure() || (cancelled() && steps.publish.outcome != 'success')`): the step ran for a failed publish and for a cancellation during the publish step, and did not run when the job was cancelled after the decision was published. The current condition also requires an empty `concluded` output, so a job cancelled after the publish step recorded its conclusion skips the fallback in either condition. A `timeout-minutes` cancellation was measured separately on 2026-10-20: a job that exceeded its one-minute timeout on a throwaway `push` branch still ran its `cancelled()` step, so the step executes on that path; whether a cancelled step's file-command outputs are exposed was not measured, and dropping them only makes that cancellation conclude `failure`.
The list response's `total_count` counts the `check_name`-filtered runs rather than every check on the head: on 2026-10-20 a head carrying 33 check runs reported `total_count` 0 for a name that matches none and 2 for an existing name, so an equality between `total_count` and the returned array length rejects a truncated page.
## Alternatives considered
**Keep the `pending` commit status.** Branch rules block correctly, but CI watchers cannot distinguish an approval wait from running checks and do not finish.
**Publish `failure` for unmet conditions.** This ends the wait, but it represents an unmet, reversible merge condition as a malfunction and conflates the policy outcome with publisher health.
**Revoke by returning the completed run to `in_progress`.** GitHub accepts that update without applying it, so the published success would stand through dependency setup and the history fetch.
**Publish nothing until success.** A required check that was never reported remains expected and is invisible to many watchers. Revoking a success already published on the same head still requires a non-success write, and each watcher's treatment of a missing required check is unspecified.
**Keep the `weighted approval` name.** When a check and a commit status share a required name, GitHub requires both to pass. Heads that carry a `pending` commit status from the earlier publisher would stay blocked until a new push.
**Create a new check run on every evaluation.** Several same-name check runs on one head make the result depend on how GitHub and each watcher select among them.
**Let approval points override a blocking review.** This makes the score the only success condition, but it permits success while a write-capable reviewer's effective decision still requests changes.
## Consequences
CI watchers finish while the decision waits on reviewers. `gh pr checks` groups `action_required` with failed checks and exits non-zero; consumers distinguish it by the check name, the `ACTION_REQUIRED` conclusion, or the output title, which carries the score or blocker. The `weighted approval publisher` Actions job and the `failure` conclusion carry operational failures.
Re-evaluating a head that already published `success` reports `action_required` until the new decision is written, so a watcher that polls inside that window sees a failed check.
The branch-rule switch leaves two gaps on heads the publisher has not evaluated since it. A check run's name does not identify its publisher. Any workflow in the pushed branch can name a job `weighted approval decision`: on 2026-10-20 a job with `if: false` on a throwaway branch produced a `github-actions` check run with that name and conclusion `skipped` on the pushed commit, and a required check accepts `skipped`. The publisher therefore revokes every passing conclusion it finds before dependency setup and adopts every same-named run when it publishes, which covers a run that exists at either write; a same-named passing run created after the publisher's last write on the head is outside its control. Requiring the check from a dedicated GitHub App is the only complete binding, and it replaces the job token with an app-installed identity.
A head without the decision check run reports the required check as expected until the next publisher event creates it; re-running a pre-switch run does not, because a re-run replays that run's earlier workflow file, which calls the removed `pending` phase against the new script. A re-run of a pre-switch run on 2026-10-20 executed the pre-switch step names and no check-run write. A pull-request edit triggers the publisher's `edited` event, so the rollout can create the run on a head that has none. A head that also carries the earlier `weighted approval` commit status keeps that pending status until the same event: the combined status resolves each context to its latest status, so the rollout posts a terminal `success` status for that context on every open head instead of waiting for its next push.
.agents/notes/implemented/process/2026-09-29-weighted-approval-decision-check-run.zh.md
-61
View File
@@ -1,61 +0,0 @@
# Agent Note: 加权批准决策使用已完成的 check run
Status: implemented
[English](2026-09-29-weighted-approval-decision-check-run.md) | 中文
## 问题
加权批准策略必须在条件满足之前阻止合并，区分尚未满足的条件与失败的评估，并让等待 PR 检查结束的工具能够退出。draft PR、批准点数不足，以及具有写权限的评审人仍然生效的 `CHANGES_REQUESTED` 评审，都是可撤销的评审状态，而不是评估故障。
commit status 只提供 `pending`、`success`、`failure` 和 `error`。`pending` commit status 会阻止合并，但 `gh pr checks --watch` 和其他 CI 监视工具把它当作仍在运行的工作，会一直等到有人批准。`failure` commit status 能结束等待，但会把尚未满足且可撤销的条件报告为故障，并混淆策略结果与 publisher 的健康状态。
## 决策
[publisher](../../../../.github/review-ownership/check-approval.mjs) 使用该 job 的 GitHub Actions token，在 PR head 上以 `weighted approval decision` check run 报告决策。分支规则要求该 check run，并把 GitHub Actions 设为预期来源。
| 条件 | Check run |
|---|---|
| head 的首次评估 | 写入决策前为 `in_progress` |
| 重新评估已发布 `success` 的 head | 写入决策前 conclusion 为 `action_required` |
| draft、点数少于要求，或存在具有写权限的评审人仍然生效的 `CHANGES_REQUESTED` 评审 | `completed`，conclusion 为 `action_required` |
| 已 ready、达到阈值且没有阻塞性评审 | `completed`，conclusion 为 `success` |
| 安装、评估或取消导致的失败 | `completed`，conclusion 为 `failure` |
已经到达 `completed` 的 check run 不会再回到 `in_progress` 或 `queued`：GitHub 对该更新返回 `200`，但让 run 保持已完成，因此只有 conclusion 可变。publisher 因此在依赖安装前用阻塞性 conclusion 撤销已发布的成功，覆盖冷启动中占主要耗时的安装与历史拉取。同一 head 上只保留一个同名 GitHub Actions run：publisher 更新 head 上已有的 run，仅在不存在时创建一个。列表不完整或 run 没有有效 ID 时发布失败。publish 步骤把自己写入的 conclusion 记录到 step output，安装失败步骤只在该 output 未记录时运行：已发布的决策在之后的失败或取消中保留，而在该写入之前被取消的 run 以 `failure` 结束。在 publish 步骤记录 conclusion 之前被取消的 run——包括 runner 丢弃其 file-command 输出的取消——因此以 `failure` 结束，仍然阻止合并。重新评估期间，先前决策若不是已发布的 `success`，其 conclusion 与标题会一直显示到新决策写入为止；已发布的 `success` 会被替换为 `action_required` 与准备中的标题。
阻塞性评审的优先级高于点数总和，因此即使计入的批准已经达到阈值，conclusion 仍为 `action_required`。GitHub 只接受 `success`、`neutral` 和 `skipped` 作为必需检查的通过结果，因此 `action_required` 会阻止合并，同时让该检查处于完成状态。
## 验证
[批准策略测试](../../../../.github/review-ownership/check-approval.test.mjs)锁定已达到阈值但仍有 blocker 的场景、准确发布的 check run payload、通过 conclusion 撤销已发布的成功、更新已有 run 而不是另建一个、记录安装失败步骤所读的 conclusion，以及无效的列表响应。[工作流测试](../../../../scripts/ci-workflow.spec.ts)锁定 `checks: write` 权限、`revoke` 和 `failure` 阶段命令、publish 步骤的身份，以及运行失败或在发布前被取消时执行的兜底步骤。
转换矩阵与该兜底条件于 2026-09-30 在本仓库用一次性 tag 运行实测，使用的是与 publisher 相同的 GitHub Actions token。以 `in_progress` 创建、以 `success` 完成的 run，接受 `status: in_progress` 和 `status: queued` 的更新时都返回 `200` 并保持 `completed`；只改 conclusion 的更新把它依次改为 `action_required`、`success` 和 `failure`。这些运行测的是 2026-09-30 当时的兜底条件（`failure() || (cancelled() && steps.publish.outcome != 'success')`）：兜底步骤在发布失败以及发布步骤执行期间被取消时运行，在决策发布之后才取消 job 时不运行。当前条件还要求 `concluded` output 为空，因此两种条件下，publish 步骤已记录结论后才取消的 job 都跳过兜底。`timeout-minutes` 取消路径于 2026-10-20 单独实测：一个在一次性 `push` 分支上超过一分钟超时的 job 仍然执行了 `cancelled()` 步骤，因此该步骤在这条路径上会执行；取消步骤的 file-command 输出是否会被暴露未实测，若被丢弃只会让该次取消以 `failure` 结束。
列表响应中的 `total_count` 统计的是经 `check_name` 过滤后的 run，而不是该 head 上的全部检查：2026-10-20 实测，一个带有 33 个 check run 的 head，对不匹配任何 run 的名称返回 `total_count` 0，对已存在的名称返回 2，因此用 `total_count` 与返回数组长度相等来拒绝被截断的分页是有效的。
## 考虑过的替代方案
**保留 `pending` commit status。** 分支规则能正确阻止合并，但 CI 监视工具无法区分等待批准与仍在运行的检查，因此不会结束。
**为尚未满足的条件发布 `failure`。** 该方案能结束等待，但会把尚未满足且可撤销的合并条件表示为故障，并混淆策略结果与 publisher 的健康状态。
**通过把已完成的 run 改回 `in_progress` 来撤销。** GitHub 接受该更新但不执行，因此已发布的成功会在依赖安装与历史拉取期间继续存在。
**在成功之前不发布任何结果。** 从未上报的必需检查保持 expected 状态，许多监视工具看不到它。撤销同一 head 上已发布的成功仍然需要写入非成功结果，而各监视工具如何对待缺失的必需检查没有规定。
**保留 `weighted approval` 名称。** 当 check 和 commit status 使用同一个必需名称时，GitHub 要求两者都通过。带有旧 publisher 所写 `pending` commit status 的 head 会一直被阻止，直到出现新的 push。
**每次评估都创建新的 check run。** 同一 head 上存在多个同名 check run 时，结果取决于 GitHub 和各监视工具如何在其中选择。
**允许批准点数覆盖阻塞性评审。** 该方案会让分数成为唯一的成功条件，但也允许在具有写权限的评审人仍然有效地要求修改时发布成功结果。
## 后果
决策等待评审人时，CI 监视工具会结束。`gh pr checks` 把 `action_required` 与失败的检查归为一组并以非零状态退出；消费方可以通过 check 名称、`ACTION_REQUIRED` conclusion 或携带分数或 blocker 的输出标题加以区分。`weighted approval publisher` Actions job 和 `failure` conclusion 承载运行故障信号。
重新评估已经发布 `success` 的 head 时，在新决策写入前会报告 `action_required`，因此在该窗口内轮询的监视工具会看到失败的检查。
分支规则切换后，对切换以来 publisher 尚未评估过的 head 存在两个缺口。check run 的名称并不标识发布者。被推送分支里的任何 workflow 都可以把一个 job 命名为 `weighted approval decision`：2026-10-20 在一次性分支上，带 `if: false` 的 job 在被推送的 commit 上产生了 `github-actions` 来源、该名称、conclusion 为 `skipped` 的 check run，而必需检查接受 `skipped`。因此 publisher 在依赖安装前撤销它找到的所有可通过的 conclusion，并在发布时接管所有同名 run，覆盖两次写入时已存在的 run；在 publisher 对该 head 最后一次写入之后新出现的同名可通过 run 不在其控制范围内。把该检查绑定到专用 GitHub App 是唯一完整的绑定方式，它会用 App 安装身份替换 job token。
没有该决策 check run 的 head 会把必需检查报告为 expected，直到下一次 publisher 事件创建它；重跑切换前的 run 不能完成迁移，因为重跑会回放该 run 当时的 workflow 文件，而它调用的 `pending` 阶段已被新脚本拒绝。2026-10-20 对切换前 run 的重跑执行的是切换前的步骤名，没有写入 check run。编辑 PR 会触发 publisher 的 `edited` 事件，因此 rollout 可以对没有该 run 的 head 制造这次事件。同时也带有先前 `weighted approval` commit status 的 head 会一直保留那个 pending status 直到同一事件：combined status 对每个 context 取最新的 status，因此 rollout 对每个开放 head 发布该 context 的终态 `success` status，而不是等它下一次 push。
.agents/skills/dsh-client-ui-ux/SKILL.md
+1 -1
View File
@@ -56,7 +56,7 @@ Scrollbars stay inside their container: use the shared scrollbar styles rather t
## Responsive layouts
Use [dsh-responsive-ui](../dsh-responsive-ui/SKILL.md) for responsive layouts, text truncation, title wrapping, icon-only controls, and focus during width changes.
Use [dsh-responsive-ui](../dsh-responsive-ui/SKILL.md) for responsive layouts, mobile and touch detection, text truncation, title wrapping, icon-only controls, and focus during width changes.
## Platform window adaptation
.agents/skills/dsh-code-review/SKILL.md
+1
View File
@@ -23,6 +23,7 @@ description: Use when reviewing a pull request in the deepseek-harness repo —
- [dsh-ci-test-reliability](../dsh-ci-test-reliability/SKILL.md): isolation and regression-proof rules for resource-owning, asynchronous, or flaky tests and fixtures.
- [dsh-client-ui-ux](../dsh-client-ui-ux/SKILL.md): feedback-surface choice, overlay safety, platform window adaptation, and visual token discipline for product-user-visible GUI changes.
- [dsh-error-handling](../dsh-error-handling/SKILL.md): failure classification, handling ownership, containment, state guarantees, and actionable reports.
- [dsh-session-projection](../dsh-session-projection/SKILL.md): projection ownership, state size, historical reads, and migrations.
- [docs/testing.md](../../../docs/testing.md) and the [quality-gates reference](../../../docs/testing.md): required test tiers and gates.
- [Agent Notes](../../notes/README.md): design rationale. Treat disagreement with an Agent Note as a design discussion, not an automatic veto.
- For bilingual changes, read [translation-rules.md](../../../docs/i18n/translation-rules.md) and [terminology.md](../../../docs/i18n/terminology.md); the extended translation skill is outside automatic review and runs only on explicit user invocation.
.agents/skills/dsh-pre-push-checks/SKILL.md
+1 -1
View File
@@ -124,7 +124,7 @@ For GitHub PRs, inspect remote CI after the push:
gh pr checks
```
Report pending checks as pending. Inspect failures before attributing them to the branch or the environment. A `weighted approval decision` check that `gh pr checks` lists as failed means the approval gate is waiting on reviewers only when its conclusion is `ACTION_REQUIRED`; the same check concluded `FAILURE` is a publisher fault, so read the `weighted approval publisher` job before attributing it to the branch.
Report pending checks as pending. Inspect failures before attributing them to the branch or the environment.
When `gh pr checks` reports "no checks reported" and `/actions/runs?head_sha=<sha>` returns `total_count: 0`, read mergeability before suspecting the push or a dropped GitHub event:
.agents/skills/dsh-responsive-ui/SKILL.md
+10 -5
View File
@@ -1,6 +1,6 @@
---
name: dsh-responsive-ui
description: Use when designing, implementing, or reviewing responsive client layouts, text truncation, title wrapping, icon-only controls, mobile hover, tooltips and keyboard behavior, and focus during width changes in DeepSeek Harness.
description: Use when designing, implementing, or reviewing responsive client layouts, mobile and touch detection, text truncation, title wrapping, icon-only controls, mobile hover, tooltips and keyboard behavior, and focus during width changes in DeepSeek Harness.
---
# Responsive client UI
@@ -39,10 +39,10 @@ Use a viewport width of 390 CSS px as the primary acceptance minimum. There must
## 6. Mobile hover behavior
1. Enable mouse-hover effects only on devices that support hovering. Use `@media (hover: hover)`, not viewport width, to determine this.
2. Tapping on a phone must not leave behind backgrounds, highlights, or action-button visibility intended only for hover.
3. Preserve normal selected, enabled, and pressed states, as well as focus indicators for keyboard interaction.
4. Try to preserve focus when the width changes, unless the focused element disappears from the page as a result.
1. Enable hover effects under `@media (hover: hover)`, independently of viewport width. This queries the primary pointing device, not the current event; JavaScript hover handlers must ignore touch events on mixed-input devices.
2. Tapping on a phone must not leave behind backgrounds, highlights, or action-button visibility intended only for hover. Without hover, actions remain visible or reachable through an explicit control. Keyboard users can reach and reveal the same actions.
3. Preserve normal selected, expanded, enabled, and pressed states, as well as focus indicators. Keep these styles independent of hover capability. Do not clear focus to remove a hover effect; `:focus-visible` can also apply to text fields focused by touch.
4. Preserve focus during width changes when the control remains visible and interactive. Avoid remounting controls only to change layout. If the focused control is hidden or removed, move focus to a suitable remaining control without opening the mobile keyboard unnecessarily.
## 7. Mobile tooltips
@@ -52,3 +52,8 @@ Use a viewport width of 390 CSS px as the primary acceptance minimum. There must
## 8. Mobile keyboard
On mobile, do not raise the input keyboard automatically unless it is explicitly intended.
## 9. Input capabilities and mobile text fields
1. Prefer CSS media features for presentation: `(pointer: coarse)` describes the primary pointer's accuracy, `(hover: none)` describes its inability to hover, and width queries select layout breaks. These queries do not detect a physical keyboard. Never branch layout on a parsed user agent.
2. Text fields take a 16px floor under `(pointer: coarse)` to avoid iOS Safari's focus zoom. Fine pointers keep the existing compact field sizes. Use the field rules in `packages/client/ui-plugin-manager/src/client/PluginManagerPage.module.css` as the reference.
.agents/skills/dsh-session-projection/SKILL.md
+19
View File
@@ -0,0 +1,19 @@
---
name: dsh-session-projection
description: Use when creating, extending, or reviewing a DeepSeek Harness session projection, or migrating a fold helper to one.
---
# DeepSeek Harness Session Projections
Follow the [projection API](../../../packages/session/session-projection/README.md) for registration and lifecycle details.
1. Decide whether to extend an existing projection or create one; check whether an existing owner already maintains the needed state.
2. Follow the README's [service ownership rules](../../../packages/session/session-projection/README.md#register-and-read): keep projection definitions and handles private to the owning service's implementation, and expose domain methods to consumers.
3. Require `sessionProjections` when the service needs projected state to function. Optional registration is for contributions the service can operate without.
4. Evaluate retained state size as the session grows. Do not accumulate state proportional to session length, including collections of event sequence numbers. Growth is acceptable only when slow enough to remain negligible. When the data would keep growing, consider incremental loading instead of retaining the complete state.
5. `apply` is synchronous; historical reads belong in asynchronous consumers. When later reads need event payloads, consider retaining sequence numbers within rule 4's state-size limits and letting consumers resolve them asynchronously.
6. When replacing a helper such as `foldSessionTitle` with a projection, remove the superseded helper.
7. After migrating to Session persistence, stop using `fold` in names and descriptions of the migrated logic.
8. Do not add calls to deprecated synchronous readers (`snapshotEvents`, `eventAt`, `ownEvents`) or wrappers exposing the same access ([decision](../../notes/implemented/architecture/2026-09-09-deprecate-synchronous-session-event-reads.md)). Assume the full event log is not resident in memory and random reads may require asynchronous disk access.
9. Consider state size, replay cost, and per-event work; add complexity only for a concrete performance need.
10. Asynchronous `register()` is usually called in `[Service.init]`. If event listeners are registered afterward, check initialization ordering: when needed, install listeners before awaiting registration; if a listener needs the returned handle, await its readiness in a callback that supports asynchronous waiting. Otherwise, an event may fire before the listener is installed and be missed. Determine whether this race exists from the actual initialization flow and event trigger paths; do not move listeners earlier or add waits mechanically.
.github/review-ownership/README.md
+6 -6
View File
@@ -16,11 +16,11 @@ The [`weighted-approval` workflow](../workflows/weighted-approval.yml) publishes
## Approval scoring
The weighted approval workflow exposes two pull-request checks. The `weighted approval publisher` Actions job reports whether evaluation and check publication completed, while the `weighted approval decision` check run carries the approval decision on the pull request head. Branch rules must require that check run with GitHub Actions as its expected source; a name-only requirement can accept a same-named check or status from another integration. The publisher updates the head's existing GitHub Actions decision check run and creates one only when none exists; a head without one starts `in_progress` and a re-evaluated head that published `success` is concluded `action_required` before Python setup or lexer installation, so failed setup or an interrupted history fetch cannot leave a previous success in place. A completed check run only ever changes conclusion, so every later write completes it again. Failures and cancellations that happen before the publish step records a conclusion conclude it as `failure`; a cancellation after that record keeps the published decision. A completed evaluation concludes `action_required` below two approval points, while the pull request is a draft, or while a write-capable reviewer has an effective `CHANGES_REQUESTED` review; the blocker keeps that conclusion even when counted approvals reach the threshold. It concludes `success` only when the threshold is met, the pull request is ready, and no such blocker exists. `action_required` blocks merging but completes the check, so tools that wait for checks to finish do not wait for reviewers; the check-run title carries the score or blocker.
The weighted approval workflow exposes two pull-request checks. The `weighted approval publisher` Actions job reports whether evaluation and status publication completed, while the `weighted approval` commit status carries the approval decision on the pull request head. Branch rules must require only the commit status with GitHub Actions as its expected source; a context-only requirement can accept a same-named status from another integration. The publisher marks the head pending before Python setup or lexer installation, so failed setup or an interrupted history fetch cannot leave a previous success in place. Setup and evaluation failures publish an error status. A completed evaluation returns `pending` below two approval points, while the pull request is a draft, or while a write-capable reviewer has an effective `CHANGES_REQUESTED` review; the blocker keeps the status pending even when counted approvals reach the threshold. It returns `success` only when the threshold is met, the pull request is ready, and no such blocker exists. If evaluation fails, the publisher writes an `error` status.
Reviewers whose calculated base repository permission is `write` or `admin` count. The [approval policy](approval-policy.json) gives `@07akioni`, `@imccyu`, `@tianyicui`, `@tianyicui-bot`, `@turtle1999`, and `@turtle2099` two points each; every other write-capable reviewer gets one point. The pull-request author’s own review and reviewers without write permission do not count.
The PR author contributes `min(100, mergedPRCount) × 11 / 1000` points: 0 merged PRs → 0 points, 50 → 0.55, and 100 or more → 1.1. The publisher counts only merged PRs in this repository using the author’s immutable account ID, excluding the current PR. It stops at 100 matches and rejects incomplete history responses. Counts refresh at the next subscribed evaluation event. Author credit is separate from reviewer approvals and cannot satisfy the two-point requirement alone; the author’s own review remains excluded. Human and bot authors, including Dependabot, use the same rule. Drafts, blocking reviews, and sufficient reviewer points skip history lookup; logs mark credit as not evaluated. Otherwise, logs show author credit and the capped count separately. Below the cap, counting may scan the repository’s entire merged history; a failed required lookup concludes the check as `failure`.
The PR author contributes `min(100, mergedPRCount) × 11 / 1000` points: 0 merged PRs → 0 points, 50 → 0.55, and 100 or more → 1.1. The publisher counts only merged PRs in this repository using the author’s immutable account ID, excluding the current PR. It stops at 100 matches and rejects incomplete history responses. Counts refresh at the next subscribed evaluation event. Author credit is separate from reviewer approvals and cannot satisfy the two-point requirement alone; the author’s own review remains excluded. Human and bot authors, including Dependabot, use the same rule. Drafts, blocking reviews, and sufficient reviewer points skip history lookup; logs mark credit as not evaluated. Otherwise, logs show author credit and the capped count separately. Below the cap, counting may scan the repository’s entire merged history; a failed required lookup publishes an error.
A one-point approval receives weight `min(2, 1 + 4 × ownedLines / totalLines)` from modified or deleted old production-code lines, attributed by `git blame` at the merge base of the live base branch and exact reviewed head. Ownership of 0%, 12.5%, and 25% gives 1, 1.5, and 2 points; higher ownership remains capped at 2. The success threshold remains 2 total points, without rounding the score. New lines do not enter the denominator, and an empty denominator gives no boost. Existing two-point weights remain unchanged. GitHub commit-author accounts identify reviewers across author emails; unlinked authors remain in the denominator without contributing to a reviewer. The publisher logs measured ownership. It skips attribution when reviewer points plus author credit already meet the threshold or a blocking review exists. Displayed scores use at most three decimal places; the decision uses unrounded scores with a tolerance of `1e-12` points for floating-point error. The curve endpoints come from the policy’s default and required points.
@@ -28,13 +28,13 @@ Production source means supported code files under `src/` in `packages/`, `apps/
Each reviewer contributes only the current `APPROVED` or `CHANGES_REQUESTED` decision that GitHub returns. A `DISMISSED` record clears that reviewer's standing decision, including earlier approvals. Comment-only and pending records do not replace a decision. Reviews from deleted accounts and reviewers without current repository access do not count. The workflow does not invalidate an approval by its review commit; the repository's native pull-request rules own stale-review and latest-push requirements.
The publisher runs when a pull request opens, synchronizes, reopens, becomes ready, becomes a draft, or is edited, including a base-branch change. Conversation comment events run the publisher only when the current or previous body contains `/delegate`. Closed or merged PRs and stale review heads are skipped before check-run writes or dependency setup; the workflow does not subscribe to master pushes. GitHub associates `workflow_run` publisher runs with the default branch even though decision check runs target the open PR head. Pull-request, review, and comment events share one concurrency group per PR. Review submissions, edits, and dismissals run the no-permission [`weighted-approval-review-event` workflow](../workflows/weighted-approval-review-event.yml); its validated run title supplies the pull-request number to the default-branch publisher. The publisher validates the current head, fetches every review and conversation comment, and resolves current repository permission before publishing the check run. Permission changes take effect on the next subscribed event.
The publisher runs when a pull request opens, synchronizes, reopens, becomes ready, becomes a draft, or is edited, including a base-branch change. Conversation comment events run the publisher only when the current or previous body contains `/delegate`. Closed or merged PRs and stale review heads are skipped before status writes or dependency setup; the workflow does not subscribe to master pushes. GitHub associates `workflow_run` publisher runs with the default branch even though approval statuses target the open PR head. Pull-request, review, and comment events share one concurrency group per PR. Review submissions, edits, and dismissals run the no-permission [`weighted-approval-review-event` workflow](../workflows/weighted-approval-review-event.yml); its validated run title supplies the pull-request number to the default-branch publisher. The publisher validates the current head, fetches every review and conversation comment, and resolves current repository permission before publishing the status. Permission changes take effect on the next subscribed event.
<a id="delegating-points"></a>
## Delegating points
Post `/delegate @username` as the entire text of a PR conversation comment to transfer your reviewer points to that user's effective `APPROVED` decision on this PR. For example, `/delegate @turtle1999` lets @turtle1999's approval count your points alongside their own. Until they approve, your points do not count, even if your own approval remains active. Each evaluation reconciles every active eligible command: it dismisses the sender's existing `APPROVED` and `CHANGES_REQUESTED` reviews, then requests review from recipients who have neither approved nor already been requested. This also handles commands posted on drafts once the PR becomes ready and commands whose original event was replaced in the concurrency queue. Other requested reviewers remain unchanged. The command does not submit a review; drafts do not dismiss or request reviews. Comment-only and pending reviews cannot be dismissed. A failed dismissal concludes the check as `failure` and prevents review requests; scores are refreshed after successful dismissal. Request failures are logged without changing the published score, and subsequent evaluations retry outstanding requests.
Post `/delegate @username` as the entire text of a PR conversation comment to transfer your reviewer points to that user's effective `APPROVED` decision on this PR. For example, `/delegate @turtle1999` lets @turtle1999's approval count your points alongside their own. Until they approve, your points do not count, even if your own approval remains active. Each evaluation reconciles every active eligible command: it dismisses the sender's existing `APPROVED` and `CHANGES_REQUESTED` reviews, then requests review from recipients who have neither approved nor already been requested. This also handles commands posted on drafts once the PR becomes ready and commands whose original event was replaced in the concurrency queue. Other requested reviewers remain unchanged. The command does not submit a review; drafts do not dismiss or request reviews. Comment-only and pending reviews cannot be dismissed. A failed dismissal publishes an error and prevents review requests; scores are refreshed after successful dismissal. Request failures are logged without changing the published score, and subsequent evaluations retry outstanding requests.
Both accounts must currently have write or admin permission, and neither may be the PR author. Commands involving ineligible accounts are ignored. A sender's newest surviving eligible, author-authorized command wins, in comment creation order; editing an older comment does not move it after newer comments. Submitting any review after delegation automatically takes the points back, including an approval, change request, or comment-only review. Submitting an inline diff comment with “Add single comment” also creates a comment-only review and reclaims the points. Pending reviews do not count; a later dismissal does not restore the delegation. When timestamps are equal, the review takes precedence. A new command after the review can delegate again; editing an older command cannot reactivate it. Post `/delegate @your-own-username` to restore your own review decision. Editing or deleting a command recomputes delegation from remaining comments, so deleting the latest command can restore an older one. Quoted commands, code blocks, review bodies, inline review comments, and commands mixed with other text do not count. Surrounding whitespace is allowed; account matching is case-insensitive. An edited command is accepted only when GitHub identifies the original author as its latest editor; edits by other accounts cannot delegate or cancel that author’s points.
@@ -44,7 +44,7 @@ Each sender contributes at most once, using their own fixed weight or production
## Security
All actions in the check-writing job are pinned to commit SHAs. The job checks out only the repository default branch. It does not check out or execute pull-request code and does not use repository secrets. Only when there is no blocker, reviewer points plus author credit are insufficient, and an approval has the policy’s default weight, it fetches complete history using the job token, passes Git objects to the trusted classifier as data, and resolves commit authors in batches of 50. Fetch credentials exist only in the Git child environment. Missing history, parsing failures, or incomplete author queries fail evaluation rather than producing a partial score. The review-event workflow has no `GITHUB_TOKEN` permissions and passes only a decimal pull-request number in its run title. The publisher accepts only successful `pull_request_review` runs from the review-event workflow file, identified by `workflow_run.path`; GitHub can populate `workflow_run.name` with the expanded run title. The publisher rejects an invalid run title and a number that does not resolve to the workflow run's current pull-request head. Pull-request reviews and comments are treated as API data and escaped in logs. All publisher triggers and steps, including dependency setup, have pull-request write permission; the policy uses it to dismiss eligible senders' prior decision reviews and request recipients. Every evaluation verifies command authors and latest editors by immutable account ID through GraphQL, matching the body to the REST comment history before counting points or acting. Missing or inconsistent editor metadata fails evaluation. Non-author edits are ignored. Only active eligible delegations are reconciled; superseded or revoked commands have no effects. Dismissal failures conclude the check as `failure`; request failures remain separate from the approval decision.
All actions in the status-writing job are pinned to commit SHAs. The job checks out only the repository default branch. It does not check out or execute pull-request code and does not use repository secrets. Only when there is no blocker, reviewer points plus author credit are insufficient, and an approval has the policy’s default weight, it fetches complete history using the job token, passes Git objects to the trusted classifier as data, and resolves commit authors in batches of 50. Fetch credentials exist only in the Git child environment. Missing history, parsing failures, or incomplete author queries fail evaluation rather than producing a partial score. The review-event workflow has no `GITHUB_TOKEN` permissions and passes only a decimal pull-request number in its run title. The publisher accepts only successful `pull_request_review` runs from the review-event workflow file, identified by `workflow_run.path`; GitHub can populate `workflow_run.name` with the expanded run title. The publisher rejects an invalid run title and a number that does not resolve to the workflow run's current pull-request head. Pull-request reviews and comments are treated as API data and escaped in logs. All publisher triggers and steps, including dependency setup, have pull-request write permission; the policy uses it to dismiss eligible senders' prior decision reviews and request recipients. Every evaluation verifies command authors and latest editors by immutable account ID through GraphQL, matching the body to the REST comment history before counting points or acting. Missing or inconsistent editor metadata fails evaluation. Non-author edits are ignored. Only active eligible delegations are reconciled; superseded or revoked commands have no effects. Dismissal failures publish an error; request failures remain separate from the approval decision.
Approval policy changes take effect only after they merge into the default branch. This prevents an untrusted pull request from changing the program or policy for its own run.
@@ -52,7 +52,7 @@ Approval policy changes take effect only after they merge into the default branc
## Verification
Run `pnpm run test:approval-policy` for policy parsing, effective review decisions, review-event validation, merged-history pagination and account matching, lazy author credit, permission filtering, weighted scoring, delegation conservation and revocation, editor authorization, comment pagination, dismissal and review-request reconciliation, blockers, drafts, closed-PR skips, check-run publication, and API failures. [Workflow tests](../../scripts/ci-workflow.spec.ts) pin the trusted checkout, no-permission review handoff, permissions, events, and commands. The repository gate graph runs the approval policy and workflow tests in CI. The Python SDK job runs `uv run --python 3.10 --with-requirements .github/review-ownership/requirements.txt python -m unittest discover -s .github/review-ownership -p 'test_*.py'` for real Git histories, lexers, renames, shallow-history rejection, and the publisher’s fetch/analysis integration.
Run `pnpm run test:approval-policy` for policy parsing, effective review decisions, review-event validation, merged-history pagination and account matching, lazy author credit, permission filtering, weighted scoring, delegation conservation and revocation, editor authorization, comment pagination, dismissal and review-request reconciliation, blockers, drafts, closed-PR skips, status publication, and API failures. [Workflow tests](../../scripts/ci-workflow.spec.ts) pin the trusted checkout, no-permission review handoff, permissions, events, and commands. The repository gate graph runs the approval policy and workflow tests in CI. The Python SDK job runs `uv run --python 3.10 --with-requirements .github/review-ownership/requirements.txt python -m unittest discover -s .github/review-ownership -p 'test_*.py'` for real Git histories, lexers, renames, shallow-history rejection, and the publisher’s fetch/analysis integration.
<a id="dev-note"></a>
.github/review-ownership/check-approval.mjs
+31 -79
View File
@@ -10,8 +10,7 @@ import { authorCreditPoints, countMergedAuthorPulls } from './author-weight.mjs'
const API_VERSION = '2026-03-10'
const MAX_PULL_REQUEST_RECORDS = 3_000
const PAGE_SIZE = 100
const CHECK_NAME = 'weighted approval decision'
const PASSING_CONCLUSIONS = new Set(['success', 'neutral', 'skipped'])
const STATUS_CONTEXT = 'weighted approval'
const WRITABLE_PERMISSIONS = new Set(['admin', 'write'])
const REVIEW_STATES = new Set(['APPROVED', 'CHANGES_REQUESTED', 'COMMENTED', 'DISMISSED', 'PENDING'])
@@ -210,13 +209,13 @@ function timestamp(value, subject) {
/**
* Evaluate approval points from current reviews and repository permissions.
* @param {{event: unknown, policySource: string, api: (path: string, options?: {method?: string, body?: unknown}) => Promise<unknown>, getOwnership?: typeof productionOwnership, getMergedCount?: typeof countMergedAuthorPulls}} options Runtime inputs.
* @returns {Promise<{pull: {repository: string, number: number, headSha: string}, state: 'action_required' | 'success', description: string, points: number, authorCredit: {mergedCount: number, points: number} | null, requiredPoints: number, approvals: Array<{login: string, points: number, delegatedTo?: string, ownership?: {ownedLines: number, totalLines: number}}>, delegations: Array<{login: string, delegatedTo: string, commentId: number, reviewIds: number[], approved: boolean}>, blockers: string[], ignoredReviewers: string[]}>} Approval decision and check-run output fields; approval login owns the points, delegatedTo supplies its decision, delegations contains eligible active commands and their superseded decision review IDs, and null author credit means history was not evaluated.
* @returns {Promise<{pull: {repository: string, number: number, headSha: string}, state: 'pending' | 'success', description: string, points: number, authorCredit: {mergedCount: number, points: number} | null, requiredPoints: number, approvals: Array<{login: string, points: number, delegatedTo?: string, ownership?: {ownedLines: number, totalLines: number}}>, delegations: Array<{login: string, delegatedTo: string, commentId: number, reviewIds: number[], approved: boolean}>, blockers: string[], ignoredReviewers: string[]}>} Approval decision and status payload fields; approval login owns the points, delegatedTo supplies its decision, delegations contains eligible active commands and their superseded decision review IDs, and null author credit means history was not evaluated.
*/
export async function evaluateApproval({ event, policySource, api, getOwnership = productionOwnership, getMergedCount = countMergedAuthorPulls }) {
const pull = pullRequestFromEvent(event)
const policy = parseApprovalPolicy(policySource)
if (pull.draft) {
return approvalResult(pull, policy.requiredPoints, [], [], [], 'action_required', 'draft pull request')
return approvalResult(pull, policy.requiredPoints, [], [], [], 'pending', 'draft pull request')
}
const reviews = await listPullRequestReviews(api, pull.repository, pull.number)
@@ -287,11 +286,11 @@ export async function evaluateApproval({ event, policySource, api, getOwnership
return next
}, authorCredit?.points ?? 0)
if (blockers.length > 0) {
return approvalResult(pull, policy.requiredPoints, approvals, blockers, ignoredReviewers, 'action_required',
return approvalResult(pull, policy.requiredPoints, approvals, blockers, ignoredReviewers, 'pending',
`${blockers.length} blocking change request${blockers.length === 1 ? '' : 's'}`, authorCredit, activeDelegations)
}
// Tolerate floating-point addition error without rounding approval scores.
const state = points + 1e-12 >= policy.requiredPoints ? 'success' : 'action_required'
const state = points + 1e-12 >= policy.requiredPoints ? 'success' : 'pending'
return approvalResult(
pull,
policy.requiredPoints,
@@ -306,12 +305,13 @@ export async function evaluateApproval({ event, policySource, api, getOwnership
}
/**
* Evaluate and publish the required check run, concluding it as failure when evaluation fails.
* @param {{event: unknown, policySource: string, api: (path: string, options?: {method?: string, body?: unknown}) => Promise<unknown>, runUrl: string, write?: (line: string) => void, record?: (conclusion: string) => void, getOwnership?: typeof productionOwnership, getMergedCount?: typeof countMergedAuthorPulls}} options Runtime inputs; record receives each conclusion the run publishes.
* Evaluate and publish the required commit status, publishing an error status when evaluation fails.
* @param {{event: unknown, policySource: string, api: (path: string, options?: {method?: string, body?: unknown}) => Promise<unknown>, runUrl: string, write?: (line: string) => void, getOwnership?: typeof productionOwnership, getMergedCount?: typeof countMergedAuthorPulls}} options Runtime inputs.
* @returns {Promise<Awaited<ReturnType<typeof evaluateApproval>>>} Published approval decision.
*/
export async function runApprovalCheck({ event, policySource, api, runUrl, getOwnership = productionOwnership, getMergedCount = countMergedAuthorPulls, write = line => process.stdout.write(`${line}\n`), record = () => {} }) {
export async function runApprovalCheck({ event, policySource, api, runUrl, getOwnership = productionOwnership, getMergedCount = countMergedAuthorPulls, write = line => process.stdout.write(`${line}\n`) }) {
const pull = pullRequestFromEvent(event)
await publishStatus(api, pull, 'pending', 'Evaluating approval points.', runUrl)
let result
try {
result = await evaluateApproval({ event, policySource, api, getOwnership, getMergedCount })
@@ -331,8 +331,7 @@ export async function runApprovalCheck({ event, policySource, api, runUrl, getOw
result = await evaluateApproval({ event, policySource, api, getOwnership, getMergedCount })
}
} catch (error) {
await publishCheck(api, pull, 'failure', 'Approval evaluation failed.', runUrl)
record('failure')
await publishStatus(api, pull, 'error', 'Approval evaluation failed.', runUrl)
throw error
}
write(result.authorCredit
@@ -343,9 +342,8 @@ export async function runApprovalCheck({ event, policySource, api, runUrl, getOw
`@${login}: ${points}${delegatedTo ? ` (delegated to @${delegatedTo})` : ''}${ownership ? ` (${ownership.ownedLines}/${ownership.totalLines} old production lines)` : ''}`))
writeList(write, 'Blocking change requests', result.blockers.map(login => `@${login}`))
writeList(write, 'Ignored reviewers without write access', result.ignoredReviewers.map(login => `@${login}`))
await publishCheck(api, pull, result.state, result.description, runUrl)
record(result.state)
write(`Published ${JSON.stringify(CHECK_NAME)} check run ${JSON.stringify(result.state)}.`)
await publishStatus(api, pull, result.state, result.description, runUrl)
write(`Published ${JSON.stringify(STATUS_CONTEXT)} status ${JSON.stringify(result.state)}.`)
try {
await requestDelegatedReviews(result, api, write)
} catch (error) {
@@ -355,15 +353,14 @@ export async function runApprovalCheck({ event, policySource, api, runUrl, getOw
}
/**
* Revoke a published success before dependency installation, or conclude a failed setup.
* @param {{event: unknown, api: (path: string, options?: {method?: string, body?: unknown}) => Promise<unknown>, runUrl: string, phase: string}} options Publication inputs.
* @returns {Promise<void>} Completion of the check-run write.
* Revoke a previous success before dependency installation, or report its failure.
* @param {{event: unknown, api: (path: string, options: object) => Promise<unknown>, runUrl: string, phase: string}} options Publication inputs.
* @returns {Promise<void>} Completion of the status write.
*/
export async function publishApprovalPhase({ event, api, runUrl, phase }) {
const pull = pullRequestFromEvent(event)
if (phase === 'revoke') return revokeApproval(api, pull, runUrl)
if (phase === 'failure') return publishCheck(api, pull, 'failure', 'Approval setup or evaluation failed.', runUrl)
throw new Error('invalid approval setup phase')
if (!['pending', 'error'].includes(phase)) throw new Error('invalid approval setup phase')
await publishStatus(api, pullRequestFromEvent(event), phase,
phase === 'pending' ? 'Preparing approval evaluation.' : 'Approval setup or evaluation failed.', runUrl)
}
/**
@@ -417,7 +414,7 @@ export async function approvalEventFromComment({ event, api }) {
}
/**
* Refresh a pull-request event before any check-run write; closed PRs and stale heads are skipped.
* Refresh a pull-request event before any status write; closed PRs and stale heads are skipped.
* @param {{event: unknown, api: (path: string) => Promise<unknown>}} options Pull-request event and API caller.
* @returns {Promise<Record<string, unknown> | null>} Live event, or null for a closed PR or superseded head.
*/
@@ -470,60 +467,18 @@ async function reviewerPermission(api, repository, login) {
return response.permission.toLowerCase()
}
// One decision per head keeps branch rules and CI watchers unambiguous; the head's existing GitHub
// Actions run is updated and a run is created only when none exists.
async function publishCheck(api, pull, conclusion, title, runUrl) {
requireHttpsRunUrl(runUrl)
const body = { status: 'completed', conclusion, details_url: runUrl, output: { title, summary: `${title} [Workflow run](${runUrl})` } }
const runs = await decisionRuns(api, pull)
if (runs.length === 0) {
await api(`/repos/${pull.repository}/check-runs`, { method: 'POST', body: { name: CHECK_NAME, head_sha: pull.headSha, ...body } })
return
}
for (const { id } of runs) {
if (!Number.isSafeInteger(id)) throw new Error('check run has no valid id')
await api(`/repos/${pull.repository}/check-runs/${id}`, { method: 'PATCH', body })
}
}
// A head without a decision starts in progress; a published success becomes action_required, so an
// interrupted evaluation cannot leave a success that its current inputs no longer support.
async function revokeApproval(api, pull, runUrl) {
requireHttpsRunUrl(runUrl)
const title = 'Preparing approval evaluation.'
const output = { title, summary: `${title} [Workflow run](${runUrl})` }
const runs = await decisionRuns(api, pull)
if (runs.length === 0) {
await api(`/repos/${pull.repository}/check-runs`, {
method: 'POST',
body: { name: CHECK_NAME, head_sha: pull.headSha, status: 'in_progress', details_url: runUrl, output },
})
return
}
for (const { id, conclusion } of runs) {
if (!Number.isSafeInteger(id)) throw new Error('check run has no valid id')
// Revoke every conclusion a required check accepts, not only the success this publisher writes:
// a workflow job added by the pull request can produce a same-named passing check run.
if (!PASSING_CONCLUSIONS.has(conclusion)) continue
await api(`/repos/${pull.repository}/check-runs/${id}`, {
method: 'PATCH',
body: { status: 'completed', conclusion: 'action_required', details_url: runUrl, output },
})
}
}
// A check run that reached completed never returns to in_progress: GitHub answers 200 and leaves the
// run completed, so a published decision can only be replaced by another conclusion.
async function decisionRuns(api, pull) {
const listed = await api(`/repos/${pull.repository}/commits/${pull.headSha}/check-runs?check_name=${encodeURIComponent(CHECK_NAME)}&filter=all&per_page=${PAGE_SIZE}`)
if (!isRecord(listed) || !Array.isArray(listed.check_runs) || listed.total_count !== listed.check_runs.length) {
throw new Error('check-run list response is invalid or incomplete')
}
return listed.check_runs.filter(run => isRecord(run) && isRecord(run.app) && run.app.slug === 'github-actions')
}
function requireHttpsRunUrl(runUrl) {
async function publishStatus(api, pull, state, description, runUrl) {
if (!/^https:\/\/[^\s]+$/u.test(runUrl)) throw new Error('workflow run URL must use HTTPS')
if (description.length > 140) throw new Error('commit status description exceeds 140 characters')
await api(`/repos/${pull.repository}/statuses/${pull.headSha}`, {
method: 'POST',
body: {
state,
context: STATUS_CONTEXT,
description,
target_url: runUrl,
},
})
}
function pullRequestFromEvent(event) {
@@ -616,7 +571,7 @@ async function main() {
const phase = process.argv[2]
if (phase) {
await publishApprovalPhase({ event, api, runUrl: process.env.GITHUB_RUN_URL ?? '', phase })
if (phase === 'revoke' && process.env.GITHUB_OUTPUT) appendFileSync(process.env.GITHUB_OUTPUT, 'active=true\n')
if (phase === 'pending' && process.env.GITHUB_OUTPUT) appendFileSync(process.env.GITHUB_OUTPUT, 'active=true\n')
return
}
await runApprovalCheck({
@@ -624,9 +579,6 @@ async function main() {
policySource,
api,
runUrl: process.env.GITHUB_RUN_URL ?? '',
record: (conclusion) => {
if (process.env.GITHUB_OUTPUT) appendFileSync(process.env.GITHUB_OUTPUT, `concluded=${conclusion}\n`)
},
})
}
.github/review-ownership/check-approval.test.mjs
+86 -173
View File
@@ -11,20 +11,14 @@ import {
evaluateApproval as evaluateWithHistory,
listPullRequestReviews,
parseApprovalPolicy,
runApprovalCheck as runPublisher,
runApprovalCheck as runWithHistory,
publishApprovalPhase,
} from './check-approval.mjs'
const policySource = readFileSync(new URL('approval-policy.json', import.meta.url), 'utf8')
const emptyCheckRuns = { total_count: 0, check_runs: [] }
const isCheckRunsList = path => path.includes('/commits/') && path.includes('/check-runs?')
const withoutCheckRuns = api => (path, options) =>
isCheckRunsList(path) ? emptyCheckRuns : api(path, options)
const withEmptyReads = api => withoutCheckRuns((path, options) => path.includes('/comments?') ? [] : api(path, options))
const evaluateApproval = options => evaluateWithHistory({ getMergedCount: async () => 0, ...options, api: withEmptyReads(options.api) })
const runWithHistory = options => runPublisher({ ...options, api: withoutCheckRuns(options.api) })
const runApprovalCheck = options => runWithHistory({ getMergedCount: async () => 0, ...options, api: withEmptyReads(options.api) })
const publishedState = body => body.conclusion ?? body.status
const withoutComments = api => (path, options) => path.includes('/comments?') ? [] : api(path, options)
const evaluateApproval = options => evaluateWithHistory({ getMergedCount: async () => 0, ...options, api: withoutComments(options.api) })
const runApprovalCheck = options => runWithHistory({ getMergedCount: async () => 0, ...options, api: withoutComments(options.api) })
const HEAD_SHA = '1234567890abcdef1234567890abcdef12345678'
const pullRequestEvent = ({ author = 'author', draft = false } = {}) => ({
@@ -204,7 +198,7 @@ test('accepts two one-point approvals and ignores reviews without write access',
assert.deepEqual(result.ignoredReviewers, ['reader'])
})
test('requires action for one one-point approval without failing the check', async () => {
test('keeps one one-point approval pending without failing the status', async () => {
const result = await evaluateApproval({
event: pullRequestEvent(),
policySource,
@@ -215,7 +209,7 @@ test('requires action for one one-point approval without failing the check', asy
throw new Error(`unexpected API path ${path}`)
},
})
assert.equal(result.state, 'action_required')
assert.equal(result.state, 'pending')
assert.equal(result.points, 1)
})
@@ -234,12 +228,12 @@ test('ignores a reviewer whose collaborator permission lookup returns 404', asyn
},
})
const result = await evaluateApproval({ event: pullRequestEvent(), policySource, api })
assert.equal(result.state, 'action_required')
assert.equal(result.state, 'pending')
assert.deepEqual(result.ignoredReviewers, ['former-writer'])
})
test('requires action on a write-capable change request while ignoring the author and read-only reviewers', async () => {
const writes = []
test('keeps the status pending on a write-capable change request while ignoring the author and read-only reviewers', async () => {
const statuses = []
const result = await runApprovalCheck({
event: pullRequestEvent({ author: 'author' }),
policySource,
@@ -257,43 +251,45 @@ test('requires action on a write-capable change request while ignoring the autho
if (path.includes('/collaborators/turtle1999/permission')) return { permission: 'admin' }
if (path.includes('/collaborators/blocker/permission')) return { permission: 'write' }
if (path.includes('/collaborators/reader/permission')) return { permission: 'read' }
if (isCheckRunsList(path)) return emptyCheckRuns
if (path.includes('/check-runs')) {
writes.push(options.body)
if (path.includes('/statuses/')) {
statuses.push(options.body)
return {}
}
throw new Error(`unexpected API path ${path}`)
},
write: () => {},
})
assert.equal(result.state, 'action_required')
assert.equal(result.state, 'pending')
assert.equal(result.description, '1 blocking change request.')
assert.equal(result.points, 2)
assert.deepEqual(result.blockers, ['blocker'])
assert.deepEqual(result.ignoredReviewers, ['reader'])
assert.deepEqual(writes, [{
name: 'weighted approval decision',
head_sha: HEAD_SHA,
status: 'completed',
conclusion: 'action_required',
details_url: 'https://github.example/actions/runs/1',
output: { title: '1 blocking change request.', summary: '1 blocking change request. [Workflow run](https://github.example/actions/runs/1)' },
assert.deepEqual(statuses, [{
state: 'pending',
context: 'weighted approval',
description: 'Evaluating approval points.',
target_url: 'https://github.example/actions/runs/1',
}, {
state: 'pending',
context: 'weighted approval',
description: '1 blocking change request.',
target_url: 'https://github.example/actions/runs/1',
}])
})
test('requires action on drafts without reading reviews', async () => {
test('keeps drafts pending without reading reviews', async () => {
const result = await evaluateApproval({
event: pullRequestEvent({ draft: true }),
policySource,
getOwnership: async () => ({ totalLines: 0, reviewerLines: {} }),
api: async () => { throw new Error('draft evaluation must not call GitHub') },
})
assert.equal(result.state, 'action_required')
assert.equal(result.state, 'pending')
assert.equal(result.points, 0)
assert.match(result.description, /draft pull request/u)
})
test('publishes the decision check run and concludes failure when evaluation fails', async () => {
test('publishes the required status and replaces stale success with error on evaluation failure', async () => {
const calls = []
const output = []
const result = await runApprovalCheck({
@@ -305,24 +301,21 @@ test('publishes the decision check run and concludes failure when evaluation fai
calls.push({ path, options })
if (path.includes('/reviews?')) return [review('turtle2099', 'APPROVED')]
if (path.includes('/collaborators/turtle2099/permission')) return { permission: 'write' }
if (isCheckRunsList(path)) return emptyCheckRuns
if (path.includes('/check-runs')) return {}
if (path.includes('/statuses/')) return {}
throw new Error(`unexpected API path ${path}`)
},
write: line => output.push(line),
})
assert.equal(result.state, 'success')
assert.deepEqual(calls.at(-1), {
path: '/repos/deepseek-harness/deepseek-harness/check-runs',
path: `/repos/deepseek-harness/deepseek-harness/statuses/${HEAD_SHA}`,
options: {
method: 'POST',
body: {
name: 'weighted approval decision',
head_sha: HEAD_SHA,
status: 'completed',
conclusion: 'success',
details_url: 'https://github.example/actions/runs/1',
output: { title: '2/2 approval points.', summary: '2/2 approval points. [Workflow run](https://github.example/actions/runs/1)' },
state: 'success',
context: 'weighted approval',
description: '2/2 approval points.',
target_url: 'https://github.example/actions/runs/1',
},
},
})
@@ -337,8 +330,7 @@ test('publishes the decision check run and concludes failure when evaluation fai
runUrl: 'https://github.example/actions/runs/2',
api: async (path, options = {}) => {
if (path.includes('/reviews?')) throw new Error('reviews unavailable')
if (isCheckRunsList(path)) return emptyCheckRuns
if (path.includes('/check-runs')) {
if (path.includes('/statuses/')) {
failures.push({ path, options })
return {}
}
@@ -346,10 +338,9 @@ test('publishes the decision check run and concludes failure when evaluation fai
},
write: () => {},
}), /reviews unavailable/u)
assert.deepEqual(failures.map(({ path }) => path), ['/repos/deepseek-harness/deepseek-harness/check-runs'])
assert.equal(failures[0].options.body.status, 'completed')
assert.equal(failures[0].options.body.conclusion, 'failure')
assert.equal(failures[0].options.body.output.title, 'Approval evaluation failed.')
assert.equal(failures[0].options.body.state, 'pending')
assert.equal(failures[1].options.body.state, 'error')
assert.equal(failures[1].options.body.description, 'Approval evaluation failed.')
})
test('sends authenticated JSON and escapes an API error body', async () => {
@@ -392,7 +383,7 @@ for (const [ownedLines, totalLines, expectedPoints] of [[9, 100, 1.3599999999999
})
assert.equal(result.points, expectedPoints)
assert.equal(result.description, `${Number(expectedPoints.toFixed(3))}/2 approval points (author 0).`)
assert.equal(result.state, expectedPoints === 2 ? 'success' : 'action_required')
assert.equal(result.state, expectedPoints === 2 ? 'success' : 'pending')
assert.equal(measurements, 1)
assert.deepEqual(result.approvals[0].ownership, { ownedLines, totalLines })
})
@@ -412,7 +403,7 @@ test('does not fetch ownership when a change request blocks approval', async ()
})
assert.equal(measurements, 0)
assert.equal(result.points, 2)
assert.equal(result.state, 'action_required')
assert.equal(result.state, 'pending')
})
test('does not fetch history when approvals already have two-point weights', async () => {
@@ -424,7 +415,7 @@ test('does not fetch history when approvals already have two-point weights', asy
assert.equal(result.points, 2)
})
test('concludes failure when production attribution fails', async () => {
test('publishes error when production attribution fails', async () => {
const states = []
await assert.rejects(runApprovalCheck({
event: pullRequestEvent(), policySource, runUrl: 'https://github.example/run/1',
@@ -433,46 +424,30 @@ test('concludes failure when production attribution fails', async () => {
if (path.includes('/reviews?')) return [review('writer', 'APPROVED')]
if (path.includes('/permission')) return { permission: 'write' }
if (path.includes('/comments?')) return []
if (isCheckRunsList(path)) return emptyCheckRuns
states.push(publishedState(options.body))
states.push(options.body.state)
return {}
},
}), /incomplete history/u)
assert.deepEqual(states, ['failure'])
assert.deepEqual(states, ['pending', 'error'])
})
test('revokes a published success without returning a completed run to in progress', async () => {
const calls = []
const checkRuns = [
{ id: 7, app: { slug: 'github-actions' }, conclusion: 'success' },
{ id: 8, app: { slug: 'github-actions' }, conclusion: 'action_required' },
{ id: 9, app: { slug: 'other-app' }, conclusion: 'success' },
{ id: 10, app: { slug: 'github-actions' }, conclusion: 'skipped' },
{ id: 11, app: { slug: 'github-actions' }, conclusion: 'neutral' },
{ id: 12, app: { slug: 'github-actions' }, conclusion: null },
]
await publishApprovalPhase({
event: pullRequestEvent(), runUrl: 'https://github.example/run/1', phase: 'revoke',
test('revokes a previous success before starting expensive attribution', async () => {
const states = []
await runApprovalCheck({
event: pullRequestEvent(), policySource, runUrl: 'https://github.example/run/1', write: () => {},
getOwnership: async () => {
assert.deepEqual(states, ['pending'])
return { totalLines: 100, reviewerLines: { writer: 25 } }
},
api: async (path, options) => {
calls.push({ path, method: options?.method, body: options?.body })
return isCheckRunsList(path) ? { total_count: checkRuns.length, check_runs: checkRuns } : {}
if (path.includes('/reviews?')) return [review('writer', 'APPROVED')]
if (path.includes('/permission')) return { permission: 'write' }
states.push(options.body.state)
return {}
},
})
assert.deepEqual(calls.map(({ path }) => path), [
`/repos/deepseek-harness/deepseek-harness/commits/${HEAD_SHA}/check-runs?check_name=weighted%20approval%20decision&filter=all&per_page=100`,
'/repos/deepseek-harness/deepseek-harness/check-runs/7',
'/repos/deepseek-harness/deepseek-harness/check-runs/10',
'/repos/deepseek-harness/deepseek-harness/check-runs/11',
])
for (const call of calls.slice(1)) {
assert.deepEqual(call.body, {
status: 'completed',
conclusion: 'action_required',
details_url: 'https://github.example/run/1',
output: { title: 'Preparing approval evaluation.', summary: 'Preparing approval evaluation. [Workflow run](https://github.example/run/1)' },
})
}
assert.deepEqual(states, ['pending', 'success'])
})
for (const reviewers of [['first', 'second'], ['turtle1999', 'first']]) {
@@ -502,67 +477,14 @@ test('publishes setup phases without evaluating or installing dependencies', asy
const states = []
const options = {
event: pullRequestEvent(), runUrl: 'https://github.example/run/1',
api: withEmptyReads(async (path, { body }) => { assert.match(path, /\/check-runs$/u); states.push(publishedState(body)) }),
api: async (path, { body }) => { assert.match(path, /\/statuses\//u); states.push(body.state) },
}
await publishApprovalPhase({ ...options, phase: 'revoke' })
await publishApprovalPhase({ ...options, phase: 'failure' })
assert.deepEqual(states, ['in_progress', 'failure'])
await publishApprovalPhase({ ...options, phase: 'pending' })
await publishApprovalPhase({ ...options, phase: 'error' })
assert.deepEqual(states, ['pending', 'error'])
await assert.rejects(publishApprovalPhase({ ...options, phase: 'success' }), /invalid approval setup phase/u)
})
test('updates the head\'s GitHub Actions decision check run instead of creating another', async () => {
const calls = []
const checkRuns = [
{ id: 7, app: { slug: 'github-actions' }, conclusion: 'action_required' },
{ id: 8, app: { slug: 'other-app' }, conclusion: 'action_required' },
]
await publishApprovalPhase({
event: pullRequestEvent(), runUrl: 'https://github.example/run/1', phase: 'failure',
api: async (path, options) => {
calls.push({ path, method: options?.method })
return { total_count: checkRuns.length, check_runs: checkRuns }
},
})
assert.deepEqual(calls, [
{ path: `/repos/deepseek-harness/deepseek-harness/commits/${HEAD_SHA}/check-runs?check_name=weighted%20approval%20decision&filter=all&per_page=100`, method: undefined },
{ path: '/repos/deepseek-harness/deepseek-harness/check-runs/7', method: 'PATCH' },
])
for (const response of [{}, { total_count: 101, check_runs: [] }, { total_count: 1, check_runs: [{ id: '7', app: { slug: 'github-actions' } }] }]) {
await assert.rejects(publishApprovalPhase({
event: pullRequestEvent(), runUrl: 'https://github.example/run/1', phase: 'failure', api: async () => response,
}), /check-run list response|valid id/u)
}
})
test('records each published conclusion so the failure step never replaces it', async () => {
const recorded = []
const api = async (path, options) => {
if (isCheckRunsList(path)) return emptyCheckRuns
if (path.includes('/reviews?')) return [review('turtle2099', 'APPROVED')]
if (path.includes('/permission')) return { permission: 'write' }
if (path.includes('/comments?')) return []
assert.match(path, /\/check-runs$/u)
return {}
}
await runApprovalCheck({
event: pullRequestEvent(), policySource, runUrl: 'https://github.example/run/1', api, write: () => {},
record: conclusion => recorded.push(conclusion),
})
await assert.rejects(runApprovalCheck({
event: pullRequestEvent(), policySource, runUrl: 'https://github.example/run/2', write: () => {},
getMergedCount: async () => { throw new Error('history unavailable') },
record: conclusion => recorded.push(conclusion),
api: async path => {
if (isCheckRunsList(path)) return emptyCheckRuns
if (path.includes('/reviews?')) return []
if (path.includes('/comments?')) return []
return {}
},
}), /history unavailable/u)
assert.deepEqual(recorded, ['success', 'failure'])
})
for (const [mergedCount, credit] of [[0, 0], [1, 0.011], [50, 0.55], [99, 1.089], [100, 1.1], [101, 1.1], [200, 1.1]]) {
test(`author with ${mergedCount} merged PRs contributes ${credit} points but cannot approve alone`, async () => {
const result = await evaluateApproval({
@@ -571,7 +493,7 @@ for (const [mergedCount, credit] of [[0, 0], [1, 0.011], [50, 0.55], [99, 1.089]
})
assert.equal(result.points, credit)
assert.equal(result.authorCredit.points, credit)
assert.equal(result.state, 'action_required')
assert.equal(result.state, 'pending')
assert.deepEqual(result.approvals, [])
})
}
@@ -587,7 +509,7 @@ test('combines author credit and reviewer ownership at the passing threshold', a
}
})
for (const [mergedCount, state] of [[90, 'action_required'], [91, 'success'], [100, 'success']]) {
for (const [mergedCount, state] of [[90, 'pending'], [91, 'success'], [100, 'success']]) {
test(`one ordinary approval with ${mergedCount} merged PRs is ${state}`, async () => {
const result = await evaluateApproval({
event: pullRequestEvent(), policySource, getMergedCount: async () => mergedCount,
@@ -610,14 +532,14 @@ test('blockers skip history and ignore non-write reviewers', async () => {
? [review('writer', 'APPROVED'), review('blocker', 'CHANGES_REQUESTED'), review('reader', 'APPROVED')]
: { permission: path.includes('/reader/') ? 'read' : 'write' },
})
assert.equal(result.state, 'action_required')
assert.equal(result.state, 'pending')
assert.equal(result.points, 1)
assert.equal(result.authorCredit, null)
assert.deepEqual(result.blockers, ['blocker'])
assert.deepEqual(result.ignoredReviewers, ['reader'])
})
test('drafts skip history and a failed history lookup concludes the check as failure', async () => {
test('drafts skip history and history failures revoke success with error', async () => {
await evaluateApproval({
event: pullRequestEvent({ draft: true }), policySource,
getMergedCount: async () => { throw new Error('draft must skip history') },
@@ -628,20 +550,20 @@ test('drafts skip history and a failed history lookup concludes the check as fai
getMergedCount: async () => { throw new Error('history unavailable') },
api: async (path, options) => {
if (path.includes('/reviews?')) return []
states.push(publishedState(options.body))
states.push(options.body.state)
},
}), /history unavailable/u)
assert.deepEqual(states, ['failure'])
assert.deepEqual(states, ['pending', 'error'])
})
test('a score just below the threshold requires action even if its display rounds to two', async () => {
test('a score just below the threshold remains pending even if its display rounds to two', async () => {
const result = await evaluateApproval({
event: pullRequestEvent(), policySource, getMergedCount: async () => 50,
getOwnership: async () => ({ totalLines: 1000000, reviewerLines: { writer: 112499 } }),
api: async path => path.includes('/reviews?') ? [review('writer', 'APPROVED')] : { permission: 'write' },
})
assert.equal(result.state, 'action_required')
assert.equal(result.state, 'pending')
assert.ok(result.points < 2)
})
@@ -665,14 +587,14 @@ test('the publisher counts merged history through the production API path', asyn
}
if (path.includes('/reviews?')) return [review('writer', 'APPROVED')]
if (path.includes('/permission')) return { permission: 'write' }
states.push(publishedState(options.body))
states.push(options.body.state)
return {}
},
})
assert.equal(result.points, 2)
assert.equal(result.authorCredit.points, 0.55)
assert.equal(result.approvals[0].points, 1.45)
assert.deepEqual(states, ['success'])
assert.deepEqual(states, ['pending', 'success'])
assert.equal(output[0], 'Author credit: 0.55 (50 merged PRs).')
})
@@ -690,7 +612,7 @@ test('sufficient reviewer points and drafts publish without querying author hist
return {}
},
})
assert.equal(result.state, draft ? 'action_required' : 'success')
assert.equal(result.state, draft ? 'pending' : 'success')
assert.equal(result.authorCredit, null)
assert.equal(output[0], `Author credit: not evaluated (${draft ? 'draft' : 'reviewer points suffice'}).`)
}
@@ -756,7 +678,7 @@ test('delegates the sender fixed points once after the recipient approves, retai
}) })
const points = recipientState === 'APPROVED' ? 3 : 0
assert.equal(result.points, points)
assert.equal(result.state, points === 3 && senderState !== 'CHANGES_REQUESTED' ? 'success' : 'action_required')
assert.equal(result.state, points === 3 && senderState !== 'CHANGES_REQUESTED' ? 'success' : 'pending')
assert.deepEqual(result.blockers, [
...(senderState === 'CHANGES_REQUESTED' ? ['turtle1999'] : []),
...(recipientState === 'CHANGES_REQUESTED' ? ['writer'] : []),
@@ -864,7 +786,7 @@ test('a sender submitted review takes back delegation, including comment-only an
}) })
assert.equal(result.delegations.length, state === 'PENDING' ? 1 : 0)
assert.equal(result.points, state === 'APPROVED' || state === 'PENDING' ? 3 : 1)
assert.equal(result.state, state === 'APPROVED' || state === 'PENDING' ? 'success' : 'action_required')
assert.equal(result.state, state === 'APPROVED' || state === 'PENDING' ? 'success' : 'pending')
if (state === 'APPROVED') assert.equal(result.approvals.find(({ login }) => login === 'Turtle1999').delegatedTo, undefined)
}
}
@@ -907,8 +829,7 @@ test('an active delegate command requests review once and preserves other reques
if (options?.method === 'POST') { requests.push(options.body); return {} }
return { users: [{ login: 'another-reviewer' }, ...(alreadyRequested ? [{ login: 'WRITER' }] : [])], teams: [] }
}
if (isCheckRunsList(path)) return emptyCheckRuns
if (path.includes('/check-runs')) return {}
if (path.includes('/statuses/')) return {}
return api(path, options)
},
})
@@ -936,8 +857,7 @@ test('inactive commands and drafts do not request review', async () => {
policySource, runUrl: 'https://github.example/run/1', getMergedCount: async () => 0, write: () => {},
api: async (path, options) => {
assert.ok(!path.endsWith('/requested_reviewers'))
if (isCheckRunsList(path)) return emptyCheckRuns
if (path.includes('/check-runs')) return {}
if (path.includes('/statuses/')) return {}
return api(path, options)
},
})
@@ -953,8 +873,7 @@ test('failed review requests preserve the computed score and report the retry',
event: { ...pullRequestEvent(), issue: { number: 42, pull_request: {} }, action: 'created', comment: { id: 1 } },
policySource, runUrl: 'https://github.example/run/1', getMergedCount: async () => 0, getOwnership: async () => ({ totalLines: 0, reviewerLines: {} }), write: line => output.push(line),
api: async (path, options) => {
if (isCheckRunsList(path)) return emptyCheckRuns
if (path.includes('/check-runs')) { statuses.push(publishedState(options.body)); return {} }
if (path.includes('/statuses/')) { statuses.push(options.body.state); return {} }
if (path.endsWith('/requested_reviewers')) {
if (options?.method === 'POST') throw new Error('request rejected')
return failedRead ? {} : { users: [] }
@@ -964,7 +883,7 @@ test('failed review requests preserve the computed score and report the retry',
})
assert.match(output.at(-1), failedRead ? /no users array/u : /request rejected/u)
assert.match(output.at(-1), /score is unchanged.*will retry/u)
assert.deepEqual(statuses, ['success'])
assert.deepEqual(statuses, ['pending', 'success'])
}
})
@@ -976,8 +895,7 @@ test('delegation dismisses only the sender old decisions and stays active across
const states = []
const read = delegationApi({ comments: [comment('turtle1999', '/delegate @writer')], reviews })
const api = async (path, options) => {
if (isCheckRunsList(path)) return emptyCheckRuns
if (path.includes('/check-runs')) { states.push(publishedState(options.body)); return {} }
if (path.includes('/statuses/')) { states.push(options.body.state); return {} }
if (path.endsWith('/dismissals')) {
calls.push(options)
const id = Number(path.split('/').at(-2))
@@ -1001,13 +919,13 @@ test('delegation dismisses only the sender old decisions and stays active across
assert.deepEqual(result.blockers, ['other'])
assert.equal(result.delegations.length, 1)
assert.deepEqual(result.delegations[0].reviewIds, [])
assert.deepEqual(states, ['action_required'])
assert.deepEqual(states, ['pending', 'pending'])
reviews.find(review => review.id === 21).state = 'DISMISSED'
for (const state of ['APPROVED', 'DISMISSED', 'COMMENTED', 'APPROVED']) {
reviews.find(review => review.id === 20).state = state
const next = await evaluateDelegation({ api })
assert.equal(next.delegations.length, 1)
assert.equal(next.state, state === 'APPROVED' ? 'success' : 'action_required')
assert.equal(next.state, state === 'APPROVED' ? 'success' : 'pending')
}
reviews.push(review('turtle1999', 'COMMENTED', '2026-09-16T00:00:00Z', 30))
const reclaimed = await evaluateDelegation({ api })
@@ -1021,8 +939,7 @@ test('a new sender review arriving during dismissal is not dismissed or delegate
const result = await runWithHistory({ event: { ...pullRequestEvent(), issue: { number: 42, pull_request: {} }, action: 'created', comment: { id: 1 } },
policySource, runUrl: 'https://github.example/run/1', write: () => {}, getMergedCount: async () => 0,
api: async (path, options) => {
if (isCheckRunsList(path)) return emptyCheckRuns
if (path.includes('/check-runs')) return {}
if (path.includes('/statuses/')) return {}
if (path.endsWith('/dismissals')) {
assert.ok(path.endsWith('/reviews/10/dismissals'))
reviews[0].state = 'DISMISSED'
@@ -1037,15 +954,14 @@ test('a new sender review arriving during dismissal is not dismissed or delegate
assert.deepEqual(result.blockers, ['turtle1999'])
})
test('a failed or unconfirmed dismissal prevents review requests and concludes failure', async () => {
test('a failed or unconfirmed dismissal prevents review requests and publishes error', async () => {
for (const rejected of [false, true]) {
const states = []
const read = delegationApi({ comments: [comment('turtle1999', '/delegate @writer')], reviews: [review('turtle1999', 'CHANGES_REQUESTED')] })
await assert.rejects(runWithHistory({ event: { ...pullRequestEvent(), issue: { number: 42, pull_request: {} }, action: 'created', comment: { id: 1 } },
policySource, runUrl: 'https://github.example/run/1', write: () => {},
api: async (path, options) => {
if (isCheckRunsList(path)) return emptyCheckRuns
if (path.includes('/check-runs')) { states.push(publishedState(options.body)); return {} }
if (path.includes('/statuses/')) { states.push(options.body.state); return {} }
if (path.endsWith('/dismissals')) {
if (rejected) throw new Error('dismissal denied')
return { id: 10, state: 'CHANGES_REQUESTED' }
@@ -1054,7 +970,7 @@ test('a failed or unconfirmed dismissal prevents review requests and concludes f
return read(path, options)
},
}), rejected ? /dismissal denied/u : /dismissal was not confirmed/u)
assert.deepEqual(states, ['failure'])
assert.deepEqual(states, ['pending', 'error'])
}
})
@@ -1067,8 +983,7 @@ test('reads all comment pages, logs the score owner, and fails closed on missing
event: pullRequestEvent(), policySource, runUrl: 'https://github.example/run/1',
getMergedCount: async () => 0, write: line => output.push(line),
api: async (path, options) => {
if (isCheckRunsList(path)) return emptyCheckRuns
if (path.includes('/check-runs')) { statuses.push(publishedState(options.body)); return {} }
if (path.includes('/statuses/')) { statuses.push(options.body.state); return {} }
if (path.includes('/comments?')) return comments(path)
return api(path, options)
},
@@ -1081,7 +996,7 @@ test('reads all comment pages, logs the score owner, and fails closed on missing
})
assert.equal(pages.length, 2)
assert.ok(output.includes('- @turtle1999: 2 (delegated to @writer)'))
assert.deepEqual(statuses.splice(0), ['success'])
assert.deepEqual(statuses.splice(0), ['pending', 'success'])
for (const [comments, message] of [
[() => { throw new Error('comment API unavailable') }, /comment API unavailable/u],
[() => ({}), /comments response is not an array/u],
@@ -1091,7 +1006,7 @@ test('reads all comment pages, logs the score owner, and fails closed on missing
[() => Array.from({ length: 100 }, () => comment('writer', 'text')), /comments exceed 3000/u],
]) {
await assert.rejects(run(comments), message)
assert.deepEqual(statuses.splice(0), ['failure'])
assert.deepEqual(statuses.splice(0), ['pending', 'error'])
}
})
@@ -1175,8 +1090,7 @@ test('later events reconcile all active delegations after a draft or replaced co
const effects = []
const requested = []
const api = async (path, options) => {
if (isCheckRunsList(path)) return emptyCheckRuns
if (path.includes('/check-runs')) return {}
if (path.includes('/statuses/')) return {}
if (path.endsWith('/dismissals')) {
const old = reviews.find(review => path.endsWith(`/reviews/${review.id}/dismissals`))
effects.push(old.id)
@@ -1208,8 +1122,7 @@ test('already approved delegates need no new request', async () => {
})
const result = await runWithHistory({ event: pullRequestEvent(), policySource, runUrl: 'https://github.example/run/1', write: () => {},
api: (path, options) => {
if (isCheckRunsList(path)) return emptyCheckRuns
if (path.includes('/check-runs')) return {}
if (path.includes('/statuses/')) return {}
assert.ok(!path.endsWith('/requested_reviewers'))
return read(path, options)
},
.github/workflows/weighted-approval.yml
+8 -12
View File
@@ -10,9 +10,9 @@ on:
types: [completed]
permissions:
checks: write
contents: read
pull-requests: write
statuses: write
concurrency:
group: weighted-approval-${{ (github.event.pull_request.number || github.event.issue.number) && format('weighted-approval-review-event:{0}', github.event.pull_request.number || github.event.issue.number) || github.event.workflow_run.display_title }}
@@ -29,20 +29,20 @@ jobs:
runs-on: ubuntu-latest
timeout-minutes: 5
steps:
# SECURITY: the check-writing job executes policy from the trusted default
# SECURITY: the status-writing job executes policy from the trusted default
# branch and reads pull-request reviews and comments only as API data.
- name: Check out trusted approval policy
uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1
with:
ref: ${{ github.event.repository.default_branch }}
persist-credentials: false
- name: Revoke previous approval decision
- name: Revoke previous approval status
id: revoke
env:
GITHUB_TOKEN: ${{ github.token }}
GITHUB_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
run: node .github/review-ownership/check-approval.mjs revoke
# SECURITY: dependency setup runs in the check-writing job.
run: node .github/review-ownership/check-approval.mjs pending
# SECURITY: dependency setup runs in the status-writing job.
- if: steps.revoke.outputs.active == 'true'
uses: actions/setup-python@ece7cb06caefa5fff74198d8649806c4678c61a1 # v6.3.0
with:
@@ -52,19 +52,15 @@ jobs:
- name: Install production lexer
if: steps.revoke.outputs.active == 'true'
run: python3 -m pip install -r .github/review-ownership/requirements.txt
- name: Publish weighted approval decision
id: publish
- name: Publish weighted approval status
if: steps.revoke.outputs.active == 'true'
env:
GITHUB_TOKEN: ${{ github.token }}
GITHUB_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
run: node .github/review-ownership/check-approval.mjs
# A failed or cancelled run must not leave an earlier conclusion standing. The publish step
# records the conclusion it wrote, so a decision already published survives a later failure or
# cancellation, and a cancelled publish step that wrote nothing still concludes failure.
- name: Publish approval setup failure
if: (failure() || cancelled()) && steps.publish.outputs.concluded == ''
if: failure() && steps.revoke.outputs.active == 'true'
env:
GITHUB_TOKEN: ${{ github.token }}
GITHUB_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
run: node .github/review-ownership/check-approval.mjs failure
run: node .github/review-ownership/check-approval.mjs error
AGENTS.md
+1
View File
@@ -135,6 +135,7 @@ Real-API tests/demos read `DEEPSEEK_API_KEY`, optional `DEEPSEEK_BASE_URL`, and
- **Switch on discriminant tags.** Closed unions end in `assertNever`; merge-extensible unions fall through a documented default.
- **Waterfall listeners MUST call `next()`** to delegate; returning without it short-circuits the chain ([semantics](docs/cordis-primer.md#cordis-waterfall-semantics)).
- **Model-visible ⟺ logged**: anything that reaches a model request must be reconstructable from the session log; a new model-visible input requires a session event.
- **Session projections:** follow [dsh-session-projection](.agents/skills/dsh-session-projection/SKILL.md) when creating, extending, or reviewing one, or migrating a fold helper to one.
- **Plugins, not loop changes**: new behavior goes on documented extension points; changing `agent-loop` requires updating docs/architecture.md.
- **A capability seam comprises Service Definition / Service Provider / Consumer roles.** It is complete, never one role; split only when roles evolve independently ([glossary](docs/glossary.md#capability-seam)).
- **Prefer maintained dependencies over hand-rolling** when they genuinely delete owned code and tests ([policy](.agents/notes/implemented/process/2026-07-26-dependencies-over-hand-rolling.md)).
apps/cli/composition.md
-3
View File
@@ -42,8 +42,6 @@ flowchart LR
cfg --> plugin_dsh_base_agent
plugin_dsh_base_agent_model["agent-model<br/>@deepseek-ai/dsh-agent-model"]
cfg --> plugin_dsh_base_agent_model
plugin_dsh_base_agent_controller["agent-controller<br/>@deepseek-ai/dsh-agent-controller"]
cfg --> plugin_dsh_base_agent_controller
plugin_dsh_base_plugin_package_inventory_deepseek["plugin-package-inventory-deepseek<br/>@deepseek-ai/dsh-plugin-package-inventory-deepseek"]
cfg --> plugin_dsh_base_plugin_package_inventory_deepseek
plugin_dsh_base_agent_default_model["agent-default-model<br/>@deepseek-ai/dsh-agent-default-model"]
@@ -223,7 +221,6 @@ flowchart LR
| `user-questions` | `@deepseek-ai/dsh-user-questions` |
| `agent` | `@deepseek-ai/dsh-agent` |
| `agent-model` | `@deepseek-ai/dsh-agent-model` |
| `agent-controller` | `@deepseek-ai/dsh-agent-controller` |
| `plugin-package-inventory-deepseek` | `@deepseek-ai/dsh-plugin-package-inventory-deepseek` |
| `agent-default-model` | `@deepseek-ai/dsh-agent-default-model` |
| `jobs` | `@deepseek-ai/dsh-jobs-local` |
apps/web/tests/README.i18n.yaml
+2 -2
View File
@@ -9,8 +9,8 @@
en: 1f9c004f3eb465fb
zh: 3e45ef8014bad48d
/apps-web-browser-e2e/these-are-host-face-tests:
en: 8c406d423c8e2137
zh: a66a144fd146f56c
en: 101d687b1d98c5b7
zh: f8c31115e8a15be7
/apps-web-browser-e2e/do-not-import-deepseek-ai-dsh-client-here:
en: 7cc75a6413647142
zh: a8c541f9b93e9f7f
apps/web/tests/README.md
+2
View File
@@ -32,6 +32,8 @@ Explicit scrolling uses `scrollIntoView` from `support.ts`: it resolves the loca
They type-check in the root `tsconfig.host.json`, not in the Client aggregate, because they read Host services directly: `ctx.connection`, the Host `SessionStore`, and `ctx.sessionProjectionCache`. Driving a browser at runtime does not make a file part of the Client program — the two faces merge Cordis `Context` under the same keys with different services, so one program cannot see both. Moving these files into the Client aggregate makes every Host-service access fail to compile.
A new scenario joins both program lists in the same change: its file in the root `tsconfig.host.json` include and in the Client aggregate's `apps/web/tsconfig.json` exclude. That project includes all of `tests`, so a file missing from the exclude compiles as Client code, and the Client compiler gate of `pnpm run typecheck` keeps failing with Host-service errors until it is listed.
## Do not import `@deepseek-ai/dsh-client-*` here
Importing a Client package — a value or a type — pulls its TypeScript project and referenced projects into the Host build graph. Generated Remote declarations are available before package compilation, but this does not make the Host and Client Cordis `Context` declarations compatible. Keep these tests on the Host face and use only the explicit built-client harness exception below.
apps/web/tests/README.zh.md
+2
View File
@@ -32,6 +32,8 @@ pnpm --filter @deepseek-ai/dsh-web-frontend exec playwright install --with-deps
它们在根 `tsconfig.host.json` 中做类型检查，而不在 Client aggregate 中，因为它们直接读取 Host 服务：`ctx.connection`、Host 侧 `SessionStore` 与 `ctx.sessionProjectionCache`。运行时驱动浏览器并不使一个文件成为 Client 程序的一部分——两个 face 在相同的键上以不同服务合并 Cordis `Context`，因此单个程序无法同时看见两者。把这些文件挪进 Client aggregate 会让每一处 Host 服务访问都无法编译。
新增场景要在同一次改动里登记到两份工程清单：根 `tsconfig.host.json` 的 include，以及 Client aggregate `apps/web/tsconfig.json` 的 exclude。该工程包含整个 `tests` 目录，漏登记 exclude 的文件会按 Client 代码编译，`pnpm run typecheck` 的 Client compiler 关卡会一直以 Host 服务错误失败，直到它被列入。
## 不要在此 import `@deepseek-ai/dsh-client-*`
import 一个 Client 包——无论值还是类型——都会把它的 TypeScript 工程及其引用的工程拉进 Host 构建图。生成的 Remote 声明在包编译之前就绪，但这并不使 Host 与 Client 的 Cordis `Context` 声明兼容。保留这些测试的 Host face 归属，只使用下文明确的 built-client harness 例外。
apps/web/tests/cua-optin.e2e.ts
+22 -3
View File
@@ -8,8 +8,8 @@ import { readProfileManifest } from '@deepseek-ai/dsh-app-boot'
import { SessionId } from '@deepseek-ai/dsh-session'
import type {} from '@deepseek-ai/dsh-ptc-runtime'
import type {} from '@deepseek-ai/dsh-agent-preset-registry'
import { expect, it } from 'vitest'
import { launchWebScaffold, type WebScaffold } from './scaffold.ts'
import { expect, it, vi } from 'vitest'
import { captureStableAria, compareOrRefreshGolden, launchWebScaffold, webSnapshotMode, type WebScaffold } from './scaffold.ts'
import { connectFreshWorkspaceZh, ZH_BROWSER_LOCALE } from './support.ts'
it.each([undefined, 'minimal'] as const)('adds CUA to PTC sessions while preserving default %s until the explicit persistent-runtime action', async (selectedDefault) => {
@@ -54,9 +54,28 @@ it.each([undefined, 'minimal'] as const)('adds CUA to PTC sessions while preserv
const toggle = card.getByRole('switch', { name: '启用 Computer Use', exact: true })
expect(await toggle.getAttribute('aria-checked')).toBe('false')
expect(await card.textContent()).toContain('请切换到 PTC 模式')
const enabling = Promise.withResolvers<undefined>()
const setEnabled = scaffold.ctx.pluginManager.setBundleEnabled.bind(scaffold.ctx.pluginManager)
let applying: ReturnType<typeof setEnabled> | undefined
// Hold the Host request so the pending UI does not depend on machine speed; release into the real operation.
const enableCall = vi.spyOn(scaffold.ctx.pluginManager, 'setBundleEnabled').mockImplementationOnce((name, enabled) => {
applying = enabling.promise.then(() => setEnabled(name, enabled))
return applying
})
try {
await toggle.click()
await card.getByRole('status').getByText('正在应用 Computer Use 的更改', { exact: true }).waitFor()
expect(await toggle.isEnabled()).toBe(false)
await compareOrRefreshGolden(fileURLToPath(new URL('./expected/cua-permissions/enabling.expected.md', import.meta.url)),
await captureStableAria(page, '[data-plugin-package="@deepseek-ai/dsh-cua-bundle"]', scaffold.workspaceCwd), webSnapshotMode())
} finally {
enabling.resolve(undefined)
enableCall.mockRestore()
await applying
}
// Bundle transitions await native cleanup under the browser action deadline.
await toggle.click()
await toggle.and(page.locator('[aria-checked="true"]')).waitFor({ state: 'visible' })
await card.getByRole('status').getByText('正在应用 Computer Use 的更改', { exact: true }).waitFor({ state: 'hidden' })
const setup = page.getByRole('dialog', { name: '使用 Computer Use 前需要设置', exact: true })
await setup.waitFor()
expect(await setup.getByText('Computer Use 插件已开启，但暂时无法检测系统权限。请前往插件详情页查看原因并重试。', { exact: true }).count()).toBe(1)
apps/web/tests/cua-permissions.e2e.ts
+94 -3
View File
@@ -7,9 +7,9 @@ import { ToolCallId } from '@deepseek-ai/dsh-llm'
import type {} from '@deepseek-ai/dsh-user-approval'
import type {} from '@deepseek-ai/dsh-agent-preset-registry'
import { chromium, type Browser } from 'playwright'
import { expect, it, onTestFailed, onTestFinished } from 'vitest'
import { expect, it, onTestFailed, onTestFinished, vi } from 'vitest'
import { captureStableAria, compareOrRefreshGolden, launchWebScaffold, webSnapshotMode, watchConsole, type WebScaffold } from './scaffold.ts'
import { connectFreshWorkspace, newEnglishPage, saveFailureShot } from './support.ts'
import { connectFreshWorkspace, newEnglishPage, PLUGIN_TOGGLE_SETTLE_MS, saveFailureShot } from './support.ts'
import { installNativeFixture, startFixtureHost } from '../../../packages/experimental/cua-macos/tests/native-fixture.ts'
const bundle = fileURLToPath(new URL('../../../packages/experimental/cua-bundle', import.meta.url))
@@ -146,5 +146,96 @@ it.skipIf(process.platform !== 'darwin')('keeps system requests pending until fr
{ method: 'openPermissionSettings', params: { permission: 'accessibility' } },
])
expect(events.every(event => ['permissionStatus', 'requestPermission', 'openPermissionSettings', 'applicationIcons', 'getApp'].includes(event.method))).toBe(true)
await plugins.locator('[data-detail-document] h1').waitFor()
const trace = await page.evaluateHandle(() => {
const errors: string[] = []
const stable = [...document.querySelectorAll('[data-cua-settings] [role="alert"]')].map(node => node.textContent?.trim())
const layouts: { height: number; sections: number[]; groups: number[] }[] = []
const documents: number[] = []
let frame = 0
const measure = () => {
const documentHeight = document.querySelector('[data-plugin-detail] [data-detail-document]')?.getBoundingClientRect().height ?? 0
if (documents.at(-1) !== documentHeight) documents.push(documentHeight)
const settings = document.querySelector('[data-cua-settings]')
if (settings) {
const height = settings.getBoundingClientRect().height
if (layouts.at(-1)?.height !== height) layouts.push({ height,
sections: [...settings.querySelectorAll('section')].map(node => node.getBoundingClientRect().height),
groups: [...settings.querySelectorAll('[role="group"]')].map(node => node.getBoundingClientRect().height),
})
}
frame = requestAnimationFrame(measure)
}
measure()
const observer = new MutationObserver(() => {
for (const node of document.querySelectorAll('[data-cua-settings] [role="alert"]')) {
const text = node.textContent?.trim()
if (text && !stable.includes(text) && errors.at(-1) !== text) errors.push(text)
}
})
observer.observe(document.body, { childList: true, subtree: true, characterData: true })
return { errors, layouts, documents, stop: () => { observer.disconnect(); cancelAnimationFrame(frame) } }
})
try {
const toggle = page.locator('[data-plugin-panel]').getByRole('switch', { name: 'Enable Computer Use', exact: true })
for (const width of [1680, 390]) {
await page.setViewportSize({ width, height: 1000 })
if (width === 390) await page.locator('[data-sidebar-collapsed]').waitFor({ state: 'attached' })
await page.locator('[data-animating]').waitFor({ state: 'hidden' })
await trace.evaluate((value) => { value.layouts.length = 0; value.documents.length = 0 })
for (let cycle = 0; cycle < 3; cycle++) {
await toggle.click()
await expect.poll(() => toggle.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('false')
await expect.poll(() => toggle.isEnabled(), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe(true)
await toggle.click()
await expect.poll(() => toggle.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('true')
await expect.poll(() => panel.getAttribute('aria-busy'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('false')
expect(await trace.evaluate(value => value.errors)).toEqual([])
}
const layouts = await trace.evaluate(value => value.layouts)
expect(layouts.length).toBeGreaterThan(0)
const documentHeights = await trace.evaluate(value => value.documents)
expect(documentHeights.length).toBeGreaterThan(0)
expect(Math.max(...documentHeights) - Math.min(...documentHeights)).toBeLessThanOrEqual(1)
const heights = layouts.map(layout => layout.height)
expect(Math.max(...heights) - Math.min(...heights), JSON.stringify(layouts)).toBeLessThanOrEqual(1)
}
} finally {
await trace.evaluate((value) => { value.stop() })
await trace.dispose()
}
await writeFile(state, JSON.stringify({ accessibility: 'unavailable', screenRecording: false }))
await page.evaluate(() => { window.dispatchEvent(new Event('focus')) })
const diagnostic = panel.getByRole('alert')
await diagnostic.waitFor()
const diagnosticText = await diagnostic.innerText()
const height = await panel.evaluate(element => element.getBoundingClientRect().height)
const rowToggle = plugins.getByRole('switch', { name: 'Enable component @deepseek-ai/dsh-experimental-cua-macos', exact: true })
const release = Promise.withResolvers<undefined>()
const changeRow = scaffold.ctx.pluginManager.setPluginEnabled.bind(scaffold.ctx.pluginManager)
let changing: ReturnType<typeof changeRow> | undefined
const hold = vi.spyOn(scaffold.ctx.pluginManager, 'setPluginEnabled').mockImplementationOnce((entryId, enabled) => {
changing = release.promise.then(() => changeRow(entryId, enabled))
return changing
})
try {
await rowToggle.click()
await expect.poll(() => hold.mock.calls.length, { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe(1)
await expect.poll(() => panel.getAttribute('aria-busy')).toBe('true')
expect(await diagnostic.innerText()).toBe(diagnosticText)
expect(await diagnostic.getByRole('button', { name: 'Retry', exact: true }).isEnabled()).toBe(false)
expect(await panel.evaluate(element => element.getBoundingClientRect().height)).toBe(height)
} finally {
release.resolve(undefined)
hold.mockRestore()
await changing
}
await expect.poll(() => rowToggle.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('false')
await expect.poll(() => rowToggle.isEnabled(), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe(true)
await writeFile(state, JSON.stringify({ accessibility: false, screenRecording: false, launchApplication }))
await rowToggle.click()
await expect.poll(() => rowToggle.getAttribute('aria-checked'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('true')
await expect.poll(() => panel.getAttribute('aria-busy'), { timeout: PLUGIN_TOGGLE_SETTLE_MS }).toBe('false')
await diagnostic.waitFor({ state: 'hidden' })
expect(tripwire.pageErrors).toEqual([])
}, 60_000)
})
apps/web/tests/expected/cot-translation-paid/paid-choices.expected.md
+1
View File
@@ -1,5 +1,6 @@
- button "返回 实验性插件": 实验性插件
- button "卸载 思考过程机器翻译" [disabled]: 卸载
- status
- switch "启用 思考过程机器翻译" [checked]
- heading "思考过程机器翻译" [level=3]
- text: v{{version}} 实验性
apps/web/tests/expected/cot-translation/preferences.expected.md
+1
View File
@@ -1,5 +1,6 @@
- button "返回 实验性插件": 实验性插件
- button "卸载 思考过程机器翻译" [disabled]: 卸载
- status
- switch "启用 思考过程机器翻译" [checked]
- heading "思考过程机器翻译" [level=3]
- text: v{{version}} 实验性
apps/web/tests/expected/cua-permissions/enabling.expected.md
+5
View File
@@ -0,0 +1,5 @@
- listitem:
- button "查看 Computer Use": Computer Use
- text: 实验性 目前仅支持 macOS，Windows 支持正在开发中。为已有及新建 PTC 会话加载 Computer Use。请切换到 PTC 模式来使用，并授予辅助功能和屏幕录制权限。
- status: 正在应用 Computer Use 的更改
- switch "启用 Computer Use" [disabled]
apps/web/tests/expected/cua-permissions/missing.expected.md
+1 -1
View File
@@ -16,7 +16,7 @@
- strong: Screen Recording
- paragraph: In System Settings, allow Visual Studio Code.app to capture application windows.
- button "Request permission"
- paragraph: Enable the application in System Settings. Follow the macOS prompt if it names another application.
- paragraph: Manage permissions in System Settings for the application named by macOS.
- region "Manage applications Computer Use can access":
- heading "Manage applications Computer Use can access" [level=3]
- heading "Always-allowed applications" [level=4]
apps/web/tests/expected/plugin-config/bundle.expected.md
+1
View File
@@ -1,6 +1,7 @@
- button "返回插件": 插件
- button "夹具操作"
- button "卸载 @fixture/live-client": 卸载
- status
- switch "启用 @fixture/live-client" [checked]
- heading "@fixture/live-client" [level=3]
- text: v0.0.1 夹具标签
apps/web/tests/expected/plugin-config/official.expected.md
+6
View File
@@ -25,24 +25,30 @@
- listitem:
- button "查看 智能体团队": 智能体团队
- text: 实验性 启用团队协作、团队工具、成员列表和共享任务看板。
- status
- switch "启用 智能体团队"
- listitem:
- button "查看 语音输入": 语音输入
- text: 实验性 在本机使用 SenseVoice 转写录音，首次使用需安装依赖。
- status
- switch "启用 语音输入"
- listitem:
- button "查看 自动授权审查": 自动授权审查
- text: 实验性 提供自动审查权限模式，由模型在每次工具调用前判断是否授权。
- status
- switch "启用 自动授权审查"
- listitem:
- button "查看 开发者工具": 开发者工具
- text: 实验性 查看调试会话原始数据、聊天消息分组数据，以及调试 NodeJS 后端
- status
- switch "启用 开发者工具"
- heading "已安装插件" [level=3]
- list:
- listitem:
- button "查看 @fixture/live-client": "@fixture/live-client"
- status
- switch "启用 @fixture/live-client"
- listitem:
- button "查看 dsh-web-scaffold-defaults": dsh-web-scaffold-defaults
- status
- switch "启用 dsh-web-scaffold-defaults" [checked]
apps/web/tests/expected/plugin-install-cancel/highlight.expected.md
+1
View File
@@ -1,5 +1,6 @@
- listitem:
- button "查看 slow-package": slow-package
- status
- switch "启用 slow-package" [checked]
[
apps/web/tests/expected/plugin-manager/exports-en.expected.md
+1
View File
@@ -1,5 +1,6 @@
- button "Back to plugins": Plugins
- button "Uninstall @fixture/bundle": Uninstall
- status
- switch "Enable @fixture/bundle"
- heading "@fixture/bundle" [level=3]
- text: v0.0.1
apps/web/tests/expected/plugin-manager/exports.expected.md
+1
View File
@@ -1,5 +1,6 @@
- button "返回插件": 插件
- button "卸载 @fixture/bundle": 卸载
- status
- switch "启用 @fixture/bundle"
- heading "@fixture/bundle" [level=3]
- text: v0.0.1
apps/web/tests/expected/plugin-manager/live-enabled.expected.md
+6
View File
@@ -25,25 +25,31 @@
- listitem:
- button "查看 智能体团队": 智能体团队
- text: 实验性 启用团队协作、团队工具、成员列表和共享任务看板。
- status
- switch "启用 智能体团队"
- listitem:
- button "查看 语音输入": 语音输入
- text: 实验性 在本机使用 SenseVoice 转写录音，首次使用需安装依赖。
- status
- switch "启用 语音输入"
- listitem:
- button "查看 自动授权审查": 自动授权审查
- text: 实验性 提供自动审查权限模式，由模型在每次工具调用前判断是否授权。
- status
- switch "启用 自动授权审查"
- listitem:
- button "查看 开发者工具": 开发者工具
- text: 实验性 查看调试会话原始数据、聊天消息分组数据，以及调试 NodeJS 后端
- status
- switch "启用 开发者工具"
- heading "已安装插件" [level=3]
- list:
- listitem:
- button "查看 @fixture/bundle": "@fixture/bundle"
- text: Registry description for the fixture bundle.
- status
- switch "启用 @fixture/bundle" [checked]
- listitem:
- button "查看 dsh-web-scaffold-defaults": dsh-web-scaffold-defaults
- status
- switch "启用 dsh-web-scaffold-defaults" [checked]
apps/web/tests/expected/plugin-manager/manager.expected.md
+6
View File
@@ -25,25 +25,31 @@
- listitem:
- button "查看 智能体团队": 智能体团队
- text: 实验性 启用团队协作、团队工具、成员列表和共享任务看板。
- status
- switch "启用 智能体团队"
- listitem:
- button "查看 语音输入": 语音输入
- text: 实验性 在本机使用 SenseVoice 转写录音，首次使用需安装依赖。
- status
- switch "启用 语音输入"
- listitem:
- button "查看 自动授权审查": 自动授权审查
- text: 实验性 提供自动审查权限模式，由模型在每次工具调用前判断是否授权。
- status
- switch "启用 自动授权审查"
- listitem:
- button "查看 开发者工具": 开发者工具
- text: 实验性 查看调试会话原始数据、聊天消息分组数据，以及调试 NodeJS 后端
- status
- switch "启用 开发者工具"
- heading "已安装插件" [level=3]
- list:
- listitem:
- button "查看 @fixture/bundle": "@fixture/bundle"
- text: Registry description for the fixture bundle.
- status
- switch "启用 @fixture/bundle"
- listitem:
- button "查看 dsh-web-scaffold-defaults": dsh-web-scaffold-defaults
- status
- switch "启用 dsh-web-scaffold-defaults" [checked]
apps/web/tests/expected/plugin-manager/missing-bundle.expected.md
+1
View File
@@ -1,5 +1,6 @@
- button "返回插件": 插件
- button "卸载 @fixture/missing-bundle": 卸载
- status
- switch "启用 @fixture/missing-bundle" [checked]
- heading "@fixture/missing-bundle" [level=3]
- text: 异常 未安装
apps/web/tests/expected/plugin-manager/preview.expected.md
+1
View File
@@ -1,4 +1,5 @@
- button "返回 实验性插件": 实验性插件
- status
- switch "启用 DSH 徽章技能"
- heading "DSH 徽章技能" [level=3]
- text: v{{version}} 实验性
apps/web/tests/expected/plugin-manager/touch-tooltips.expected.md
+8
View File
@@ -0,0 +1,8 @@
# Touch tooltips
Touch refresh: one refresh request; no tooltip
Touch explanation: visible within the 390px viewport
- tooltip "在这里配置官方插件，安装和管理其他插件。内置插件列表及运行状态可在「设置 → 内置插件」中查看"
Explanation closes on another tap or an outside touch
Keyboard focus after touch: refresh tooltip visible
Mouse hover after touch: refresh tooltip visible
apps/web/tests/expected/tooltip-touch/copy.expected.md
+5
View File
@@ -0,0 +1,5 @@
# Touch copy feedback
Copy succeeds and the button returns from Copied to Copy.
Tooltips during copy feedback: 0
Tooltips after emulated mouse entry: 0
apps/web/tests/hover-input.e2e.ts
+242
View File
@@ -0,0 +1,242 @@
/** Pointer, keyboard, and resize behavior in the assembled client. */
import { fileURLToPath } from 'node:url'
import { chromium, type Locator } from 'playwright'
import { expect, it, onTestFailed, onTestFinished } from 'vitest'
import { launchWebScaffold, seedSession, watchConsole } from './scaffold.ts'
import { createChatScrollFixture } from './chat-scroll-fixture.ts'
import { connectFreshWorkspace, saveFailureShot } from './support.ts'
it.each([false, true])('shows the new-session shortcut on keyboard focus with touch=%s', async (hasTouch) => {
const scaffold = await launchWebScaffold()
onTestFinished(() => scaffold.close())
const browser = await chromium.launch()
onTestFinished(() => browser.close())
const context = await browser.newContext({ viewport: { width: 1280, height: 900 }, hasTouch, locale: 'en-US' })
await context.addInitScript(() => { Object.defineProperty(navigator, 'platform', { value: 'MacIntel' }) })
const page = await context.newPage()
onTestFailed(() => saveFailureShot(page, `new-session-focus-${String(hasTouch)}`))
await page.goto(scaffold.authenticatedUrl)
const button = page.getByRole('button', { name: 'New session', exact: true }).filter({ has: page.locator('kbd') })
await button.waitFor()
const shortcut = button.locator('[class*="newSessionShortcut"]')
const mask = button.locator('[class*="newSessionLabelMask"]')
const opacity = () => shortcut.evaluate(element => getComputedStyle(element).opacity)
const maskImage = () => mask.evaluate(element => getComputedStyle(element).maskImage)
expect(await page.evaluate(() => matchMedia('(hover: none)').matches)).toBe(hasTouch)
expect(await opacity()).toBe('0')
expect(await maskImage()).toBe('none')
await button.focus()
await page.keyboard.press('Shift+Tab')
await page.keyboard.press('Tab')
expect(await button.evaluate(element => element === document.activeElement && element.matches(':focus-visible'))).toBe(true)
expect(await opacity()).toBe('1')
expect(await maskImage()).not.toBe('none')
await page.keyboard.press('Tab')
expect(await opacity()).toBe('0')
expect(await maskImage()).toBe('none')
if (hasTouch) {
await button.tap()
expect(await button.evaluate(element => element.matches(':focus-visible'))).toBe(false)
expect(await opacity()).toBe('0')
expect(await maskImage()).toBe('none')
} else {
await button.hover()
expect(await opacity()).toBe('1')
expect(await maskImage()).not.toBe('none')
}
})
it.each([false, true])('keeps workspace and archived search actions reachable with touch=%s', async (hasTouch) => {
const scaffold = await launchWebScaffold()
onTestFinished(() => scaffold.close())
const fixture = createChatScrollFixture({ markerPrefix: 'ROW_ACTIONS', title: 'Archived row actions', turns: 1 })
const id = await seedSession(scaffold, fixture.log, 'row-actions')
const browser = await chromium.launch()
onTestFinished(() => browser.close())
const page = await browser.newPage({ viewport: { width: 1280, height: 900 }, hasTouch, locale: 'en-US' })
await page.addInitScript(() => { Object.defineProperty(navigator, 'platform', { value: 'MacIntel' }) })
onTestFailed(() => saveFailureShot(page, `row-actions-${String(hasTouch)}`))
const activate = async (target: Locator) => {
if (hasTouch) await target.tap()
else await target.click()
}
const openFixture = async () => {
await activate(page.getByRole('button', { name: 'Search sessions', exact: true }))
await page.getByRole('textbox', { name: 'Search session names', exact: true }).fill(fixture.markers.user(1))
await activate(page.getByRole('tree', { name: 'Search results' }).getByRole('treeitem'))
await expect.poll(() => row.getAttribute('aria-selected')).toBe('true')
}
const row = page.locator(`[data-row-key="session:${id}"]`)
await page.goto(scaffold.authenticatedUrl)
await connectFreshWorkspace(page, scaffold.workspaceCwd)
await openFixture()
await activate(page.getByRole('button', { name: 'View options', exact: true }))
await activate(page.getByRole('menuitem', { name: 'WorkSpace', exact: true }))
const menu = page.getByRole('button', { name: 'Workspace actions for workspace', exact: true, includeHidden: true })
const project = page.locator('[data-row-key^="workspace:"]').filter({ has: menu })
await page.mouse.move(0, 0)
expect(await page.evaluate(() => matchMedia('(hover: none)').matches)).toBe(hasTouch)
expect(await menu.isVisible()).toBe(hasTouch)
if (!hasTouch) await project.hover()
await activate(menu)
await page.getByRole('menuitem', { name: 'Rename', exact: true }).waitFor()
await page.keyboard.press('Escape')
if (!hasTouch) await project.hover()
await activate(page.getByRole('button', { name: 'New session in workspace', exact: true }))
await expect.poll(() => row.getAttribute('aria-selected')).not.toBe('true')
expect(await project.getAttribute('aria-expanded')).toBe('true')
await openFixture()
if (!hasTouch) await row.hover()
const actions = row.getByRole('button', { name: /^Session actions for / })
await activate(actions)
const archive = page.getByRole('menuitem', { name: 'Archive session', exact: true })
const shortcut = archive.locator('[aria-hidden="true"]').filter({ has: page.locator('kbd') })
await archive.waitFor()
expect(await shortcut.count()).toBe(1)
expect(await page.evaluate(() => matchMedia('(pointer: coarse)').matches)).toBe(hasTouch)
expect(await shortcut.isVisible()).toBe(!hasTouch)
for (const key of ['ArrowDown', 'Tab']) {
await page.keyboard.press(key)
expect(await page.locator('html').getAttribute('data-input-modality')).toBe('keyboard')
expect(await page.getByRole('menu').evaluate(menu => menu.contains(document.activeElement))).toBe(true)
expect(await shortcut.isVisible()).toBe(true)
if (hasTouch) {
await page.setViewportSize({ width: 390, height: 900 })
await expect.poll(() => shortcut.evaluate((element) => {
const bounds = element.getBoundingClientRect()
return bounds.left >= 0 && bounds.right <= window.innerWidth
})).toBe(true)
await page.setViewportSize({ width: 1280, height: 900 })
}
await page.keyboard.press('Escape')
if (!hasTouch) await row.hover()
await activate(actions)
await archive.waitFor()
expect(await shortcut.isVisible()).toBe(!hasTouch)
}
await activate(archive)
await row.waitFor({ state: 'detached' })
await activate(page.getByRole('button', { name: 'View options', exact: true }))
await activate(page.getByRole('menuitem', { name: 'All conversations (show archived)', exact: true }))
await activate(page.getByRole('button', { name: 'Search sessions', exact: true }))
await page.getByRole('textbox', { name: 'Search session names', exact: true }).fill(fixture.markers.user(1))
const result = page.getByRole('tree', { name: 'Search results' }).getByRole('treeitem')
await result.waitFor()
const restore = result.getByRole('button', { name: 'Unarchive session', exact: true, includeHidden: true })
expect(await restore.isVisible()).toBe(hasTouch)
if (!hasTouch) await result.hover()
await activate(restore)
await expect.poll(() => scaffold.ctx.workspaceRegistry.archivedSessionIds.includes(id)).toBe(false)
await restore.waitFor({ state: 'detached' })
await activate(result)
await expect.poll(() => row.getAttribute('aria-selected')).toBe('true')
})
it.each([false, true])('keeps actions and focus usable with touch=%s across layout widths', async (hasTouch) => {
const scaffold = await launchWebScaffold()
onTestFinished(() => scaffold.close())
const fixture = createChatScrollFixture({ markerPrefix: 'HOVER', title: 'Hover input', turns: 1 })
const id = await seedSession(scaffold, fixture.log, 'hover-input')
const browser = await chromium.launch()
onTestFinished(() => browser.close())
const context = await browser.newContext({
viewport: { width: 1280, height: 900 }, hasTouch, locale: 'en-US', timezoneId: 'Asia/Shanghai',
})
const page = await context.newPage()
const consoleWatch = watchConsole(page)
onTestFailed(() => saveFailureShot(page, `hover-input-${String(hasTouch)}`))
await page.goto(scaffold.authenticatedUrl)
await connectFreshWorkspace(page, scaffold.workspaceCwd)
expect(await page.evaluate(() => matchMedia('(hover: hover)').matches)).toBe(!hasTouch)
await page.getByRole('button', { name: 'Search sessions' }).click()
await page.getByRole('textbox', { name: 'Search session names', exact: true }).fill(fixture.markers.user(1))
const result = page.getByRole('tree', { name: 'Search results' }).getByRole('treeitem')
await expect.poll(() => result.count()).toBe(1)
await result.click()
await page.getByText(fixture.markers.assistant(1), { exact: false }).last().waitFor()
const row = page.locator(`[data-row-key="session:${id}"]`)
await row.waitFor({ state: 'visible' })
expect(await row.getAttribute('aria-selected')).toBe('true')
const copy = page.locator('[data-clock="end"] > button').first()
const background = () => copy.evaluate(element => getComputedStyle(element).backgroundColor)
await page.mouse.move(0, 0)
const resting = await background()
if (hasTouch) {
await copy.tap()
expect(await background()).toBe(resting)
expect(await row.getAttribute('aria-selected')).toBe('true')
expect(await page.getByRole('tooltip').count()).toBe(0)
} else {
await copy.hover()
await expect.poll(background).not.toBe(resting)
await page.mouse.move(0, 0)
await expect.poll(background).toBe(resting)
}
await page.keyboard.press('Tab')
await copy.focus()
expect(await copy.evaluate(element => element.matches(':focus-visible'))).toBe(true)
await row.focus()
expect(await row.evaluate(element => document.activeElement === element)).toBe(true)
await page.keyboard.press('Enter')
expect(await row.getAttribute('aria-selected')).toBe('true')
const image = fileURLToPath(new URL('../../../snapshots/session/read-image/workspace/red.png', import.meta.url))
await page.locator('input[type="file"]').setInputFiles(image)
const remove = page.getByRole('button', { name: 'Remove image red.png', exact: true })
await remove.waitFor({ state: 'visible' })
if (hasTouch) {
expect(await remove.evaluate(element => getComputedStyle(element).opacity)).toBe('1')
await remove.tap()
} else {
await remove.focus()
await expect.poll(() => remove.evaluate(element => getComputedStyle(element).opacity)).toBe('1')
await page.keyboard.press('Enter')
}
await remove.waitFor({ state: 'detached' })
const input = page.locator('[data-composer-input]')
await input.fill('keep this draft')
await input.press('Home')
await input.press('Shift+End')
const element = await input.elementHandle()
const selection = () => input.evaluate(node => ({
focused: document.activeElement === node,
text: node.textContent,
selected: window.getSelection()?.toString(),
}))
const before = await selection()
expect(before.focused).toBe(true)
for (const width of [390, 768, 1280]) {
await page.setViewportSize({ width, height: 900 })
expect(await element!.evaluate(node => node.isConnected)).toBe(true)
expect(await selection()).toEqual(before)
}
await page.getByRole('button', { name: 'Open right sidebar', exact: true }).click()
const panel = page.locator('[data-sidebar-right-panel][data-sidebar-right-open]')
await panel.getByRole('button', { name: 'Fullscreen', exact: true }).click()
await panel.locator('[data-dockkit-split-button]').click()
await expect.poll(() => panel.locator('[data-dockkit-pane]').count()).toBe(2)
await page.mouse.move(0, 0)
await page.keyboard.press('Escape')
await page.getByRole('tooltip').waitFor({ state: 'detached' })
const blockedSplit = panel.locator('[data-dockkit-split-button]').last()
expect(await blockedSplit.isDisabled()).toBe(true)
const hit = await blockedSplit.evaluate((button) => {
const bounds = button.getBoundingClientRect()
const x = bounds.x + bounds.width / 2
const y = bounds.y + bounds.height / 2
const target = document.elementFromPoint(x, y)
return { x, y, wrapperReceivesInput: target === button.parentElement, cursor: target && getComputedStyle(target).cursor }
})
expect(hit.wrapperReceivesInput).toBe(true)
expect(['auto', 'default']).toContain(hit.cursor)
// Coordinate input exercises CSS hit testing over the disabled button itself.
if (hasTouch) await page.touchscreen.tap(hit.x, hit.y)
else await page.mouse.click(hit.x, hit.y)
await page.getByRole('tooltip').filter({ hasText: 'Two panes is the limit' }).waitFor()
expect(await panel.locator('[data-dockkit-pane]').count()).toBe(2)
expect(consoleWatch.pageErrors).toEqual([])
})
apps/web/tests/markdown-wide-table.e2e.ts
+40 -2
View File
@@ -20,8 +20,8 @@
// NO_ADAPTER.
import { fileURLToPath } from 'node:url'
import type { Browser, Page } from 'playwright'
import { chromium } from 'playwright'
import { afterAll, beforeAll, describe, expect, it, onTestFailed } from 'vitest'
import { chromium, webkit } from 'playwright'
import { afterAll, beforeAll, describe, expect, it, onTestFailed, onTestFinished } from 'vitest'
import { createMessage, createUserMessage } from '@deepseek-ai/dsh-llm'
import {
SESSION_FORMAT_VERSION,
@@ -374,6 +374,44 @@ describe('web e2e: markdown tables fill the column, wide ones break out and scro
expect(tripwire.pageErrors).toEqual([])
}, 120_000)
it.each(['chromium', 'webkit'] as const)('keeps wide tables horizontally scrollable without hover in %s', async (engine) => {
const touchBrowser = await ({ chromium, webkit })[engine].launch()
onTestFinished(() => touchBrowser.close())
// Mobile WebKit has no wheel driver; touch emulation also disables hover in its desktop viewport.
const context = await touchBrowser.newContext({
viewport: { width: 390, height: 844 }, hasTouch: true, isMobile: engine === 'chromium', locale: 'en-US',
})
const touchPage = await context.newPage()
const touchConsole = watchConsole(touchPage)
onTestFailed(() => saveFailureShot(touchPage, `markdown-wide-table-touch-${engine}`))
await touchPage.goto(scaffold.authenticatedUrl, { waitUntil: 'load' })
const group = touchPage.getByRole('treeitem').first()
await touchPage.getByRole('button', { name: 'Open sidebar', exact: true }).first().click()
await group.click()
await touchPage.getByRole('treeitem').filter({ hasText: 'Markdown wide tables' }).click()
await touchPage.getByText(TAIL_MARKER, { exact: true }).waitFor()
expect(await touchPage.evaluate(() => matchMedia('(hover: hover)').matches)).toBe(false)
const wide = touchPage.locator('[class*="tableScroll"]', { hasText: WIDE_MARKER })
await wide.scrollIntoViewIfNeeded()
expect(await wide.evaluate(element => ({
overflow: getComputedStyle(element).overflowX,
overflowing: element.scrollWidth > element.clientWidth,
focused: element.matches(':focus-visible'),
}))).toEqual({ overflow: 'auto', overflowing: true, focused: false })
const box = await wide.boundingBox()
if (box === null) throw new Error('Wide table is not visible')
// A wheel input must scroll the native overflow container without hover or focus.
// Assigning scrollLeft would also work with overflow:hidden and miss this regression.
await touchPage.mouse.move(box.x + box.width / 2, box.y + box.height / 2)
await touchPage.mouse.wheel(5_000, 0)
await expect.poll(() => wide.evaluate(element => element.scrollLeft)).toBeGreaterThan(0)
await expect.poll(() => wide.evaluate((element) => {
const last = element.querySelector('thead th:last-child')!
return last.getBoundingClientRect().right <= element.getBoundingClientRect().right + 1
})).toBe(true)
expect(touchConsole.pageErrors).toEqual([])
})
it('reveals the wide table scrollbar on hover only', async () => {
onTestFailed(() => saveFailureShot(page, 'web-e2e-markdown-wide-table-scrollbar'))
await sweep()
apps/web/tests/plugin-manager.e2e.ts
+84 -2
View File
@@ -298,7 +298,7 @@ describe('web e2e: plugin manager', () => {
width: node.getBoundingClientRect().width,
height: node.getBoundingClientRect().height,
})))
expect(blankActions).toEqual(Array.from({ length: 4 }, () => ({ children: 0, background: 'rgba(0, 0, 0, 0)', width: 36, height: 20 })))
expect(blankActions).toEqual(Array.from({ length: 4 }, () => ({ children: 0, background: 'rgba(0, 0, 0, 0)', width: 58, height: 20 })))
if (shots !== undefined) {
for (const colorScheme of ['light', 'dark'] as const) {
await probe.emulateMedia({ colorScheme })
@@ -488,6 +488,88 @@ describe('web e2e: plugin manager', () => {
}
})
it('keeps touch actions direct before keyboard focus and mouse hover', async () => {
const context = await browser.newContext({
viewport: { width: 1280, height: 900 }, locale: ZH_BROWSER_LOCALE,
})
try {
const probe = await context.newPage()
const input = await context.newCDPSession(probe)
// Native touch input preserves desktop hover capability for mixed-input devices.
const tap = async (target: Locator): Promise<void> => {
await target.scrollIntoViewIfNeeded()
const bounds = await target.boundingBox()
if (bounds === null) throw new Error('Touch target is not visible')
await input.send('Input.dispatchTouchEvent', {
type: 'touchStart', touchPoints: [{ x: bounds.x + bounds.width / 2, y: bounds.y + bounds.height / 2 }],
})
await input.send('Input.dispatchTouchEvent', { type: 'touchEnd', touchPoints: [] })
}
const consoleWatch = watchConsole(probe)
onTestFailed(() => saveFailureShot(probe, 'web-e2e-touch-tooltips'))
await probe.clock.install()
await probe.goto(scaffold.authenticatedUrl, { waitUntil: 'load' })
await probe.getByRole('button', { name: '插件', exact: true }).click()
const panel = probe.locator('[data-plugin-panel]')
const refresh = panel.getByRole('button', { name: '刷新', exact: true })
const info = panel.getByRole('button', { name: '插件说明', exact: true })
await expect.poll(() => refresh.isEnabled()).toBe(true)
await probe.setViewportSize({ width: 390, height: 844 })
await probe.evaluate(() => document.fonts.ready)
const calls = vi.spyOn(scaffold.ctx.pluginManager, 'listBundles')
try {
await tap(refresh)
await expect.poll(() => calls.mock.calls.length).toBe(1)
await expect.poll(() => refresh.isEnabled()).toBe(true)
} finally {
calls.mockRestore()
}
await probe.clock.pauseAt(await probe.evaluate(() => Date.now() + 1000))
await probe.clock.runFor(500)
expect(await probe.getByRole('tooltip').count()).toBe(0)
const facts = ['Touch refresh: one refresh request; no tooltip']
await tap(info)
const help = probe.getByRole('tooltip')
await help.waitFor()
const bounds = await help.boundingBox()
if (bounds === null) throw new Error('Tapped plugin explanation is not visible')
expect(bounds.x).toBeGreaterThanOrEqual(12)
expect(bounds.x + bounds.width).toBeLessThanOrEqual(390 - 12)
expect(bounds.y).toBeGreaterThanOrEqual(12)
expect(bounds.y + bounds.height).toBeLessThanOrEqual(844 - 12)
facts.push('Touch explanation: visible within the 390px viewport', await help.ariaSnapshot())
await tap(info)
expect(await help.count()).toBe(0)
await tap(info)
await help.waitFor()
await tap(panel.getByRole('heading', { name: '插件', exact: true }))
expect(await help.count()).toBe(0)
facts.push('Explanation closes on another tap or an outside touch')
await probe.keyboard.press('Tab')
await refresh.focus()
await probe.clock.runFor(500)
await probe.getByRole('tooltip').filter({ hasText: '刷新' }).waitFor()
facts.push('Keyboard focus after touch: refresh tooltip visible')
await probe.keyboard.press('Tab')
expect(await probe.evaluate(() => matchMedia('(any-hover: hover)').matches)).toBe(true)
await refresh.hover()
await probe.clock.runFor(500)
await probe.getByRole('tooltip').filter({ hasText: '刷新' }).waitFor()
facts.push('Mouse hover after touch: refresh tooltip visible')
await probe.mouse.move(0, 843)
expect(await help.count()).toBe(0)
await input.detach()
await probe.clock.resume()
await compareOrRefreshGolden(join(SNAPSHOT_DIR, 'touch-tooltips.expected.md'),
['# Touch tooltips', '', ...facts].join('\n'), MODE)
expect(consoleWatch.pageErrors).toEqual([])
} finally {
await context.close()
}
})
it('starts with an unavailable selected bundle and lets the user clear its error', async () => {
onTestFailed(() => saveFailureShot(page, 'web-e2e-plugin-manager-missing-bundle'))
const panel = await openPluginsPanel()
@@ -1139,7 +1221,7 @@ describe('web e2e: plugin manager', () => {
expect(tripwire.warnings).toEqual([])
await assertFixtureInventory(SNAPSHOT_DIR, [
'manager.expected.md', 'live-enabled.expected.md', 'missing-bundle.expected.md', 'exports.expected.md', 'exports-en.expected.md', 'icons.expected.md', 'loading.expected.md', 'refresh.expected.md',
'ime-enter.expected.md', 'header-responsive.expected.md', 'preview.expected.md',
'ime-enter.expected.md', 'header-responsive.expected.md', 'preview.expected.md', 'touch-tooltips.expected.md',
])
})
})
apps/web/tests/present.e2e.ts
+25
View File
@@ -175,6 +175,31 @@ fs.appendFileSync(${JSON.stringify(openLog)}, JSON.stringify({ path, action, con
expect(await opened()).toHaveLength(count + 1)
expect((await opened()).at(-1)).toEqual({ action: 'open', path: await realpath(join(cwd, name)), content: bytes })
}
const success = row.locator('[data-presented-file]').filter({ hasText: '说明.txt' }).locator('[data-success]')
await success.waitFor()
const animation = await success.evaluate((element) => {
const style = getComputedStyle(element)
const fade = element.getAnimations()[0]
if (fade === undefined) throw new Error('Successful file feedback has no fade animation')
// Seek the CSS animation so the 200 ms fade does not race the assertion.
fade.pause()
fade.currentTime = 5100
return {
delay: style.animationDelay, duration: style.animationDuration,
fill: style.animationFillMode, opacity: Number(getComputedStyle(element).opacity),
}
})
expect(animation.delay).toBe('5s')
expect(animation.duration).toBe('0.2s')
expect(animation.fill).toBe('forwards')
expect(animation.opacity).toBeGreaterThan(0)
expect(animation.opacity).toBeLessThan(1)
try {
await page.emulateMedia({ reducedMotion: 'reduce' })
expect(await success.evaluate(element => getComputedStyle(element).animationName)).toBe('none')
} finally {
await page.emulateMedia({ reducedMotion: 'no-preference' })
}
}
expect(downloads).toEqual([])
const response = await page.request.get(new URL(`/api/session.export?sessionId=${sessionId}`, scaffold.authenticatedUrl).href)
apps/web/tests/shipped-composition.e2e.ts
+20 -20
View File
@@ -47,12 +47,13 @@ const AUTO_PARENT_ADJUST = 'AUTO_PARENT_ADJUST'
/** Identify the one-shot Auto Review request and check its request-only outer input. */
function isAutoReviewRequest(options: GenerateOptions): boolean {
if (options.system?.startsWith('REVIEW_POLICY\n') !== true) return false
expect(options.messages).toHaveLength(1)
const message = options.messages[0]
expect(message).toMatchObject({ role: 'user', content: [{ type: 'text' }] })
expect(message?.content).toHaveLength(1)
expect(message).not.toHaveProperty('id')
expect(message).not.toHaveProperty('source')
expect(options.messages.length).toBeGreaterThanOrEqual(2)
for (const message of options.messages) {
expect(message).toMatchObject({ role: 'user', content: [{ type: 'text' }] })
expect(message.content).toHaveLength(1)
expect(message).not.toHaveProperty('id')
expect(message).not.toHaveProperty('source')
}
return true
}
@@ -150,21 +151,19 @@ interface ChildScriptState {
phase: number
}
/** Parse the fixed review request sections emitted by the production plugin. */
/** Parse the history items and pending action emitted by the production plugin. */
function childReviewSections(options: GenerateOptions): Record<string, unknown> {
const block = options.messages[0]?.content[0]
if (block?.type !== 'text') throw new Error('shipped child review request has no text body')
const labels = ['ENVIRONMENT', 'PROJECT_INSTRUCTIONS', 'FILTERED_HISTORY', 'PENDING_ACTION'] as const
const sections: Record<string, unknown> = {}
for (const [index, label] of labels.entries()) {
const prefix = `${label}\n`
const start = block.text.indexOf(prefix)
if (start < 0) throw new Error(`shipped child review request is missing ${label}`)
const next = labels[index + 1]
const end = next === undefined ? block.text.length : block.text.indexOf(`\n\n${next}\n`, start)
sections[label] = JSON.parse(block.text.slice(start + prefix.length, end)) as unknown
const items = options.messages.slice(1).map((message) => {
const block = message.content[0]
if (block?.type !== 'text') throw new Error('shipped child review message has no text body')
return JSON.parse(block.text) as Record<string, unknown>
})
const call = items.at(-1)
if (call?.kind !== 'tool-call') throw new Error('shipped child review request does not end with its pending call')
return {
FILTERED_HISTORY: items.slice(0, -1).filter(item => item.kind !== 'tool-schema'),
PENDING_ACTION: { name: call.name, arguments: JSON.parse(call.arguments as string) as Record<string, unknown> },
}
return sections
}
/** One native tool-call completion in the provider-neutral stream vocabulary. */
@@ -790,7 +789,8 @@ it('routes one browser-authored Auto request through the same model and asks the
.filter(block => block.type === 'text')
.map(block => block.text)
.join('') ?? ''
expect(reviewInput).toContain('PENDING_ACTION')
expect(reviewInput).toContain('ENVIRONMENT')
expect(reviewInput).toContain('"kind": "tool-schema"')
expect(reviewInput).toContain(requestId)
expect(reviewInput).toContain(targetPath)
expect(approvalReasons).toEqual([`Auto review denied tool "bash": ${AUTO_RAW_REASON}`])
apps/web/tests/sidebar-scrollbar.e2e.ts
+8 -3
View File
@@ -71,17 +71,22 @@ function measureList(page: Page): Promise<ListMetrics> {
probe.remove()
return value
}
// Computed pseudo style folds in hover even at rest, so inspect the cascade.
// Computed pseudo style folds in hover even at rest, so inspect active CSS groups.
// Cross-origin sheets may throw and cannot contain the app-owned rule.
const styleRules = (rules: CSSRuleList): CSSStyleRule[] => [...rules].flatMap((rule) => {
if (rule instanceof CSSStyleRule) return [rule]
if (rule instanceof CSSMediaRule && !matchMedia(rule.conditionText).matches) return []
if (rule instanceof CSSSupportsRule && !CSS.supports(rule.conditionText)) return []
return rule instanceof CSSGroupingRule ? styleRules(rule.cssRules) : []
})
const hoverRules = [...document.styleSheets]
.flatMap((sheet) => {
try {
return [...sheet.cssRules]
return styleRules(sheet.cssRules)
} catch {
return []
}
})
.filter((rule): rule is CSSStyleRule => rule instanceof CSSStyleRule)
.filter(rule => rule.selectorText === '::-webkit-scrollbar-thumb:hover')
.map(rule => rule.style.getPropertyValue('background-color'))
const style = getComputedStyle(list)
apps/web/tests/turn-tail-actions.e2e.ts
+50 -3
View File
@@ -12,14 +12,15 @@ import { tmpdir } from 'node:os'
import { fileURLToPath } from 'node:url'
import { join } from 'node:path'
import type { Browser, Locator, Page } from 'playwright'
import { chromium } from 'playwright'
import { afterEach, describe, expect, it, onTestFailed, vi } from 'vitest'
import { chromium, devices, webkit } from 'playwright'
import { afterEach, describe, expect, it, onTestFailed, onTestFinished, vi } from 'vitest'
import type { ReplayOverrideDoc } from '@deepseek-ai/dsh-llm-replay'
import type { SessionEvent } from '@deepseek-ai/dsh-session'
import {
acknowledgeReloadConnectionLoss, assertFixtureInventory, captureStableAria, compareOrRefreshGolden, fixtureUserPrompts,
launchWebScaffold, recordFixture, watchConsole, webSnapshotMode, type WebScaffold,
launchWebScaffold, recordFixture, seedSession, watchConsole, webSnapshotMode, type WebScaffold,
} from './scaffold.ts'
import { createChatScrollFixture } from './chat-scroll-fixture.ts'
import { openSettings, connectFreshWorkspace, expandOwningTurnProcess, newEnglishPage, saveFailureShot, WEB_FIXTURE_TIME } from './support.ts'
const SNAPSHOT_DIR = fileURLToPath(new URL('../../../snapshots/web/turn-tail-actions', import.meta.url))
@@ -347,3 +348,49 @@ describe('web e2e: assistant IconActions wait for the turn to end', () => {
)
})
})
it('keeps copy tooltips hidden during a WebKit touch copy', async () => {
const scaffold = await launchWebScaffold({})
const browserPromise = webkit.launch()
onTestFinished(async () => {
try { await (await browserPromise).close() }
finally { await scaffold.close() }
})
const browser = await browserPromise
const context = await browser.newContext({
...devices['iPhone 13'], locale: 'en-US', viewport: { width: 1280, height: 900 },
})
const page = await context.newPage()
// LAN HTTP pages use the clipboard fallback on iOS.
await page.addInitScript(() => {
Object.defineProperty(navigator, 'clipboard', { configurable: true, value: undefined })
})
const fixture = createChatScrollFixture({ markerPrefix: 'TOOLTIP_TOUCH', title: 'Touch tooltips', turns: 1 })
await seedSession(scaffold, fixture.log, 'tooltip-touch-web-e2e')
await page.goto(scaffold.authenticatedUrl, { waitUntil: 'load' })
await connectFreshWorkspace(page, scaffold.workspaceCwd, 'tooltip-touch')
await page.getByRole('button', { name: 'Search sessions' }).click()
await page.getByRole('textbox', { name: 'Search session names', exact: true }).fill(fixture.markers.user(1))
const result = page.getByRole('tree', { name: 'Search results' }).getByRole('treeitem')
await expect.poll(() => result.count(), { timeout: 60_000 }).toBe(1)
await result.click()
await page.getByText(fixture.markers.assistant(1), { exact: false }).last().waitFor()
await page.setViewportSize({ width: 390, height: 844 })
expect(await page.evaluate(() => matchMedia('(any-hover: hover)').matches)).toBe(false)
const copy = page.getByRole('button', { name: 'Copy', exact: true }).last()
await copy.tap()
await expect.poll(() => page.getByRole('button', { name: 'Copied', exact: true }).count()).toBe(1)
const duringCopy = await page.getByRole('tooltip').count()
expect(duringCopy).toBe(0)
await expect.poll(() => page.getByRole('button', { name: 'Copied', exact: true }).count(), { timeout: 10_000 }).toBe(0)
// iOS may synthesize mouse entry when the checkmark is replaced by the copy icon.
await copy.dispatchEvent('pointerover', { pointerType: 'mouse', bubbles: true })
const afterCopy = await page.getByRole('tooltip').count()
expect(afterCopy).toBe(0)
await compareOrRefreshGolden(
fileURLToPath(new URL('./expected/tooltip-touch/copy.expected.md', import.meta.url)),
['# Touch copy feedback', '', 'Copy succeeds and the button returns from Copied to Copy.',
`Tooltips during copy feedback: ${duringCopy}`, `Tooltips after emulated mouse entry: ${afterCopy}`].join('\n'),
webSnapshotMode(),
)
})
apps/web/tests/voice-setup.e2e.ts
+1 -1
View File
@@ -32,7 +32,7 @@ it('guides a newly enabled voice plugin to installation and lets the user postpo
await dialog.waitFor({ state: 'hidden' })
expect(await toggle.getAttribute('aria-checked')).toBe('true')
const card = page.locator('[data-plugin-package="@deepseek-ai/dsh-experimental-voice-input-bundle"]')
expect(await card.getByRole('status').count()).toBe(0)
expect(await card.getByRole('status').textContent()).toBe('')
await toggle.click()
await page.getByRole('switch', { name: 'Enable Voice input', exact: true, checked: false }).click()
await dialog.getByRole('button', { name: 'Go to setup', exact: true }).click()
apps/web/tsconfig.json
+1
View File
@@ -23,6 +23,7 @@
// this client-registered project must not also hold them (one program
// cannot see both sides of the cordis Context merges).
"exclude": [
"tests/hover-input.e2e.ts",
"tests/collapse-timing.e2e.ts",
"tests/notice-wake-tail.e2e.ts",
"tests/fold-blank-matrix.e2e.ts",
docs/architecture.i18n.yaml
+4 -4
View File
@@ -18,11 +18,11 @@
en: b9eeb18645121a8b
zh: 70d87a12c43e042a
/deepseek-harness-architecture/core-packages:
en: f4dde8809a713bad
zh: ccbee814b4e8b4b7
en: 20ef882e44b06a10
zh: c10c3934ed5570e9
/deepseek-harness-architecture/events:
en: c00467ff3a210ef5
zh: 7619d7cca723fa6b
en: e89802f066e4955d
zh: ea2fb4bef94e086d
/deepseek-harness-architecture/turn-flow:
en: f270fd4ca334d417
zh: 3c789548e555508e
docs/architecture.md
-2
View File
@@ -61,7 +61,6 @@ Electron starts the private Desktop Host in Electron Node mode. The Host invokes
| [`core/tools`](subsystems/tools.md) | The scoped tool registry and guarded execution pipeline | `ctx.tools` |
| [`core/agent`](subsystems/core.md) | The `Agent` interface, live registry, and `agent/*` events | `ctx.agents` |
| [`core/agent-loop`](subsystems/core.md) | The default driver implementing that interface | `ctx.agentLoop` |
| [`core/agent-controller`](subsystems/core.md#messages-from-outside-an-agent) | Admission and delivery of outside messages | `ctx.agentController` |
| [`core/scope`](subsystems/scope.md) | The per-agent scoped-registration primitive | library, no key |
| [`llm/llm`](subsystems/llm-streaming.md) | Message and stream vocabulary plus the adapter seam | `ctx.llm` |
@@ -71,7 +70,6 @@ Events are the extension points, and picking the right domain is the first decis
- **Session events** are durable facts appended to the log and broadcast through `session/event`. Use one when the fact must survive a reload.
- **Agent events** (`agent/*`) carry a live `Agent`: inbox, step, status, request, validation, continuation. Use one to observe or intercept work in flight.
- **Delivery admission** (`agent-controller/admission`) holds or refuses outside input before it enters the inbox.
- **Capability events** attach policy and adapters to a seam (`fs/*`, `tools/*`, `telemetry/*`) without importing the loop.
AgentLoop awaits serial `agent/created` before queued work and rolls back failed initialization. Unload retains resources through Agent teardown; [agent-loop](../packages/core/agent-loop/README.md#understand-the-implementation) defines ordering.
docs/architecture.zh.md
-2
View File
@@ -61,7 +61,6 @@ Electron 使用 Electron Node 模式启动私有 Desktop Host。Host 调用共
| [`core/tools`](subsystems/tools.zh.md) | 作用域化的工具注册表和带把关的执行流水线 | `ctx.tools` |
| [`core/agent`](subsystems/core.zh.md) | `Agent` 接口、活跃 agent 注册表和 `agent/*` 事件 | `ctx.agents` |
| [`core/agent-loop`](subsystems/core.zh.md) | 实现该接口的默认驱动器 | `ctx.agentLoop` |
| [`core/agent-controller`](subsystems/core.zh.md#messages-from-outside-an-agent) | 外部消息的准入与投递 | `ctx.agentController` |
| [`core/scope`](subsystems/scope.zh.md) | 按 agent 划分作用域的注册原语 | 库，无 ctx 键 |
| [`llm/llm`](subsystems/llm-streaming.zh.md) | 消息与流式词汇表，以及适配器 seam | `ctx.llm` |
@@ -73,7 +72,6 @@ Electron 使用 Electron Node 模式启动私有 Desktop Host。Host 调用共
- **会话事件**是追加到日志并通过 `session/event` 广播的持久事实。当某个事实必须在重新加载后仍然存在时，使用它。
- **Agent 事件**（`agent/*`）携带活跃 `Agent`：inbox、步骤、状态、请求、验证、续跑。要观察或拦截进行中的工作时，使用它。
- **投递准入**（`agent-controller/admission`）在外部输入进入收件箱之前暂留或拒绝它。
- **能力事件**无需导入循环即可向某个 seam（`fs/*`、`tools/*`、`telemetry/*`）附加策略和适配器。
AgentLoop 在启动已排队工作前等待串行 `agent/created`，初始化失败则回滚。卸载在 Agent teardown 期间保留所需资源；[agent-loop](../packages/core/agent-loop/README.zh.md#understand-the-implementation)定义顺序。
docs/capability-seams.i18n.yaml
+1 -1
View File
@@ -6,5 +6,5 @@
en: 41d772192075e133
zh: 5ef3da69b4de1d8a
/capability-seams-and-core-services:
en: fc474f60e98a897e
en: b6d09bcd8359cbd2
zh: b64ec0217ca2bd3f
docs/capability-seams.md
-5
View File
@@ -193,8 +193,6 @@ flowchart LR
pkg_acp["acp"]
pkg_agent_model["agent-model"]
svc_agentModel["ctx.agentModel<br/>Session model selection"]
pkg_agent_controller["agent-controller"]
svc_agentController["ctx.agentController<br/>Outside-message delivery"]
svc_agentDefaultModel["ctx.agentDefaultModel<br/>Default Agent model selection"]
svc_agentLoop["ctx.agentLoop<br/>Concrete loop driver"]
pkg_base["base"]
@@ -302,7 +300,6 @@ flowchart LR
svc_dynamicCordisRunner["ctx.dynamicCordisRunner<br/>Dynamic Cordis package host runner"]
svc_cordisInspect["ctx.cordisInspect<br/>Dynamic Cordis inspect registry"]
pkg_agent --> svc_agents
pkg_agent_controller --> svc_agentController
pkg_agent_default_model --> svc_agentDefaultModel
pkg_agent_loop --> svc_agentLoop
pkg_agent_model --> svc_agentModel
@@ -462,7 +459,6 @@ flowchart LR
pkg_working_directory --> svc_workingDirectory
pkg_workspace --> svc_workspaceRegistry
pkg_workspace_changes --> svc_workspaceChanges
svc_agentController --> pkg_api_session_controller
svc_agentDefaultModel --> pkg_api_session_controller
svc_agentDefaultModel --> pkg_headless
svc_agentLoop --> pkg_base
@@ -699,7 +695,6 @@ flowchart LR
| `ctx.skills` | `seam` | [`skill`](../packages/skill/skill) | [`sandbox-windows-acl`](../packages/sandbox/sandbox-windows-acl), [`skill-badge`](../packages/experimental/skill-badge), [`skill-filesystem`](../packages/skill/skill-filesystem), [`skill-office`](../packages/skill/skill-office) | [`tool-skill`](../packages/skill/tool-skill) | - | Merges provider skill catalogs; tool-skill renders the session-prefix catalog and loads complete skill bodies. |
| `ctx.agents` | `core` | [`agent`](../packages/core/agent) | - | [`agent-loop`](../packages/core/agent-loop), [`acp`](../packages/acp/acp), [`subagent`](../packages/subagent/subagent) | - | Owns live Agent handles, the create/resume factory seam, and process-local initiator propagation. |
| `ctx.agentModel` | `core` | [`agent-model`](../packages/core/agent-model) | - | [`agent-loop`](../packages/core/agent-loop), [`api-session-controller`](../packages/api/session-controller), [`acp`](../packages/acp/acp), [`sdk-jsonrpc-server`](../packages/sdk/server), [`headless`](../packages/bundle/headless) | - | Reads and records Session model choices; owns the modelSelection projection and optional scoped route-switch notices. |
| `ctx.agentController` | `core` | [`agent-controller`](../packages/core/agent-controller) | - | [`api-session-controller`](../packages/api/session-controller) | - | Prepares, admits through the agent-controller/admission stage, and appends messages that senders bring on someone's behalf, one at a time per Agent. |
| `ctx.agentDefaultModel` | `core` | [`agent-default-model`](../packages/core/agent-default-model) | - | [`api-session-controller`](../packages/api/session-controller), [`headless`](../packages/bundle/headless) | - | Reads the default ModelSelection from volatile Config and saves selections through the profile editor. |
| `ctx.agentLoop` | `bundle` | [`agent-loop`](../packages/core/agent-loop) | - | [`base`](../packages/bundle/base), [`sdk-minimal`](../packages/bundle/sdk-minimal) | - | The one concrete loop plugin; extension packages depend on dsh-agent events and services, not on this package. |
| `ctx.schedule` | `core` | [`schedule`](../packages/schedule/schedule) | - | - | - | Stores tasks independently of Session activation and queues due messages in the original Session. |
docs/capability-seams.zh.md
-2
View File
@@ -304,7 +304,6 @@ flowchart LR
svc_dynamicCordisRunner["ctx.dynamicCordisRunner<br/>Dynamic Cordis package host runner"]
svc_cordisInspect["ctx.cordisInspect<br/>Dynamic Cordis inspect registry"]
pkg_agent --> svc_agents
pkg_agent_controller --> svc_agentController
pkg_agent_default_model --> svc_agentDefaultModel
pkg_agent_loop --> svc_agentLoop
pkg_agent_model --> svc_agentModel
@@ -464,7 +463,6 @@ flowchart LR
pkg_working_directory --> svc_workingDirectory
pkg_workspace --> svc_workspaceRegistry
pkg_workspace_changes --> svc_workspaceChanges
svc_agentController --> pkg_api_session_controller
svc_agentDefaultModel --> pkg_api_session_controller
svc_agentDefaultModel --> pkg_headless
svc_agentLoop --> pkg_base
docs/config-catalog.md
+1 -2
View File
@@ -225,7 +225,7 @@ export interface Config {
## `@deepseek-ai/dsh-api-session-controller`
- `inject`: `agentController` · `agentDefaultModel` · `agents` · `attachments` · `fileUploads` · `fs` · `llm` · `remoteAccess` · `sessions` · `sessionProjections` · `agentModel` · `sessionQuery` · `typert` · `workspaceRegistry`
- `inject`: `agentDefaultModel` · `agents` · `attachments` · `fileUploads` · `fs` · `llm` · `remoteAccess` · `sessions` · `sessionProjections` · `agentModel` · `sessionQuery` · `typert` · `workspaceRegistry`
- `source`: [`packages/api/session-controller/src/index.ts:94`](../packages/api/session-controller/src/index.ts)
```ts config-catalog
@@ -4698,7 +4698,6 @@ These load from a `cordis.yml` entry with no `config:` block; they declare no co
| --- | --- | --- |
| `@deepseek-ai/dsh-acp-app` | `cmdlineArgs` | [`packages/bundle/acp-app/src/index.ts`](../packages/bundle/acp-app/src/index.ts) |
| `@deepseek-ai/dsh-agent` | `sessionProjections` | [`packages/core/agent/src/index.ts`](../packages/core/agent/src/index.ts) |
| `@deepseek-ai/dsh-agent-controller` | `agents` | [`packages/core/agent-controller/src/index.ts`](../packages/core/agent-controller/src/index.ts) |
| `@deepseek-ai/dsh-agent-model` | `sessionProjections` | [`packages/core/agent-model/src/index.ts`](../packages/core/agent-model/src/index.ts) |
| `@deepseek-ai/dsh-api-account-controller` | `deepseekAccount` · `agents` | [`packages/api/account-controller/src/index.ts`](../packages/api/account-controller/src/index.ts) |
| `@deepseek-ai/dsh-api-remote-access` | — | [`packages/api/remote-access/src/index.ts`](../packages/api/remote-access/src/index.ts) |
docs/config-catalog.zh.md
+1 -2
View File
@@ -227,7 +227,7 @@ export interface Config {
## `@deepseek-ai/dsh-api-session-controller`
- `inject`: `agentController` · `agentDefaultModel` · `agents` · `attachments` · `fileUploads` · `fs` · `llm` · `remoteAccess` · `sessions` · `sessionProjections` · `agentModel` · `sessionQuery` · `typert` · `workspaceRegistry`
- `inject`: `agentDefaultModel` · `agents` · `attachments` · `fileUploads` · `fs` · `llm` · `remoteAccess` · `sessions` · `sessionProjections` · `agentModel` · `sessionQuery` · `typert` · `workspaceRegistry`
- `source`: [`packages/api/session-controller/src/index.ts:94`](../packages/api/session-controller/src/index.ts)
```ts config-catalog
@@ -4700,7 +4700,6 @@ export interface Config {
| --- | --- | --- |
| `@deepseek-ai/dsh-acp-app` | `cmdlineArgs` | [`packages/bundle/acp-app/src/index.ts`](../packages/bundle/acp-app/src/index.ts) |
| `@deepseek-ai/dsh-agent` | `sessionProjections` | [`packages/core/agent/src/index.ts`](../packages/core/agent/src/index.ts) |
| `@deepseek-ai/dsh-agent-controller` | `agents` | [`packages/core/agent-controller/src/index.ts`](../packages/core/agent-controller/src/index.ts) |
| `@deepseek-ai/dsh-agent-model` | `sessionProjections` | [`packages/core/agent-model/src/index.ts`](../packages/core/agent-model/src/index.ts) |
| `@deepseek-ai/dsh-api-account-controller` | `deepseekAccount` · `agents` | [`packages/api/account-controller/src/index.ts`](../packages/api/account-controller/src/index.ts) |
| `@deepseek-ai/dsh-api-remote-access` | — | [`packages/api/remote-access/src/index.ts`](../packages/api/remote-access/src/index.ts) |
docs/event-producer-consumer.md
+5 -6
View File
@@ -8,7 +8,6 @@ This matrix shows which packages dispatch each harness-owned event and which pac
<!-- BEGIN GENERATED event-producer-consumer:events -->
| Event | Mode | Declared in | Dispatchers | Listeners |
| --- | --- | --- | --- | --- |
| `agent-controller/admission` | `waterfall` | [`packages/core/agent-controller/src/types.ts:79`](../packages/core/agent-controller/src/types.ts) | [`agent-controller`](../packages/core/agent-controller) (`waterfall`) | - |
| `agent-loop/config-start-failed` | `emit` | [`packages/core/agent-loop/src/index.ts:197`](../packages/core/agent-loop/src/index.ts) | [`agent-loop`](../packages/core/agent-loop) (`events.dispatch`) | - |
| `agent-preset/selected` | `emit` | [`packages/preset/agent-preset-registry/src/types.ts:66`](../packages/preset/agent-preset-registry/src/types.ts) | [`agent-preset-registry`](../packages/preset/agent-preset-registry) (`emit`) | [`cua`](../packages/experimental/cua), `remotes` |
| `agent/assistant-stream` | `emit` | [`packages/core/agent/src/runtime-types.ts:376`](../packages/core/agent/src/runtime-types.ts) | [`agent-loop`](../packages/core/agent-loop) (`emit`) | [`headless`](../packages/bundle/headless), `session-controller` |
@@ -22,11 +21,11 @@ This matrix shows which packages dispatch each harness-owned event and which pac
| `agent/request-error` | `waterfall` | [`packages/core/agent/src/runtime-types.ts:366`](../packages/core/agent/src/runtime-types.ts) | [`agent-loop`](../packages/core/agent-loop) (`waterfall`) | [`compaction-basic`](../packages/compaction/compaction-basic), [`compaction-image-offload`](../packages/compaction/compaction-image-offload), [`llm-retry`](../packages/llm/llm-retry) |
| `agent/status` | `emit` | [`packages/core/agent/src/runtime-types.ts:286`](../packages/core/agent/src/runtime-types.ts) | [`agent-loop`](../packages/core/agent-loop) (`emit`) | [`acp`](../packages/acp/acp), `agent-team`, [`compaction-basic`](../packages/compaction/compaction-basic), [`goal-round-driver`](../packages/goal/goal-round-driver), `server`, `session-controller` |
| `agent/turn-stopping` | `serial` | [`packages/core/agent/src/runtime-types.ts:394`](../packages/core/agent/src/runtime-types.ts) | [`agent-loop`](../packages/core/agent-loop) (`serial`) | [`hooks-claude-code`](../packages/experimental/hooks-claude-code), [`hooks-codex`](../packages/experimental/hooks-codex), [`workspace-changes`](../packages/deliverables/workspace-changes) |
| `api-session/activity` | `emit` | [`packages/api/session-controller/src/types.ts:605`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/added` | `emit` | [`packages/api/session-controller/src/types.ts:585`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/error` | `emit` | [`packages/api/session-controller/src/types.ts:612`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/removed` | `emit` | [`packages/api/session-controller/src/types.ts:591`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/status` | `emit` | [`packages/api/session-controller/src/types.ts:598`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/activity` | `emit` | [`packages/api/session-controller/src/types.ts:599`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/added` | `emit` | [`packages/api/session-controller/src/types.ts:579`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/error` | `emit` | [`packages/api/session-controller/src/types.ts:606`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/removed` | `emit` | [`packages/api/session-controller/src/types.ts:585`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/status` | `emit` | [`packages/api/session-controller/src/types.ts:592`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `app-boot/config-reload` | `emit` | [`packages/boot/app-boot/src/index.ts:54`](../packages/boot/app-boot/src/index.ts) | [`app-boot`](../packages/boot/app-boot) (`emit`) | [`settings`](../packages/settings/settings) |
| `approval/request` | `waterfall` | [`packages/interaction/user-approval/src/types.ts:87`](../packages/interaction/user-approval/src/types.ts) | [`user-approval`](../packages/interaction/user-approval) (`waterfall`) | [`acp`](../packages/acp/acp), `remotes` |
| `authorization/settled` | `emit` | [`packages/credentials/authorization/src/index.ts:57`](../packages/credentials/authorization/src/index.ts) | [`authorization`](../packages/credentials/authorization) (`events.dispatch`) | - |
docs/event-producer-consumer.zh.md
+5 -6
View File
@@ -10,7 +10,6 @@
<!-- BEGIN GENERATED event-producer-consumer:events -->
| Event | Mode | Declared in | Dispatchers | Listeners |
| --- | --- | --- | --- | --- |
| `agent-controller/admission` | `waterfall` | [`packages/core/agent-controller/src/types.ts:79`](../packages/core/agent-controller/src/types.ts) | [`agent-controller`](../packages/core/agent-controller) (`waterfall`) | - |
| `agent-loop/config-start-failed` | `emit` | [`packages/core/agent-loop/src/index.ts:197`](../packages/core/agent-loop/src/index.ts) | [`agent-loop`](../packages/core/agent-loop) (`events.dispatch`) | - |
| `agent-preset/selected` | `emit` | [`packages/preset/agent-preset-registry/src/types.ts:66`](../packages/preset/agent-preset-registry/src/types.ts) | [`agent-preset-registry`](../packages/preset/agent-preset-registry) (`emit`) | [`cua`](../packages/experimental/cua), `remotes` |
| `agent/assistant-stream` | `emit` | [`packages/core/agent/src/runtime-types.ts:376`](../packages/core/agent/src/runtime-types.ts) | [`agent-loop`](../packages/core/agent-loop) (`emit`) | [`headless`](../packages/bundle/headless), `session-controller` |
@@ -24,11 +23,11 @@
| `agent/request-error` | `waterfall` | [`packages/core/agent/src/runtime-types.ts:366`](../packages/core/agent/src/runtime-types.ts) | [`agent-loop`](../packages/core/agent-loop) (`waterfall`) | [`compaction-basic`](../packages/compaction/compaction-basic), [`compaction-image-offload`](../packages/compaction/compaction-image-offload), [`llm-retry`](../packages/llm/llm-retry) |
| `agent/status` | `emit` | [`packages/core/agent/src/runtime-types.ts:286`](../packages/core/agent/src/runtime-types.ts) | [`agent-loop`](../packages/core/agent-loop) (`emit`) | [`acp`](../packages/acp/acp), `agent-team`, [`compaction-basic`](../packages/compaction/compaction-basic), [`goal-round-driver`](../packages/goal/goal-round-driver), `server`, `session-controller` |
| `agent/turn-stopping` | `serial` | [`packages/core/agent/src/runtime-types.ts:394`](../packages/core/agent/src/runtime-types.ts) | [`agent-loop`](../packages/core/agent-loop) (`serial`) | [`hooks-claude-code`](../packages/experimental/hooks-claude-code), [`hooks-codex`](../packages/experimental/hooks-codex), [`workspace-changes`](../packages/deliverables/workspace-changes) |
| `api-session/activity` | `emit` | [`packages/api/session-controller/src/types.ts:605`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/added` | `emit` | [`packages/api/session-controller/src/types.ts:585`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/error` | `emit` | [`packages/api/session-controller/src/types.ts:612`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/removed` | `emit` | [`packages/api/session-controller/src/types.ts:591`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/status` | `emit` | [`packages/api/session-controller/src/types.ts:598`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/activity` | `emit` | [`packages/api/session-controller/src/types.ts:599`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/added` | `emit` | [`packages/api/session-controller/src/types.ts:579`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/error` | `emit` | [`packages/api/session-controller/src/types.ts:606`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/removed` | `emit` | [`packages/api/session-controller/src/types.ts:585`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `api-session/status` | `emit` | [`packages/api/session-controller/src/types.ts:592`](../packages/api/session-controller/src/types.ts) | `session-controller` (`emit`) | `remotes` |
| `app-boot/config-reload` | `emit` | [`packages/boot/app-boot/src/index.ts:54`](../packages/boot/app-boot/src/index.ts) | [`app-boot`](../packages/boot/app-boot) (`emit`) | [`settings`](../packages/settings/settings) |
| `approval/request` | `waterfall` | [`packages/interaction/user-approval/src/types.ts:87`](../packages/interaction/user-approval/src/types.ts) | [`user-approval`](../packages/interaction/user-approval) (`waterfall`) | [`acp`](../packages/acp/acp), `remotes` |
| `authorization/settled` | `emit` | [`packages/credentials/authorization/src/index.ts:57`](../packages/credentials/authorization/src/index.ts) | [`authorization`](../packages/credentials/authorization) (`events.dispatch`) | - |
docs/module-graph.md
-6
View File
@@ -41,7 +41,6 @@ flowchart TD
end
subgraph group_core["packages/core"]
pkg_agent["agent"]
pkg_agent_controller["agent-controller"]
pkg_agent_default_model["agent-default-model"]
pkg_agent_loop["agent-loop"]
pkg_agent_model["agent-model"]
@@ -642,9 +641,6 @@ flowchart TD
pkg_llm_retry --> pkg_session
pkg_llm_retry --> pkg_session_projection
pkg_llm_retry --> pkg_timeout
pkg_agent_controller --> pkg_agent
pkg_agent_controller --> pkg_llm
pkg_agent_controller --> pkg_scope
pkg_agent_default_model --> pkg_agent
pkg_agent_default_model --> pkg_llm
pkg_agent_model --> pkg_agent
@@ -1283,7 +1279,6 @@ flowchart TD
pkg_tool_subagent_control --> pkg_subagent
pkg_tool_subagent_control --> pkg_tools
pkg_api_session_controller --> pkg_agent
pkg_api_session_controller --> pkg_agent_controller
pkg_api_session_controller --> pkg_agent_default_model
pkg_api_session_controller --> pkg_agent_model
pkg_api_session_controller --> pkg_agent_preset_registry
@@ -1667,7 +1662,6 @@ flowchart TD
| [`pwsh-local`](../packages/shell/pwsh-local) | `shell` | [`shell`](../packages/shell/shell), [`subprocess`](../packages/subprocess/subprocess), [`timeout`](../packages/util/timeout) |
| [`llm-deepseek-api-key`](../packages/llm/llm-deepseek-api-key) | `llm` | [`credentials`](../packages/credentials/credentials), [`launch-environment`](../packages/util/launch-environment), [`llm`](../packages/llm/llm), [`llm-deepseek`](../packages/llm/llm-deepseek) |
| [`llm-retry`](../packages/llm/llm-retry) | `llm` | [`agent`](../packages/core/agent), [`brand`](../packages/util/brand), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`session-projection`](../packages/session/session-projection), [`timeout`](../packages/util/timeout) |
| [`agent-controller`](../packages/core/agent-controller) | `core` | [`agent`](../packages/core/agent), [`llm`](../packages/llm/llm), [`scope`](../packages/core/scope) |
| [`agent-default-model`](../packages/core/agent-default-model) | `core` | [`agent`](../packages/core/agent), [`llm`](../packages/llm/llm) |
| [`agent-model`](../packages/core/agent-model) | `core` | [`agent`](../packages/core/agent), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`session-projection`](../packages/session/session-projection) |
| [`goal`](../packages/goal/goal) | `goal` | [`agent`](../packages/core/agent), [`brand`](../packages/util/brand), [`llm`](../packages/llm/llm), [`scope`](../packages/core/scope), [`session`](../packages/core/session), [`session-projection`](../packages/session/session-projection), [`typert-protocol`](../packages/typert/protocol) |
docs/module-graph.zh.md
-6
View File
@@ -43,7 +43,6 @@ flowchart TD
end
subgraph group_core["packages/core"]
pkg_agent["agent"]
pkg_agent_controller["agent-controller"]
pkg_agent_default_model["agent-default-model"]
pkg_agent_loop["agent-loop"]
pkg_agent_model["agent-model"]
@@ -644,9 +643,6 @@ flowchart TD
pkg_llm_retry --> pkg_session
pkg_llm_retry --> pkg_session_projection
pkg_llm_retry --> pkg_timeout
pkg_agent_controller --> pkg_agent
pkg_agent_controller --> pkg_llm
pkg_agent_controller --> pkg_scope
pkg_agent_default_model --> pkg_agent
pkg_agent_default_model --> pkg_llm
pkg_agent_model --> pkg_agent
@@ -1285,7 +1281,6 @@ flowchart TD
pkg_tool_subagent_control --> pkg_subagent
pkg_tool_subagent_control --> pkg_tools
pkg_api_session_controller --> pkg_agent
pkg_api_session_controller --> pkg_agent_controller
pkg_api_session_controller --> pkg_agent_default_model
pkg_api_session_controller --> pkg_agent_model
pkg_api_session_controller --> pkg_agent_preset_registry
@@ -1669,7 +1664,6 @@ flowchart TD
| [`pwsh-local`](../packages/shell/pwsh-local) | `shell` | [`shell`](../packages/shell/shell), [`subprocess`](../packages/subprocess/subprocess), [`timeout`](../packages/util/timeout) |
| [`llm-deepseek-api-key`](../packages/llm/llm-deepseek-api-key) | `llm` | [`credentials`](../packages/credentials/credentials), [`launch-environment`](../packages/util/launch-environment), [`llm`](../packages/llm/llm), [`llm-deepseek`](../packages/llm/llm-deepseek) |
| [`llm-retry`](../packages/llm/llm-retry) | `llm` | [`agent`](../packages/core/agent), [`brand`](../packages/util/brand), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`session-projection`](../packages/session/session-projection), [`timeout`](../packages/util/timeout) |
| [`agent-controller`](../packages/core/agent-controller) | `core` | [`agent`](../packages/core/agent), [`llm`](../packages/llm/llm), [`scope`](../packages/core/scope) |
| [`agent-default-model`](../packages/core/agent-default-model) | `core` | [`agent`](../packages/core/agent), [`llm`](../packages/llm/llm) |
| [`agent-model`](../packages/core/agent-model) | `core` | [`agent`](../packages/core/agent), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`session-projection`](../packages/session/session-projection) |
| [`goal`](../packages/goal/goal) | `goal` | [`agent`](../packages/core/agent), [`brand`](../packages/util/brand), [`llm`](../packages/llm/llm), [`scope`](../packages/core/scope), [`session`](../packages/core/session), [`session-projection`](../packages/session/session-projection), [`typert-protocol`](../packages/typert/protocol) |
docs/persistence-catalog.i18n.yaml
+4 -4
View File
@@ -1347,8 +1347,8 @@
en: 2786fef560851bb4
zh: 5d787475356b2606
/session-persistence-event-catalog/resolved-persistence-types/string:
en: 0780373a5ca32b81
zh: 84c50d1091fb03b9
en: 967168de7dce17f6
zh: c43778dcbfea301e
/session-persistence-event-catalog/resolved-persistence-types/string~2:
en: 0089620bf4638263
zh: 1922bef1713094bb
@@ -1833,8 +1833,8 @@
en: 809a23c25ded8497
zh: e9cd23e281f8d31f
/session-persistence-event-catalog/resolved-persistence-types/kind-user:
en: 5fe56e85e30e0688
zh: e99526919f202f59
en: 43f0c6042d410afd
zh: a0b68de4e5d220bd
/session-persistence-event-catalog/resolved-persistence-types/kind-user-approval:
en: a52d2dfbaf844127
zh: 4c5306ca40f20873
docs/persistence-catalog.md
+2 -2
View File
@@ -5583,7 +5583,7 @@ Array of `number`.
SHA-256: `4cd48f3e5108bd6ebedf7301c3638839b8b2b9db8dbd493974177fbdeb91a01b`
Sources: [`packages/api/session-controller/src/types.ts:385`](../packages/api/session-controller/src/types.ts) · [`packages/attachment/attachment/src/brand.ts:6`](../packages/attachment/attachment/src/brand.ts) · [`packages/compaction/compaction/src/brand.ts:4`](../packages/compaction/compaction/src/brand.ts) · [`packages/core/session/src/types.ts:21`](../packages/core/session/src/types.ts) · [`packages/experimental/agent-team/src/types.ts:21`](../packages/experimental/agent-team/src/types.ts) · [`packages/experimental/agent-team/src/types.ts:33`](../packages/experimental/agent-team/src/types.ts) · [`packages/experimental/agent-team/src/types.ts:9`](../packages/experimental/agent-team/src/types.ts) · [`packages/experimental/webhook/src/brand.ts:12`](../packages/experimental/webhook/src/brand.ts) · [`packages/experimental/webhook/src/brand.ts:6`](../packages/experimental/webhook/src/brand.ts) · [`packages/experimental/webhook/src/brand.ts:9`](../packages/experimental/webhook/src/brand.ts) · [`packages/feedback/message-feedback/src/types.ts:14`](../packages/feedback/message-feedback/src/types.ts) · [`packages/goal/goal/src/types.ts:17`](../packages/goal/goal/src/types.ts) · [`packages/interaction/commands/src/brand.ts:31`](../packages/interaction/commands/src/brand.ts) · [`packages/interaction/user-approval/src/types.ts:17`](../packages/interaction/user-approval/src/types.ts) · [`packages/jobs/jobs/src/brand.ts:19`](../packages/jobs/jobs/src/brand.ts) · [`packages/llm/llm-retry/src/brand.ts:4`](../packages/llm/llm-retry/src/brand.ts) · [`packages/llm/llm/src/brand.ts:16`](../packages/llm/llm/src/brand.ts) · [`packages/llm/llm/src/brand.ts:31`](../packages/llm/llm/src/brand.ts) · [`packages/llm/llm/src/brand.ts:43`](../packages/llm/llm/src/brand.ts) · [`packages/llm/llm/src/brand.ts:67`](../packages/llm/llm/src/brand.ts) · [`packages/schedule/schedule/src/types.ts:14`](../packages/schedule/schedule/src/types.ts) · [`packages/session/session-title/src/index.ts:42`](../packages/session/session-title/src/index.ts) · [`packages/session/session-title/src/types.ts:17`](../packages/session/session-title/src/types.ts) · [`packages/workflow/workflow/src/types.ts:13`](../packages/workflow/workflow/src/types.ts)
Sources: [`packages/api/session-controller/src/types.ts:379`](../packages/api/session-controller/src/types.ts) · [`packages/attachment/attachment/src/brand.ts:6`](../packages/attachment/attachment/src/brand.ts) · [`packages/compaction/compaction/src/brand.ts:4`](../packages/compaction/compaction/src/brand.ts) · [`packages/core/session/src/types.ts:21`](../packages/core/session/src/types.ts) · [`packages/experimental/agent-team/src/types.ts:21`](../packages/experimental/agent-team/src/types.ts) · [`packages/experimental/agent-team/src/types.ts:33`](../packages/experimental/agent-team/src/types.ts) · [`packages/experimental/agent-team/src/types.ts:9`](../packages/experimental/agent-team/src/types.ts) · [`packages/experimental/webhook/src/brand.ts:12`](../packages/experimental/webhook/src/brand.ts) · [`packages/experimental/webhook/src/brand.ts:6`](../packages/experimental/webhook/src/brand.ts) · [`packages/experimental/webhook/src/brand.ts:9`](../packages/experimental/webhook/src/brand.ts) · [`packages/feedback/message-feedback/src/types.ts:14`](../packages/feedback/message-feedback/src/types.ts) · [`packages/goal/goal/src/types.ts:17`](../packages/goal/goal/src/types.ts) · [`packages/interaction/commands/src/brand.ts:31`](../packages/interaction/commands/src/brand.ts) · [`packages/interaction/user-approval/src/types.ts:17`](../packages/interaction/user-approval/src/types.ts) · [`packages/jobs/jobs/src/brand.ts:19`](../packages/jobs/jobs/src/brand.ts) · [`packages/llm/llm-retry/src/brand.ts:4`](../packages/llm/llm-retry/src/brand.ts) · [`packages/llm/llm/src/brand.ts:16`](../packages/llm/llm/src/brand.ts) · [`packages/llm/llm/src/brand.ts:31`](../packages/llm/llm/src/brand.ts) · [`packages/llm/llm/src/brand.ts:43`](../packages/llm/llm/src/brand.ts) · [`packages/llm/llm/src/brand.ts:67`](../packages/llm/llm/src/brand.ts) · [`packages/schedule/schedule/src/types.ts:14`](../packages/schedule/schedule/src/types.ts) · [`packages/session/session-title/src/index.ts:42`](../packages/session/session-title/src/index.ts) · [`packages/session/session-title/src/types.ts:17`](../packages/session/session-title/src/types.ts) · [`packages/workflow/workflow/src/types.ts:13`](../packages/workflow/workflow/src/types.ts)
`string`
@@ -7554,7 +7554,7 @@ Sources: [`packages/core/tools/src/index.ts:35`](../packages/core/tools/src/inde
SHA-256: `6dacb1584f2bfbe209dc6b6f578d51da6d55595f64d0e63c727c98f359278def`
Sources: [`packages/api/session-controller/src/types.ts:394`](../packages/api/session-controller/src/types.ts)
Sources: [`packages/api/session-controller/src/types.ts:388`](../packages/api/session-controller/src/types.ts)
| Property | Presence | Type |
|---|---|---|
docs/persistence-catalog.zh.md
+2 -2
View File
@@ -5585,7 +5585,7 @@ SHA-256: `5d03ba38734809bcbd2a55221bd938b3b3fc34b49bb686c6113e9e4931e3aa78`
SHA-256: `4cd48f3e5108bd6ebedf7301c3638839b8b2b9db8dbd493974177fbdeb91a01b`
来源：[`packages/api/session-controller/src/types.ts:385`](../packages/api/session-controller/src/types.ts) · [`packages/attachment/attachment/src/brand.ts:6`](../packages/attachment/attachment/src/brand.ts) · [`packages/compaction/compaction/src/brand.ts:4`](../packages/compaction/compaction/src/brand.ts) · [`packages/core/session/src/types.ts:21`](../packages/core/session/src/types.ts) · [`packages/experimental/agent-team/src/types.ts:21`](../packages/experimental/agent-team/src/types.ts) · [`packages/experimental/agent-team/src/types.ts:33`](../packages/experimental/agent-team/src/types.ts) · [`packages/experimental/agent-team/src/types.ts:9`](../packages/experimental/agent-team/src/types.ts) · [`packages/experimental/webhook/src/brand.ts:12`](../packages/experimental/webhook/src/brand.ts) · [`packages/experimental/webhook/src/brand.ts:6`](../packages/experimental/webhook/src/brand.ts) · [`packages/experimental/webhook/src/brand.ts:9`](../packages/experimental/webhook/src/brand.ts) · [`packages/feedback/message-feedback/src/types.ts:14`](../packages/feedback/message-feedback/src/types.ts) · [`packages/goal/goal/src/types.ts:17`](../packages/goal/goal/src/types.ts) · [`packages/interaction/commands/src/brand.ts:31`](../packages/interaction/commands/src/brand.ts) · [`packages/interaction/user-approval/src/types.ts:17`](../packages/interaction/user-approval/src/types.ts) · [`packages/jobs/jobs/src/brand.ts:19`](../packages/jobs/jobs/src/brand.ts) · [`packages/llm/llm-retry/src/brand.ts:4`](../packages/llm/llm-retry/src/brand.ts) · [`packages/llm/llm/src/brand.ts:16`](../packages/llm/llm/src/brand.ts) · [`packages/llm/llm/src/brand.ts:31`](../packages/llm/llm/src/brand.ts) · [`packages/llm/llm/src/brand.ts:43`](../packages/llm/llm/src/brand.ts) · [`packages/llm/llm/src/brand.ts:67`](../packages/llm/llm/src/brand.ts) · [`packages/schedule/schedule/src/types.ts:14`](../packages/schedule/schedule/src/types.ts) · [`packages/session/session-title/src/index.ts:42`](../packages/session/session-title/src/index.ts) · [`packages/session/session-title/src/types.ts:17`](../packages/session/session-title/src/types.ts) · [`packages/workflow/workflow/src/types.ts:13`](../packages/workflow/workflow/src/types.ts)
来源：[`packages/api/session-controller/src/types.ts:379`](../packages/api/session-controller/src/types.ts) · [`packages/attachment/attachment/src/brand.ts:6`](../packages/attachment/attachment/src/brand.ts) · [`packages/compaction/compaction/src/brand.ts:4`](../packages/compaction/compaction/src/brand.ts) · [`packages/core/session/src/types.ts:21`](../packages/core/session/src/types.ts) · [`packages/experimental/agent-team/src/types.ts:21`](../packages/experimental/agent-team/src/types.ts) · [`packages/experimental/agent-team/src/types.ts:33`](../packages/experimental/agent-team/src/types.ts) · [`packages/experimental/agent-team/src/types.ts:9`](../packages/experimental/agent-team/src/types.ts) · [`packages/experimental/webhook/src/brand.ts:12`](../packages/experimental/webhook/src/brand.ts) · [`packages/experimental/webhook/src/brand.ts:6`](../packages/experimental/webhook/src/brand.ts) · [`packages/experimental/webhook/src/brand.ts:9`](../packages/experimental/webhook/src/brand.ts) · [`packages/feedback/message-feedback/src/types.ts:14`](../packages/feedback/message-feedback/src/types.ts) · [`packages/goal/goal/src/types.ts:17`](../packages/goal/goal/src/types.ts) · [`packages/interaction/commands/src/brand.ts:31`](../packages/interaction/commands/src/brand.ts) · [`packages/interaction/user-approval/src/types.ts:17`](../packages/interaction/user-approval/src/types.ts) · [`packages/jobs/jobs/src/brand.ts:19`](../packages/jobs/jobs/src/brand.ts) · [`packages/llm/llm-retry/src/brand.ts:4`](../packages/llm/llm-retry/src/brand.ts) · [`packages/llm/llm/src/brand.ts:16`](../packages/llm/llm/src/brand.ts) · [`packages/llm/llm/src/brand.ts:31`](../packages/llm/llm/src/brand.ts) · [`packages/llm/llm/src/brand.ts:43`](../packages/llm/llm/src/brand.ts) · [`packages/llm/llm/src/brand.ts:67`](../packages/llm/llm/src/brand.ts) · [`packages/schedule/schedule/src/types.ts:14`](../packages/schedule/schedule/src/types.ts) · [`packages/session/session-title/src/index.ts:42`](../packages/session/session-title/src/index.ts) · [`packages/session/session-title/src/types.ts:17`](../packages/session/session-title/src/types.ts) · [`packages/workflow/workflow/src/types.ts:13`](../packages/workflow/workflow/src/types.ts)
`string`
@@ -7556,7 +7556,7 @@ SHA-256: `8fcb0b44a575e998665c053643d1ab04c82f39253ebd190ac0c975b78e40b543`
SHA-256: `6dacb1584f2bfbe209dc6b6f578d51da6d55595f64d0e63c727c98f359278def`
来源：[`packages/api/session-controller/src/types.ts:394`](../packages/api/session-controller/src/types.ts)
来源：[`packages/api/session-controller/src/types.ts:388`](../packages/api/session-controller/src/types.ts)
| 属性 | 存在性 | 类型 |
|---|---|---|
docs/persistence-schema.json
Generated
+2 -2
View File
@@ -23574,7 +23574,7 @@
"packages/workflow/workflow/src/types.ts#WorkflowRunId"
],
"sources": [
"packages/api/session-controller/src/types.ts:385",
"packages/api/session-controller/src/types.ts:379",
"packages/attachment/attachment/src/brand.ts:6",
"packages/compaction/compaction/src/brand.ts:4",
"packages/core/session/src/types.ts:21",
@@ -24096,7 +24096,7 @@
"digest": "6dacb1584f2bfbe209dc6b6f578d51da6d55595f64d0e63c727c98f359278def",
"names": [],
"sources": [
"packages/api/session-controller/src/types.ts:394"
"packages/api/session-controller/src/types.ts:388"
]
},
{
docs/subsystems/core.i18n.yaml
+3 -6
View File
@@ -6,8 +6,8 @@
en: f6ae07d8902e8edf
zh: c92fef531c6da331
/core/the-spine-package-by-package:
en: 60926dc99f41024d
zh: f40370b034f91463
en: c90cb23528ddb6c8
zh: 62724b7255b586b6
/core/creation-and-ownership:
en: ab09a38b4e1f0883
zh: 81e2d14c4036ab6c
@@ -16,10 +16,7 @@
zh: 5f4dd4fa9e9a88f0
/core/initiating-agent:
en: 6c8c47ac78de2dde
zh: acde99c8808ea891
/core/messages-from-outside-an-agent:
en: cf087fd9ff5edb87
zh: c3bb56b17cd76c52
zh: c9d527a25a85d1d9
/core/interception-decisions:
en: 6fccceb0a5f1cad9
zh: f43c30c2386d8cc8
docs/subsystems/core.md
+1 -125
View File
@@ -15,10 +15,9 @@ A turn flows through the six packages in one loop: the driver in [`agent-loop`](
| `tools/` | The scoped tool registry and guarded execution pipeline (`ctx.tools`) | [tools.md](tools.md) |
| `agent/` | The `Agent` interface, live registry, initiator scope, and `agent/*` event vocabulary (`ctx.agents`) | this page |
| `agent-loop/` | The concrete driver implementing the public `Agent` contract (`ctx.agentLoop`) | this page |
| `agent-controller/` | Delivery of messages from outside an Agent through the `agent-controller/admission` stage (`ctx.agentController`) | this page |
| `scope/` | The scoped-registration primitive the registries and loop build per-agent scoping on | [scope.md](scope.md) |
`scope/` is the one non-service package: a dependency-free library (`createScope`/`scopeOf`/`scopeTarget`) that sits below `session/` and `system-prompt/` in the module graph precisely so they can consume it without a cycle. `agent-loop` is the one concrete implementation of the public `Agent` contract and lives here because it is the harness's default product loop; it runs each driver inside `ctx.agents.withInitiator()`. Extension plugins depend on `agent` — including when they need the initiating Agent — and never on `agent-loop` directly, so the loop stays swappable. `agent-controller` sits in front of the loop: a sender that brings a message on someone's behalf delivers it through `ctx.agentController` instead of appending to the inbox itself. [`dsh-base`](../../packages/bundle/base/README.md) is the default product composition, while [`dsh-sdk-minimal`](../../packages/bundle/sdk-minimal/README.md) declares a smaller standalone tree.
`scope/` is the one non-service package: a dependency-free library (`createScope`/`scopeOf`/`scopeTarget`) that sits below `session/` and `system-prompt/` in the module graph precisely so they can consume it without a cycle. `agent-loop` is the one concrete implementation of the public `Agent` contract and lives here because it is the harness's default product loop; it runs each driver inside `ctx.agents.withInitiator()`. Extension plugins depend on `agent` — including when they need the initiating Agent — and never on `agent-loop` directly, so the loop stays swappable. [`dsh-base`](../../packages/bundle/base/README.md) is the default product composition, while [`dsh-sdk-minimal`](../../packages/bundle/sdk-minimal/README.md) declares a smaller standalone tree.
## Creation and ownership
@@ -348,65 +347,6 @@ The [event taxonomy](../architecture.md#events) owns the `agent/*` lifecycle, ch
The process-local initiator carried by `ctx.agents` is the exact `Agent` above, not a separate frame or copied identity. Ambient presence is neither liveness proof nor authorization; the [initiator-scope decision](../../.agents/notes/implemented/architecture/2026-07-15-agent-initiator-scope.md) defines its lifetime and scope rules.
## Messages from outside an Agent
A sender that brings a message on someone's behalf, such as the Web Session Remote, calls `ctx.agentController.deliver` from [`agent-controller`](../../packages/core/agent-controller/README.md) instead of `followup()` or `steer()`. The controller prepares the content in the Agent's delivery order, asks the `agent-controller/admission` waterfall for a decision, and appends an admitted message; messages the Agent's own machinery produces enter without the stage. The [delivery decision](../../.agents/notes/implemented/architecture/2026-10-17-agent-controller-message-delivery.md) explains why the stage lives outside `agent`.
Source: [`packages/core/agent-controller/src/types.ts`](../../packages/core/agent-controller/src/types.ts)
```ts type-equiv
/** How a delivered message enters the inbox: `followup` for its own turn, `steer` for the nearest step. */
type DeliveryMode = 'followup' | 'steer'
```
```ts type-equiv
/** One message a sender delivers on someone's behalf. */
interface DeliveryRequest {
/** Where the message enters once it is admitted. */
readonly mode: DeliveryMode
/** Durable attribution recorded on the message. */
readonly source: MessageSource
/** The sender's cancellation; it ends the sender's wait at once and stops a message that has not entered the inbox. */
readonly signal: AbortSignal
/**
* Turn the sender's content into durable message content, such as by
* storing attachments. It runs in the Agent's delivery order, after the
* earlier messages to the same Agent settled; its failure reaches the
* sender unchanged.
* @param signal - aborted when the sender cancels or the controller stops; controller disposal waits for the preparation to settle.
* @returns the content of the message.
*/
readonly prepare: (signal: AbortSignal) => Promise<UserMessage['content']>
/**
* Record what the sender needs before the Agent wakes. It runs
* synchronously after admission, immediately before the message enters the
* inbox. A throw keeps the message out of the inbox and reaches the sender
* unchanged; the sender reverts its own effects when `deliver` rejects.
* @param message - the admitted message.
*/
readonly beforeInsert?: (message: UserMessage) => void
}
```
A listener admits by returning the decision of `next()`, holds the message by awaiting before `next()`, and refuses it by returning a `reject` decision:
```ts type-equiv
/** The outcome of the admission stage for one message. */
type AdmissionDecision =
/** The message enters the inbox. */
| { readonly kind: 'admit' }
/** The message stays out of the inbox. */
| {
readonly kind: 'reject'
/** Code of the refusal, owned by its source, such as `forge/worktree-missing`. */
readonly reason: string
/** Explanation for a sender that shows nothing of the refusing plugin. */
readonly message: string
/** Whether the refusing plugin explains the refusal in its own interface. */
readonly presented: boolean
}
```
## Interception decisions
Pre-step decisions use the same identified `UserMessage` type as durable user-role input. The entered batch is authoritative and preserves every message's `id` and `source`. Hook bridges map their native decision fields onto this typed result.
@@ -539,33 +479,6 @@ interface TurnBoundaryProjection {
Generated from source by `scripts/gen-cordis-catalog.ts` (verified fresh by `pnpm run verify-cordis-catalog` in doc-sync; regenerate with `pnpm run gen-cordis-catalog`) — the language sides differ only in locale-specific paired document paths. Signature blocks use a `ts cordis-catalog` fence and keep the original source JSDoc; dispatch modes are defined in the [primer](../cordis-primer.md#dispatch-modes), and the framework-inherited `ctx` API lives in [cordis-api/inherited.md](../cordis-api/inherited.md).
<a id="ctxagentcontroller--agentcontroller"></a>
### `ctx.agentController` — `AgentController`
Delivers the messages that senders, such as the Web Session Remote, bring on someone's behalf. Messages to one Agent are prepared, decided, and appended one at a time in arrival order; messages to different Agents do not wait for each other. Disposal stops every message that has not entered an inbox and finishes once each delivery it stopped has settled.
```ts cordis-catalog
/**
* Deliver one message from outside the Agent: prepare its content, ask the
* `agent-controller/admission` stage for a decision, then append an
* admitted message with `followup` or `steer`.
* @param agent - the receiving Agent.
* @param request - delivery mode, source, cancellation, and the sender's content preparation.
* @returns the id of the message once it is in the inbox.
* @throws {@link AgentAdmissionRejected} when a listener rejects the message
* or throws, or when the Agent left the registry first; the error of
* `prepare`, `beforeInsert`, `followup`, or `steer` unchanged; the sender's
* cancellation reason as soon as it cancels, also while the message waits
* behind another one; the stop reason once the controller stops.
*/
async deliver(agent: Agent, request: DeliveryRequest): Promise<MessageId>
```
Types: [MessageId](llm-streaming.md)
Source: [`packages/core/agent-controller/src/index.ts`](../../packages/core/agent-controller/src/index.ts)
<a id="ctxagentdefaultmodel--agentdefaultmodelconfig"></a>
### `ctx.agentDefaultModel` — `AgentDefaultModelConfig`
@@ -1238,43 +1151,6 @@ Types: [Scoped](scope.md)
Source: [`packages/core/agent/src/runtime-types.ts`](../../packages/core/agent/src/runtime-types.ts)
<a id="agent-controller-events"></a>
### `agent-controller/*` events
<a id="agent-controlleradmission--waterfall"></a>
#### `agent-controller/admission` — waterfall
Decide whether one message from outside an Agent enters its inbox. `ctx.agentController.deliver` runs this stage and appends the message only after the whole chain admitted it; messages the Agent's own machinery produces enter through `followup`, `steer`, or `inject` without it. Listeners run in registration order, and messages to one Agent are decided one at a time in arrival order. A listener admits by returning the decision of `next()`, whose last step admits; awaiting before `next()` holds the message, and returning a `reject` decision refuses it without asking later listeners. A thrown error means the listener's check itself failed. A listener must not wait for a person when the message source is unattended, such as a schedule or a webhook.
```ts cordis-catalog
/**
* Decide whether one message from outside an Agent enters its inbox.
* `ctx.agentController.deliver` runs this stage and appends the message
* only after the whole chain admitted it; messages the Agent's own
* machinery produces enter through `followup`, `steer`, or `inject`
* without it. Listeners run in registration order, and messages to one
* Agent are decided one at a time in arrival order. A listener admits by
* returning the decision of `next()`, whose last step admits; awaiting
* before `next()` holds the message, and returning a `reject` decision
* refuses it without asking later listeners. A thrown error means the
* listener's check itself failed. A listener must not wait for a person
* when the message source is unattended, such as a schedule or a webhook.
* @param payload.agent - the receiving Agent.
* @param payload.message - the message as it will enter the inbox.
* @param payload.signal - aborted when the sender cancels or the controller stops; a holding
* listener must settle once it aborts, because controller disposal waits for it.
* Scope-filtered dispatch (`@deepseek-ai/dsh-scope`): Agent-scoped listeners receive only that Agent.
* @mode waterfall
*/
'agent-controller/admission'(this: Scoped<Agent>, payload: { agent: Agent; message: UserMessage; signal: AbortSignal }, next: () => Promise<AdmissionDecision>): Promise<AdmissionDecision>
```
Types: [Scoped](scope.md) · [UserMessage](session.md)
Source: [`packages/core/agent-controller/src/types.ts`](../../packages/core/agent-controller/src/types.ts)
<a id="agent-loop-events"></a>
### `agent-loop/*` events
docs/subsystems/core.zh.md
+1 -127
View File
@@ -15,10 +15,9 @@
| `tools/` | 带作用域的工具注册表与受保护的执行流水线（`ctx.tools`） | [tools.md](tools.zh.md) |
| `agent/` | `Agent` 接口、实时注册表、发起者作用域与 `agent/*` 事件词汇（`ctx.agents`） | 本页 |
| `agent-loop/` | 实现公开 `Agent` 约定的具体 driver（`ctx.agentLoop`） | 本页 |
| `agent-controller/` | 外部消息经 `agent-controller/admission` 阶段投递进 Agent（`ctx.agentController`） | 本页 |
| `scope/` | 注册表与循环用于构建按 agent 作用域的注册原语 | [scope.md](scope.zh.md) |
`scope/` 是这里唯一的非服务包：一个零依赖库（`createScope`/`scopeOf`/`scopeTarget`），在模块图中位于 `session/` 与 `system-prompt/` 之下，正是为了让它们消费它而不形成环。`agent-loop` 是公开 `Agent` 约定的唯一具体实现，放在这里因为它是 harness 的默认产品循环；它在 `ctx.agents.withInitiator()` 内运行每个 driver。扩展插件依赖 `agent`——包括需要发起 Agent 时——而绝不直接依赖 `agent-loop`，因此循环保持可替换。`agent-controller` 位于循环之前：代表他人送来消息的发送方经 `ctx.agentController` 投递，而不是自己写入收件箱。[`dsh-base`](../../packages/bundle/base/README.zh.md) 是默认产品组合，[`dsh-sdk-minimal`](../../packages/bundle/sdk-minimal/README.zh.md) 则声明一棵更小的独立配置树。
`scope/` 是这里唯一的非服务包：一个零依赖库（`createScope`/`scopeOf`/`scopeTarget`），在模块图中位于 `session/` 与 `system-prompt/` 之下，正是为了让它们消费它而不形成环。`agent-loop` 是公开 `Agent` 约定的唯一具体实现，放在这里因为它是 harness 的默认产品循环；它在 `ctx.agents.withInitiator()` 内运行每个 driver。扩展插件依赖 `agent`——包括需要发起 Agent 时——而绝不直接依赖 `agent-loop`，因此循环保持可替换。[`dsh-base`](../../packages/bundle/base/README.zh.md) 是默认产品组合，[`dsh-sdk-minimal`](../../packages/bundle/sdk-minimal/README.zh.md) 则声明一棵更小的独立配置树。
<a id="creation-and-ownership"></a>
@@ -354,67 +353,6 @@ cause 是由 TypeScript 强制约束的同进程输入。活跃的取消持有
`ctx.agents` 携带的进程本地 initiator 就是上面的确切 `Agent`，不是单独的 frame 或复制的标识。环境中存在该值既不能证明存活，也不代表授权；[initiator 作用域决策](../../.agents/notes/implemented/architecture/2026-07-15-agent-initiator-scope.zh.md)定义其生命周期和作用域规则。
<a id="messages-from-outside-an-agent"></a>
## 来自 Agent 外部的消息
代表他人送来消息的发送方（例如 Web Session Remote）调用 [`agent-controller`](../../packages/core/agent-controller/README.zh.md) 的 `ctx.agentController.deliver`，而不是 `followup()` 或 `steer()`。控制器按该 Agent 的投递顺序准备内容，向 `agent-controller/admission` waterfall 请求决定，再放入被接纳的消息；Agent 自身机制产生的消息不经过这个阶段。[投递决策](../../.agents/notes/implemented/architecture/2026-10-17-agent-controller-message-delivery.zh.md)说明了这个阶段为什么不放在 `agent` 里。
源码：[`packages/core/agent-controller/src/types.ts`](../../packages/core/agent-controller/src/types.ts)
```ts type-equiv
/** How a delivered message enters the inbox: `followup` for its own turn, `steer` for the nearest step. */
type DeliveryMode = 'followup' | 'steer'
```
```ts type-equiv
/** One message a sender delivers on someone's behalf. */
interface DeliveryRequest {
/** Where the message enters once it is admitted. */
readonly mode: DeliveryMode
/** Durable attribution recorded on the message. */
readonly source: MessageSource
/** The sender's cancellation; it ends the sender's wait at once and stops a message that has not entered the inbox. */
readonly signal: AbortSignal
/**
* Turn the sender's content into durable message content, such as by
* storing attachments. It runs in the Agent's delivery order, after the
* earlier messages to the same Agent settled; its failure reaches the
* sender unchanged.
* @param signal - aborted when the sender cancels or the controller stops; controller disposal waits for the preparation to settle.
* @returns the content of the message.
*/
readonly prepare: (signal: AbortSignal) => Promise<UserMessage['content']>
/**
* Record what the sender needs before the Agent wakes. It runs
* synchronously after admission, immediately before the message enters the
* inbox. A throw keeps the message out of the inbox and reaches the sender
* unchanged; the sender reverts its own effects when `deliver` rejects.
* @param message - the admitted message.
*/
readonly beforeInsert?: (message: UserMessage) => void
}
```
监听器返回 `next()` 的决定即放行，在 `next()` 之前 await 即暂留消息，返回 `reject` 决定即拒绝：
```ts type-equiv
/** The outcome of the admission stage for one message. */
type AdmissionDecision =
/** The message enters the inbox. */
| { readonly kind: 'admit' }
/** The message stays out of the inbox. */
| {
readonly kind: 'reject'
/** Code of the refusal, owned by its source, such as `forge/worktree-missing`. */
readonly reason: string
/** Explanation for a sender that shows nothing of the refusing plugin. */
readonly message: string
/** Whether the refusing plugin explains the refusal in its own interface. */
readonly presented: boolean
}
```
<a id="interception-decisions"></a>
## 拦截决策
@@ -553,33 +491,6 @@ interface TurnBoundaryProjection {
Generated from source by `scripts/gen-cordis-catalog.ts` (verified fresh by `pnpm run verify-cordis-catalog` in doc-sync; regenerate with `pnpm run gen-cordis-catalog`) — the language sides differ only in locale-specific paired document paths. Signature blocks use a `ts cordis-catalog` fence and keep the original source JSDoc; dispatch modes are defined in the [primer](../cordis-primer.zh.md#dispatch-modes), and the framework-inherited `ctx` API lives in [cordis-api/inherited.md](../cordis-api/inherited.md).
<a id="ctxagentcontroller--agentcontroller"></a>
### `ctx.agentController` — `AgentController`
Delivers the messages that senders, such as the Web Session Remote, bring on someone's behalf. Messages to one Agent are prepared, decided, and appended one at a time in arrival order; messages to different Agents do not wait for each other. Disposal stops every message that has not entered an inbox and finishes once each delivery it stopped has settled.
```ts cordis-catalog
/**
* Deliver one message from outside the Agent: prepare its content, ask the
* `agent-controller/admission` stage for a decision, then append an
* admitted message with `followup` or `steer`.
* @param agent - the receiving Agent.
* @param request - delivery mode, source, cancellation, and the sender's content preparation.
* @returns the id of the message once it is in the inbox.
* @throws {@link AgentAdmissionRejected} when a listener rejects the message
* or throws, or when the Agent left the registry first; the error of
* `prepare`, `beforeInsert`, `followup`, or `steer` unchanged; the sender's
* cancellation reason as soon as it cancels, also while the message waits
* behind another one; the stop reason once the controller stops.
*/
async deliver(agent: Agent, request: DeliveryRequest): Promise<MessageId>
```
Types: [MessageId](llm-streaming.zh.md)
Source: [`packages/core/agent-controller/src/index.ts`](../../packages/core/agent-controller/src/index.ts)
<a id="ctxagentdefaultmodel--agentdefaultmodelconfig"></a>
### `ctx.agentDefaultModel` — `AgentDefaultModelConfig`
@@ -1252,43 +1163,6 @@ Types: [Scoped](scope.zh.md)
Source: [`packages/core/agent/src/runtime-types.ts`](../../packages/core/agent/src/runtime-types.ts)
<a id="agent-controller-events"></a>
### `agent-controller/*` events
<a id="agent-controlleradmission--waterfall"></a>
#### `agent-controller/admission` — waterfall
Decide whether one message from outside an Agent enters its inbox. `ctx.agentController.deliver` runs this stage and appends the message only after the whole chain admitted it; messages the Agent's own machinery produces enter through `followup`, `steer`, or `inject` without it. Listeners run in registration order, and messages to one Agent are decided one at a time in arrival order. A listener admits by returning the decision of `next()`, whose last step admits; awaiting before `next()` holds the message, and returning a `reject` decision refuses it without asking later listeners. A thrown error means the listener's check itself failed. A listener must not wait for a person when the message source is unattended, such as a schedule or a webhook.
```ts cordis-catalog
/**
* Decide whether one message from outside an Agent enters its inbox.
* `ctx.agentController.deliver` runs this stage and appends the message
* only after the whole chain admitted it; messages the Agent's own
* machinery produces enter through `followup`, `steer`, or `inject`
* without it. Listeners run in registration order, and messages to one
* Agent are decided one at a time in arrival order. A listener admits by
* returning the decision of `next()`, whose last step admits; awaiting
* before `next()` holds the message, and returning a `reject` decision
* refuses it without asking later listeners. A thrown error means the
* listener's check itself failed. A listener must not wait for a person
* when the message source is unattended, such as a schedule or a webhook.
* @param payload.agent - the receiving Agent.
* @param payload.message - the message as it will enter the inbox.
* @param payload.signal - aborted when the sender cancels or the controller stops; a holding
* listener must settle once it aborts, because controller disposal waits for it.
* Scope-filtered dispatch (`@deepseek-ai/dsh-scope`): Agent-scoped listeners receive only that Agent.
* @mode waterfall
*/
'agent-controller/admission'(this: Scoped<Agent>, payload: { agent: Agent; message: UserMessage; signal: AbortSignal }, next: () => Promise<AdmissionDecision>): Promise<AdmissionDecision>
```
Types: [Scoped](scope.zh.md) · [UserMessage](session.zh.md)
Source: [`packages/core/agent-controller/src/types.ts`](../../packages/core/agent-controller/src/types.ts)
<a id="agent-loop-events"></a>
### `agent-loop/*` events
docs/subsystems/session.md
+2 -6
View File
@@ -930,14 +930,10 @@ workspaceDesktop(): { name: string; available: boolean; fileManager: 'finder' |
@Remote('fork') async fork(request: SessionForkRequest): Promise<SessionForkValue>
/**
* Admit one prompt after explicitly resuming its Session. The prompt is
* delivered through `ctx.agentController`, whose `agent-controller/admission`
* stage runs before it enters the inbox, so the call can wait for a listener
* that holds it.
* Admit one prompt after explicitly resuming its Session.
* @param request - Session identity, prompt content, source metadata, and delivery mode.
* @param signal - caller cancellation; it also stops a prompt waiting for or held at the admission stage.
* @param signal - caller cancellation before prompt admission begins.
* @returns acknowledgement that the Agent accepted the prompt.
* @throws `session/prompt-rejected` when an admission listener refuses the prompt or fails.
*/
@Remote('prompt') async prompt(request: SessionPromptRequest, signal: AbortSignal): Promise<SessionPromptValue>
docs/subsystems/session.zh.md
+2 -6
View File
@@ -934,14 +934,10 @@ workspaceDesktop(): { name: string; available: boolean; fileManager: 'finder' |
@Remote('fork') async fork(request: SessionForkRequest): Promise<SessionForkValue>
/**
* Admit one prompt after explicitly resuming its Session. The prompt is
* delivered through `ctx.agentController`, whose `agent-controller/admission`
* stage runs before it enters the inbox, so the call can wait for a listener
* that holds it.
* Admit one prompt after explicitly resuming its Session.
* @param request - Session identity, prompt content, source metadata, and delivery mode.
* @param signal - caller cancellation; it also stops a prompt waiting for or held at the admission stage.
* @param signal - caller cancellation before prompt admission begins.
* @returns acknowledgement that the Agent accepted the prompt.
* @throws `session/prompt-rejected` when an admission listener refuses the prompt or fails.
*/
@Remote('prompt') async prompt(request: SessionPromptRequest, signal: AbortSignal): Promise<SessionPromptValue>
docs/web-styling.i18n.yaml
+5 -2
View File
@@ -9,8 +9,11 @@
en: dc8c41fbf9869179
zh: e45976a585cb9c86
/web-ui-style-reference/component-rules:
en: 6715806a6f84d141
zh: 94b7a6ede4919512
en: 4ee4a6fd2917306a
zh: d7d68c140a698ab0
/web-ui-style-reference/input-capabilities:
en: bf2e98bce68743e2
zh: fc58dcbb60bb944a
/web-ui-style-reference/corner-radii-and-settings-cards:
en: 118972f6c2bd6ce4
zh: 85f8416b03e4dd8b
docs/web-styling.md
+12 -1
View File
@@ -21,7 +21,7 @@ Global style sheets belong in `ui-theme/src/styles/`. Component styles live besi
- Keep source text, terminal output, and diff lines unwrapped when their component contract requires column preservation; use the shared scrollbar styles rather than component-specific scrollbar selectors.
- Put presentation in CSS. Inline React styles may pass component-local custom-property values but must not encode theme branches.
- A relative `url()` in a plugin stylesheet points to a file in the package's `src/client/assets/`, and the build rewrites it to the [package asset](../packages/client/AGENTS.md#dependency-declaration) URL. `data:`, `#fragment`, `/`-rooted, scheme, and empty `url('')` references stay unchanged, a relative file reference with a fragment or query fails the build, and stylesheets from dependencies are not rewritten.
- Preserve keyboard focus visibility and reduced-motion behavior when adding transitions or hover-only controls.
- Put `:hover` selectors under `@media (hover: hover)`, independently of viewport width. Keep selected, expanded, pressed, and keyboard-focus styles outside that condition. Without hover, actions remain visible or accessible through an explicit control; pointer handlers ignore touch when opening hover-only previews. Preserve keyboard focus visibility and reduced-motion behavior.
- Rounded corners inherit the global superellipse smoothing from ui-theme's `corner-shape.css` on supporting engines. Pair `corner-shape: round` with every full-round `border-radius` (`50%`, `100%`, or a pill radius) so circles and capsules keep circular arcs; the ui-theme corner-shape spec enforces the pairing.
- Elevated surfaces (menus, popovers, modals, panels, floating buttons, the composer) set `border: 0` and take `box-shadow: var(--dsw-elevation-panel)`, `var(--dsw-elevation-prominent)`, or the composer's `var(--dsw-elevation-soft)` (larger blur at lower alpha): the 0.5px hairline stroke is the first shadow layer, and `--dsw-elevation-stroke-color` rebinds or suppresses it per surface or state. Never pair a `--dsw-alias-border-*` border with an lv/elevation shadow — the ui-theme elevation spec rejects the pairing; state-colored borders (warn panels) stay real borders.
- Existing and new dropdown, context, submenu, and selection menus use `Menu` or wrap custom content in `MenuSurface`. The material is the theme-owned `--dsw-menu-surface-fill` plus `--dsw-menu-backdrop-filter`; feature and platform CSS must not override either. The ui-theme menu spec rejects unwrapped menu/listbox renderers, material overrides, and token redefinitions. The schedule-owned `TaskMenu` and `ClockPicker` retain their existing containers and material as explicit exceptions ([owner](../packages/client/ui-schedule/README.md)).
@@ -30,6 +30,17 @@ Global style sheets belong in `ui-theme/src/styles/`. Component styles live besi
- Flat borders and separators that use a neutral `--dsw-alias-border-*` token draw at `0.5px` — buttons, inputs, cards, row dividers, and separators drawn as filled boxes (menu separators, the conversation header seam, markdown `hr`, vertical rails) share the hairline weight, which Chromium paints as one device pixel. Dashed affordances and state-colored borders keep 1px; spinner ring tracks keep their width through the spec's explicit allowlist. The ui-theme elevation spec rejects wider neutral solid borders.
- Clickable artifact links (markdown anchors, prose file mentions, web source and fetch links, produced-file chips, workflow member links) color through `--dsw-alias-link` at `font-weight: 500`, with no underline at rest and a dotted 3px-offset underline on hover/focus. Compact Thinking Markdown keeps tertiary text color and a resting dotted underline ([compact presentation](../packages/client/ui-primitives/README.md#rendering-agent-output)). Text-leading anchors also lead with the ui-primitives `LinkIconMedium` category glyph riding `currentColor`, which for a well-known external host is that site's own mark instead of the globe; workflow member links and image-only anchors carry no glyph, and tool-row file links keep their grey dotted affordance.
<a id="input-capabilities"></a>
## Input capabilities
`pointer` and `hover` describe the primary pointing device chosen by the browser; `any-hover` considers every available pointing device. None detects a physical keyboard.
- Use `(pointer: coarse)` for presentation that depends on pointing accuracy, such as touch-sized controls and text fields.
- Keep CSS hover effects under `(hover: hover)` so a touch-first layout remains usable without hover.
- JavaScript hover previews use `(any-hover: hover)` when accepting mouse entry and ignore touch entry. A secondary mouse can therefore open a preview even when the touchscreen remains the primary pointer. Enabling all CSS hover styles through `any-hover` would also enable them for touch-generated `:hover` states.
On a mixed-input device, a mouse preview can appear while the primary-pointer CSS hover style stays inactive. Keyboard focus and shortcut hints follow actual keyboard interaction independently of these capability queries; [Menu and input modality](../packages/client/ui-primitives/README.md#input-modality) define the shared behavior.
<a id="corner-radii-and-settings-cards"></a>
## Corner radii and settings cards
docs/web-styling.zh.md
+12 -1
View File
@@ -21,7 +21,7 @@
- 当组件约定要求保留列结构时，源码文本、终端输出和 diff 行不得换行；使用共享滚动条样式，不得定义组件专用滚动条选择器。
- 呈现规则写在 CSS 中。React 内联样式可以传递组件局部自定义属性值，但不得编码主题分支。
- 插件样式表中的相对 `url()` 指向本包 `src/client/assets/` 中的文件，构建会将其改写为[包资源](../packages/client/AGENTS.md#dependency-declaration) URL。`data:`、`#fragment`、以 `/` 开头、带 scheme 的引用以及空 `url('')` 引用保持不变，带片段或查询串的相对文件引用会让构建失败，依赖包中的样式表不会被改写。
- 添加过渡动画或仅悬停可见的控件时，保留清晰可见的键盘焦点和减少动态效果行为。
- 将 `:hover` 选择器放在 `@media (hover: hover)` 内，不依赖视口宽度。选中、展开、按下和键盘焦点样式保留在该条件之外。无 hover 时，操作保持可见或通过明确控件访问；打开纯悬停预览的指针事件处理忽略触摸。保留清晰可见的键盘焦点和减少动态效果行为。
- 支持的引擎上，圆角继承 ui-theme `corner-shape.css` 的全局超级椭圆平滑。每个正圆 `border-radius`（`50%`、`100%` 或胶囊半径）必须配对 `corner-shape: round`，使圆形与胶囊保持圆弧；ui-theme 的 corner-shape spec 强制这一配对。
- 高层级表面（菜单、浮层、对话框、面板、悬浮按钮、输入框）设 `border: 0` 并使用 `box-shadow: var(--dsw-elevation-panel)`、`var(--dsw-elevation-prominent)` 或输入框专用的 `var(--dsw-elevation-soft)`（更大模糊、更低透明度）：0.5px 发丝描边是第一层投影，`--dsw-elevation-stroke-color` 可按表面或状态重绑或抑制描边。不得将 `--dsw-alias-border-*` border 与 lv/elevation 投影配对——ui-theme 的 elevation spec 会拒绝；状态色 border（warn 面板）保持真 border。
- 存量和新增的下拉菜单、上下文菜单、子菜单及选择菜单使用 `Menu`，自定义内容用 `MenuSurface` 包裹。材质由主题的 `--dsw-menu-surface-fill` 和 `--dsw-menu-backdrop-filter` 组成；功能及平台 CSS 不得覆盖两者。ui-theme 菜单检查拒绝未使用共享容器的 menu/listbox 渲染、材质覆盖和 token 重定义。计划任务自有的 `TaskMenu` 与 `ClockPicker` 是明确例外，保留现有容器和材质（[所属包](../packages/client/ui-schedule/README.zh.md)）。
@@ -30,6 +30,17 @@
- 使用中性 `--dsw-alias-border-*` token 的平面边框与分割线一律 `0.5px`——按钮、输入框、卡片、行分割线，以及以填充盒绘制的分隔线（菜单分隔、对话标题栏接缝、markdown `hr`、竖向轨道线）共用发丝线粗细，Chromium 将其绘制为一个设备像素。dashed 记号与状态色 border 保持 1px；spinner 圆环经 spec 的显式豁免保留原宽度。更宽的中性 solid border 会被 ui-theme elevation spec 拒绝。
- 可点击产物链接（Markdown 锚点、正文文件引用、网页来源与抓取链接、产物 chips、workflow 成员链接）经 `--dsw-alias-link` 着色、`font-weight: 500`，默认无下划线，hover/focus 时为 3px offset 的点状下划线。紧凑 Thinking Markdown 保持 tertiary 文字色和默认点状下划线（[紧凑展示](../packages/client/ui-primitives/README.zh.md#rendering-agent-output)）。带文字的锚点另以 ui-primitives 的 `LinkIconMedium` 分类图形（随 `currentColor`）作前置；目的地是已知外部站点时改用该站点自己的标记而非地球；workflow 成员链接与只包图片的锚点不带图形，工具行文件链接保持其灰色点线示能。
<a id="input-capabilities"></a>
## 输入能力
`pointer` 与 `hover` 描述浏览器选定的主指针，`any-hover` 则考虑所有可用指针。三者都不检测物理键盘。
- `(pointer: coarse)` 用于依赖指针精度的呈现，例如适合触摸的控件和文本框尺寸。
- CSS 悬停效果保留在 `(hover: hover)` 下，使以触屏为主的布局无需悬停也能使用。
- JavaScript 悬停预览在接受 mouse 进入事件时使用 `(any-hover: hover)`，并忽略 touch 进入事件。因此，即使触屏仍是主指针，外接鼠标也能打开预览。如果通过 `any-hover` 启用全部 CSS 悬停样式，触摸产生的 `:hover` 状态也会应用这些样式。
混合输入设备上，鼠标预览可以出现，而基于主指针的 CSS 悬停样式保持关闭。键盘焦点和快捷键提示根据实际键盘交互显示，不依赖这些能力查询；共享行为见 [Menu 与输入方式](../packages/client/ui-primitives/README.zh.md#input-modality)。
<a id="corner-radii-and-settings-cards"></a>
## 圆角与设置卡片
packages/api/session-controller/README.i18n.yaml
+2 -2
View File
@@ -15,8 +15,8 @@
en: 1fdf707322fb6996
zh: 0204f7e9d4aa3cce
/session-controller/use-this-package:
en: fe242dda6579245e
zh: cbd8c012135623de
en: 8fcfa14545bd9d1c
zh: c10c2bba792699ef
/session-controller/client-references:
en: 41bdda1ac32df558
zh: aa5d0e325924066b
packages/api/session-controller/README.md
+1 -1
View File
@@ -33,7 +33,7 @@ History pages and follow opening snapshots carry one `{ type: 'event', event: Se
The Client journal validates current Session event envelopes before publishing follow snapshots, live entries, or history pages. It reuses the browser-safe Session validators for required surface markers, exact replacement endpoints, earlier unique source seqs, embedded Assistant provider metadata, request-header omissions, and tool-error consistency. Invalid records fail without field stripping or normalization; range membership and source existence remain durable-log checks on the Host.
Each endpoint states its activation policy. List reads only stored headers and projection-cache rows: it never calls per-session stat or opens a cold Session body. A current-format cache identity may supply every list hint; a lifecycle-matching predecessor cache may supply only its version-compatible title as a stale display fact, never as an authoritative fold seed. Search, attachment, history pages, log following, skill discovery, and workspace-path opening can inspect persistence without activating an Agent; `canOpenWorkspacePath()` reports native-opening availability without addressing a Session. Cancellation requires live state; queue mutation, model, rename, prompt, and file-reference operations may resolve or resume an ordinary Session. Prompt rejects content with neither non-whitespace text nor an attachment before resolving the Agent or appending Session events; queue edits accept only non-empty text content. Prompt admission consumes opaque receipts from the injected [`fileUploads`](../../client/file-upload/README.md) Host service and resolves every same-Agent receipt before sending the complete ordered content list through `ctx.attachments`. Prompt retries whose `requestId` is already queued or logged return the original acceptance without inserting another message. A prompt enters the Agent through [`ctx.agentController.deliver`](../../core/agent-controller/README.md#use-this-package), which runs the receipt resolution and attachment storage in the Agent's delivery order and binds the receipts immediately before the message enters the inbox, so the call stays open while an `agent-controller/admission` listener holds it. An image prompt takes the Session's image admission turn when its preparation starts and keeps it until the prompt settles, so a model selection lands either before its image check or after it entered the inbox, and waits while a listener holds that image prompt. A refusal returns `session/prompt-rejected` with the listener's `reason` and `presented`, a listener failure returns the same code with reason `agent-controller/admission-failed`, and an Agent disposed meanwhile returns `session/not-found`. A client that predates `session/prompt-rejected` shows it as an ordinary failure with its message and code. Create and fork are the only operations that create a new Agent directly. The service applies one preset-aware resume policy and subagent ownership fence to its own methods and to the Typert Agent and Session lookups used by other Remote namespaces. Queue mutation has one narrow exception: a live child whose current projected identity is continuable and comes from its own non-seed suffix accepts the ordinary Edit, Remove, and QueueDock Steer actions across both inbox destinations. One-shot, missing, unknown, corrupt, seed-only, or cold children remain rejected without resume. The skill catalog uses a live Agent's effective working directory and scoped registry when present. Cold listing uses the projected current directory, falling back to the recorded origin and then the configured directory default. It retains the preset standing scope without activating an Agent. The authenticated delivery routes use `workspaceDesktop()` for the serving Host name and file-manager behavior. `openWorkspacePath({ path, action: "reveal" })` delegates file-manager navigation to the native adapter; omitting `action` opens the file-type association, including HTML and SVG. Both operations require the composed filesystem to map the requested Host path to the same canonical process path; unmapped remote paths are refused before a native command runs. `session.projections` normally reads an exact baseline, including current-format cold logs, and returns `kind: sequenced`. Only a read requiring migration returns `kind: migration-required` hints without `asOfSeq`; missing keys remain absent. It returns null for a missing Session. Fork callers may set `allowMigration: false` to receive `session/migration-required` instead of starting source migration; omission retains automatic migration. The Client exposes shared values and explicit-read state through `projectionsBySession`; domains select their own keys. Session-list summaries carry `agentAvailable`, updated through existing summary and status events independently of durable projections. Sequenced reads and live projection frames use the same sequence ordering; cached reads fill hints without replacing sequenced values.
Each endpoint states its activation policy. List reads only stored headers and projection-cache rows: it never calls per-session stat or opens a cold Session body. A current-format cache identity may supply every list hint; a lifecycle-matching predecessor cache may supply only its version-compatible title as a stale display fact, never as an authoritative fold seed. Search, attachment, history pages, log following, skill discovery, and workspace-path opening can inspect persistence without activating an Agent; `canOpenWorkspacePath()` reports native-opening availability without addressing a Session. Cancellation requires live state; queue mutation, model, rename, prompt, and file-reference operations may resolve or resume an ordinary Session. Prompt rejects content with neither non-whitespace text nor an attachment before resolving the Agent or appending Session events; queue edits accept only non-empty text content. Prompt admission consumes opaque receipts from the injected [`fileUploads`](../../client/file-upload/README.md) Host service and resolves every same-Agent receipt before sending the complete ordered content list through `ctx.attachments`. Prompt retries whose `requestId` is already queued or logged return the original acceptance without inserting another message. Create and fork are the only operations that create a new Agent directly. The service applies one preset-aware resume policy and subagent ownership fence to its own methods and to the Typert Agent and Session lookups used by other Remote namespaces. Queue mutation has one narrow exception: a live child whose current projected identity is continuable and comes from its own non-seed suffix accepts the ordinary Edit, Remove, and QueueDock Steer actions across both inbox destinations. One-shot, missing, unknown, corrupt, seed-only, or cold children remain rejected without resume. The skill catalog uses a live Agent's effective working directory and scoped registry when present. Cold listing uses the projected current directory, falling back to the recorded origin and then the configured directory default. It retains the preset standing scope without activating an Agent. The authenticated delivery routes use `workspaceDesktop()` for the serving Host name and file-manager behavior. `openWorkspacePath({ path, action: "reveal" })` delegates file-manager navigation to the native adapter; omitting `action` opens the file-type association, including HTML and SVG. Both operations require the composed filesystem to map the requested Host path to the same canonical process path; unmapped remote paths are refused before a native command runs. `session.projections` normally reads an exact baseline, including current-format cold logs, and returns `kind: sequenced`. Only a read requiring migration returns `kind: migration-required` hints without `asOfSeq`; missing keys remain absent. It returns null for a missing Session. Fork callers may set `allowMigration: false` to receive `session/migration-required` instead of starting source migration; omission retains automatic migration. The Client exposes shared values and explicit-read state through `projectionsBySession`; domains select their own keys. Session-list summaries carry `agentAvailable`, updated through existing summary and status events independently of durable projections. Sequenced reads and live projection frames use the same sequence ordering; cached reads fill hints without replacing sequenced values.
Every Remote endpoint is access-controlled: a Peer whose principal is not the owner reaches an endpoint only when, on every resource the call touches, a binding of that principal kind's policy in [`dsh-api-remote-access`](../remote-access/README.md) grants a role listing the action ([Peer principals Agent Note](../../../.agents/notes/implemented/architecture/2026-10-12-peer-principal-and-remote-access.md)). The controller injects `remoteAccess` and registers the `session` resource kind with the default field `sessionId` through `typert.resources`. `session/page` and `session/follow` declare a `session` reader in `@Remote({ resources })` that reads `request.address.sessionId` for an ordinary Session address and `request.address.childSessionId` for a subagent address; whether access to the parent Session covers its children is the policy's rule. `session/projections` reads the Session resource from `request.sessionId`, so a reconnecting reader reloads only the projections of Sessions it may read. `session/list`, `session/search`, and `session/control` read no resource, so their calls touch the `global` resource and need a `global` role; for a non-owner caller, `session/list` returns only the rows, `session/search` only the hits, and `session/control` only the baseline projections and projection frames, whose Session `ctx.remoteAccess.allows()` admits for the same action, so that action also belongs to a `session` role; owner results are unfiltered. `session/prompt` builds the durable `user-rpc` message source in its Remote method and records `principal` only for a non-owner caller, read through `principalOf(this.ctx.invocation)`; a source without `principal` belongs to the owner.
packages/api/session-controller/README.zh.md
+1 -1
View File
@@ -33,7 +33,7 @@ kind: "package-reference"
Client journal 在发布 follow 快照、live entry 或历史页之前验证当前 Session 事件 envelope。它复用浏览器安全的 Session validator，检查必需的 surface marker、精确的 replacement endpoint、更早且唯一的 source seq、内嵌 Assistant 提供方元数据、request header 可选字段的省略规则以及工具错误一致性。无效 record 直接失败，不删除字段或归一化；范围成员与来源存在性仍由 Host 的持久日志检查。
每个 endpoint 都声明自己的激活策略。列表只读取持久化 header 与 projection cache row，绝不调用逐 Session stat 或打开冷 Session body。当前格式 cache identity 可以提供全部列表 hint；生命周期匹配的 predecessor cache 只能提供版本兼容的 title，作为可能过时的展示事实，绝不能作为权威 fold seed。搜索、附件、历史页、日志跟随、skill 发现和工作区路径打开可以在不激活 Agent 的情况下检查 persistence；`canOpenWorkspacePath()` 无需指定 Session 即可报告原生打开能力。取消要求 live 状态；queue 变更、模型、重命名、prompt 和文件引用操作可以解析或恢复普通 Session。提示词会在解析 Agent 或追加 Session 事件前，拒绝既没有非空白文本也没有附件的 content；queue edit 只接受非空文本 content。prompt 准入从注入的 [`fileUploads`](../../client/file-upload/README.zh.md) Host 服务取得不透明凭证，在把完整有序内容列表交给 `ctx.attachments` 前解析每个属于同一 Agent 的凭证。`requestId` 已进入 queue 或日志时，prompt 重试直接返回原来的接受结果，不会重复插入消息。提示词经 [`ctx.agentController.deliver`](../../core/agent-controller/README.zh.md#use-this-package) 进入 Agent：控制器按该 Agent 的投递顺序完成凭证解析与附件存储，并在消息进入收件箱前一刻绑定凭证，因此 `agent-controller/admission` 监听器暂留它期间调用保持未完成。图片提示词在开始准备时取得该 Session 的图片准入顺位，并持有到提示词结束，因此模型切换要么发生在它的图片检查之前，要么发生在它进入收件箱之后；监听器暂留这条图片提示词期间，模型切换会等待。被拒绝时返回 `session/prompt-rejected`，并带上监听器给出的 `reason` 与 `presented`；监听器失败时返回同一错误码，reason 为 `agent-controller/admission-failed`；期间 Agent 被销毁则返回 `session/not-found`。早于 `session/prompt-rejected` 的旧客户端把它显示为普通失败，带消息与错误码。只有 create 与 fork 会直接创建新 Agent。该服务把同一套感知 preset 的恢复策略和 subagent ownership fence 同时用于自身方法，以及其他 Remote namespace 使用的 Typert Agent 与 Session lookup。Queue 变更只有一个狭窄例外：当前 projection identity 为 continuable 且来自自身非 seed suffix 的在线 child，可以在两个 inbox 目标上使用普通 Edit、Remove 与 QueueDock Steer action。One-shot、缺失、未知、损坏、仅含 seed identity 或冷 child 继续被拒绝，且不会恢复。skill 目录优先使用已有 live Agent 的有效工作目录与限定范围内的注册表。冷态列表使用投影中的当前目录，依次回退到记录的起始目录与配置的默认目录。它保留 preset 常驻 scope，不会激活 Agent。经过鉴权的文件交付路由通过 `workspaceDesktop()` 获取提供服务的 Host 名称和文件管理器行为。`openWorkspacePath({ path, action: "reveal" })` 将文件管理器导航委托给原生适配器；省略 `action` 时按文件类型关联打开，包括 HTML 和 SVG。两种操作都要求当前文件系统将请求的 Host 路径映射到同一个规范进程路径；无法映射的远端路径会在执行原生命令之前被拒绝。 `session.projections` 正常读取准确基线，包括当前格式的冷日志，返回 `kind: sequenced`。只有需要迁移时才返回不带 `asOfSeq` 的 `kind: migration-required` 提示，缺少的键保持缺席。Session 不存在时返回 null。Fork 调用者可设置 `allowMigration: false`，在源历史需要迁移时收到 `session/migration-required`；省略则保持自动迁移。Client 通过 `projectionsBySession` 暴露共享值和显式读取状态，由领域选择自身的 key。Session 列表摘要携带 `agentAvailable`，通过已有摘要与状态事件更新，与持久化 projection 相互独立。有序读取与实时 projection 帧使用相同的序号排序规则；缓存读取只补充提示，不替换有序值。
每个 endpoint 都声明自己的激活策略。列表只读取持久化 header 与 projection cache row，绝不调用逐 Session stat 或打开冷 Session body。当前格式 cache identity 可以提供全部列表 hint；生命周期匹配的 predecessor cache 只能提供版本兼容的 title，作为可能过时的展示事实，绝不能作为权威 fold seed。搜索、附件、历史页、日志跟随、skill 发现和工作区路径打开可以在不激活 Agent 的情况下检查 persistence；`canOpenWorkspacePath()` 无需指定 Session 即可报告原生打开能力。取消要求 live 状态；queue 变更、模型、重命名、prompt 和文件引用操作可以解析或恢复普通 Session。提示词会在解析 Agent 或追加 Session 事件前，拒绝既没有非空白文本也没有附件的 content；queue edit 只接受非空文本 content。prompt 准入从注入的 [`fileUploads`](../../client/file-upload/README.zh.md) Host 服务取得不透明凭证，在把完整有序内容列表交给 `ctx.attachments` 前解析每个属于同一 Agent 的凭证。`requestId` 已进入 queue 或日志时，prompt 重试直接返回原来的接受结果，不会重复插入消息。只有 create 与 fork 会直接创建新 Agent。该服务把同一套感知 preset 的恢复策略和 subagent ownership fence 同时用于自身方法，以及其他 Remote namespace 使用的 Typert Agent 与 Session lookup。Queue 变更只有一个狭窄例外：当前 projection identity 为 continuable 且来自自身非 seed suffix 的在线 child，可以在两个 inbox 目标上使用普通 Edit、Remove 与 QueueDock Steer action。One-shot、缺失、未知、损坏、仅含 seed identity 或冷 child 继续被拒绝，且不会恢复。skill 目录优先使用已有 live Agent 的有效工作目录与限定范围内的注册表。冷态列表使用投影中的当前目录，依次回退到记录的起始目录与配置的默认目录。它保留 preset 常驻 scope，不会激活 Agent。经过鉴权的文件交付路由通过 `workspaceDesktop()` 获取提供服务的 Host 名称和文件管理器行为。`openWorkspacePath({ path, action: "reveal" })` 将文件管理器导航委托给原生适配器；省略 `action` 时按文件类型关联打开，包括 HTML 和 SVG。两种操作都要求当前文件系统将请求的 Host 路径映射到同一个规范进程路径；无法映射的远端路径会在执行原生命令之前被拒绝。 `session.projections` 正常读取准确基线，包括当前格式的冷日志，返回 `kind: sequenced`。只有需要迁移时才返回不带 `asOfSeq` 的 `kind: migration-required` 提示，缺少的键保持缺席。Session 不存在时返回 null。Fork 调用者可设置 `allowMigration: false`，在源历史需要迁移时收到 `session/migration-required`；省略则保持自动迁移。Client 通过 `projectionsBySession` 暴露共享值和显式读取状态，由领域选择自身的 key。Session 列表摘要携带 `agentAvailable`，通过已有摘要与状态事件更新，与持久化 projection 相互独立。有序读取与实时 projection 帧使用相同的序号排序规则；缓存读取只补充提示，不替换有序值。
每个 Remote 端点都受访问控制：principal 不是 owner 的 Peer 只有在本次调用访问的每个资源上，该 principal kind 在 [`dsh-api-remote-access`](../remote-access/README.zh.md) 中的 policy 都有某个 binding 授予列出该 action 的 role 时，才能调用该端点（[Peer 身份属性 Agent Note](../../../.agents/notes/implemented/architecture/2026-10-12-peer-principal-and-remote-access.zh.md)）。控制器注入 `remoteAccess`，并经 `typert.resources` 以默认字段 `sessionId` 登记 `session` 资源种类。`session/page` 与 `session/follow` 在 `@Remote({ resources })` 中声明 `session` 读取函数：对普通 Session 地址读 `request.address.sessionId`，对 subagent 地址读 `request.address.childSessionId`；父会话的访问权是否覆盖其子会话由 policy 决定。`session/projections` 从 `request.sessionId` 读取会话资源，因此重连的读者只会重新读取它可读的会话的 projections。`session/list`、`session/search` 与 `session/control` 不读取资源，其调用访问 `global` 资源，需要 `global` role；对非 owner 调用方，`session/list` 只返回其 Session 被 `ctx.remoteAccess.allows()` 按同一 action 允许的行，`session/search` 只返回这类 Session 的命中，`session/control` 也只返回这类 Session 的 baseline projection 与 projection 帧，因此该 action 也要列在 `session` role 中；owner 的结果不过滤。`session/prompt` 在其 Remote 方法内构造持久的 `user-rpc` 消息 source，且只为非 owner 调用方记录经 `principalOf(this.ctx.invocation)` 读取的 `principal`；不带 `principal` 的 source 属于 owner。
packages/api/session-controller/package.json
-2
View File
@@ -80,7 +80,6 @@
"peerDependencies": {
"@deepseek-ai/cordis": "workspace:~",
"@deepseek-ai/dsh-agent": "workspace:*",
"@deepseek-ai/dsh-agent-controller": "workspace:*",
"@deepseek-ai/dsh-agent-default-model": "workspace:*",
"@deepseek-ai/dsh-agent-preset-registry": "workspace:*",
"@deepseek-ai/dsh-api-gateway": "workspace:*",
@@ -127,7 +126,6 @@
"@deepseek-ai/dsh-api-remote-access": "workspace:*",
"@deepseek-ai/cordis": "workspace:~",
"@deepseek-ai/dsh-agent": "workspace:*",
"@deepseek-ai/dsh-agent-controller": "workspace:*",
"@deepseek-ai/dsh-agent-default-model": "workspace:*",
"@deepseek-ai/dsh-agent-loop": "workspace:*",
"@deepseek-ai/dsh-agent-loop-testkit": "workspace:*",
packages/api/session-controller/src/agent.ts
-21
View File
@@ -298,27 +298,6 @@ export class ApiSessionAgentController {
return result
}
/**
* Take the image admission turn for one Agent: wait until earlier image
* admissions and model selections settled, then hold the turn until the
* returned release runs.
* @param agent - live Agent that owns the serialization chain.
* @param signal - the waiting caller's cancellation; a caller cancelled by its turn passes the turn on.
* @returns the release, which the caller runs once its image prompt settled.
* @throws the cancellation reason when the caller was cancelled by its turn.
*/
acquireImageAdmission(agent: Agent, signal: AbortSignal): Promise<() => void> {
const previous = this.imageAdmissionChains.get(agent) ?? Promise.resolve()
const held = Promise.withResolvers<undefined>()
this.imageAdmissionChains.set(agent, previous.then(() => held.promise))
const release = (): void => { held.resolve(undefined) }
return previous.then(() => {
if (signal.aborted) release()
signal.throwIfAborted()
return release
})
}
/**
* Resolve the preset id and pre-publication Agent setup for a create or resume.
* @param presetId - requested preset or the configured default when omitted.
packages/api/session-controller/src/commands.ts
+44 -90
View File
@@ -5,15 +5,14 @@ import { randomUUID } from 'node:crypto'
import type { Context } from '@deepseek-ai/cordis'
import { brandString } from '@deepseek-ai/dsh-brand'
import type { Agent, ModelSelection as AgentModelSelection } from '@deepseek-ai/dsh-agent'
import type { ADMISSION_AGENT_GONE, ADMISSION_REJECTED_CODE, AgentAdmissionRejected } from '@deepseek-ai/dsh-agent-controller'
import { AttachmentError } from '@deepseek-ai/dsh-attachment'
import type {
AttachmentAdmissionPart, FileAttachmentRef, ImageAttachmentRef,
} from '@deepseek-ai/dsh-attachment'
import type { FileUploadReceiptId } from '@deepseek-ai/dsh-client-file-upload/types'
import type { PromptFileBinding } from '@deepseek-ai/dsh-client-file-upload'
import type {} from '@deepseek-ai/dsh-client-file-upload'
import {
ReasoningEffortId, assistantStreamChunks, freezeMessage,
ReasoningEffortId, assistantStreamChunks, createUserMessage, freezeMessage,
} from '@deepseek-ai/dsh-llm'
import type { MessageSource } from '@deepseek-ai/dsh-llm'
import { buildForkSeed } from '@deepseek-ai/dsh-session/fork'
@@ -311,73 +310,58 @@ export class SessionCommandController {
}
/**
* Deliver one prompt through the Agent controller, whose admission stage
* may hold or refuse it; the Remote method has already validated the
* content and built the durable message source. The controller runs this
* Session's content preparation (image model check, upload receipts,
* attachment storage) in the Agent's delivery order and binds the upload
* receipts immediately before the message enters the inbox. A repeated
* request identity is acknowledged without delivering it again. An image
* prompt takes the Session's image admission turn when its preparation
* starts and keeps it until the prompt settles, so a model selection lands
* either before its image check or after it entered the inbox; a selection
* waits while a listener holds that image prompt.
* Admit one prompt after Agent and attachment validation; the Remote method
* has already validated the content and built the durable message source.
* @param request - Session identity, prompt content, and delivery mode.
* @param source - the `user-rpc` source recorded on the accepted message.
* @param signal - the requesting client's cancellation; it stops a prompt that waits for its turn or is held at the admission stage.
* @returns acknowledgement that the Agent accepted the prompt.
* @throws `session/prompt-rejected` when an admission listener refuses the prompt or fails.
*/
async prompt(request: SessionPromptRequest, source: MessageSource, signal: AbortSignal): Promise<SessionPromptValue> {
async prompt(request: SessionPromptRequest, source: MessageSource): Promise<SessionPromptValue> {
const agent = await this.resolveAgent(request.sessionId)
if (hasPromptRequest(agent, request.requestId)) return { accepted: true }
const hasImage = request.content.some(part => part.type === 'image')
const uploads: { receiptIds: readonly FileUploadReceiptId[]; binding?: PromptFileBinding } = { receiptIds: [] }
const imageTurn: { release?: () => void } = {}
try {
await this.ctx.agentController.deliver(agent, {
mode: request.mode === 'steer' ? 'steer' : 'followup',
source,
signal,
prepare: async (preparing) => {
if (hasImage) {
// Taken in the Agent's delivery order, so a waiting image prompt keeps its place among the others.
imageTurn.release = await this.agents.acquireImageAdmission(agent, preparing)
await this.requireImageInput(agent)
const admit = async (): Promise<SessionPromptValue> => {
try {
if (hasImage) {
const current = this.ctx.agentModel.resolveForNextStep(agent)
if (current !== undefined) {
const model = await this.ctx.llm.resolveModelInfo(current.provider, current.model)
if (model.inputModalities !== undefined && !model.inputModalities.includes('image')) {
throw new RemoteError(
'session/attachment-invalid',
`Model "${current.model}" does not support image input.`,
{ reason: 'MODEL_DOES_NOT_SUPPORT_IMAGES' },
)
}
}
const admission = resolvePromptFileReceipts(
request.content,
receiptId => this.ctx.fileUploads.resolve(agent, receiptId),
}
const admission = resolvePromptFileReceipts(
request.content,
receiptId => this.ctx.fileUploads.resolve(agent, receiptId),
)
const content = await this.ctx.attachments.admitPromptContent(admission.content)
const message: UserMessage = createUserMessage({ content, source })
if (this.ctx.agents.get(agent.id) !== agent) {
throw new RemoteError(
'session/not-found',
`session "${agent.id}" was disposed during prompt admission`,
{ sessionId: agent.id },
)
uploads.receiptIds = admission.receiptIds
return await this.ctx.attachments.admitPromptContent(admission.content)
},
beforeInsert: () => {
uploads.binding = this.ctx.fileUploads.bindPrompt(agent, uploads.receiptIds, request.requestId)
},
})
uploads.binding?.commit()
} catch (error) {
throw promptFailure(agent, error, signal)
} finally {
uploads.binding?.[Symbol.dispose]()
imageTurn.release?.()
}
return { accepted: true }
}
/** Refuse an image prompt when the Session's current model does not declare image input. */
private async requireImageInput(agent: Agent): Promise<void> {
const current = this.ctx.agentModel.resolveForNextStep(agent)
if (current === undefined) return
const model = await this.ctx.llm.resolveModelInfo(current.provider, current.model)
if (model.inputModalities !== undefined && !model.inputModalities.includes('image')) {
throw new RemoteError(
'session/attachment-invalid',
`Model "${current.model}" does not support image input.`,
{ reason: 'MODEL_DOES_NOT_SUPPORT_IMAGES' },
)
}
using binding = this.ctx.fileUploads.bindPrompt(agent, admission.receiptIds, request.requestId)
if (request.mode === 'steer') agent.steer(message)
else agent.followup(message)
binding.commit()
} catch (error) {
if (remoteErrorOf(error) !== undefined) throw error
if (error instanceof AttachmentError) {
throw new RemoteError('session/attachment-invalid', error.message, { reason: error.code })
}
throw new RemoteError('session/agent-busy', 'prompt rejected', { reason: String(error) })
}
return { accepted: true }
}
return hasImage ? this.agents.serializeImageAdmission(agent, admit) : admit()
}
private async requireModel(selection: Pick<AgentModelSelection, 'provider' | 'model'>): Promise<void> {
@@ -578,36 +562,6 @@ export class SessionCommandController {
}
}
// Typed by the agent-controller constants, so a changed value fails this module's build while it imports no agent-controller value.
const ADMISSION_REJECTED: typeof ADMISSION_REJECTED_CODE = 'AGENT_ADMISSION_REJECTED'
const AGENT_GONE: typeof ADMISSION_AGENT_GONE = 'agent-controller/agent-gone'
/** Recognize an admission refusal by the code that the error of every copy of the agent-controller module carries. */
function isAdmissionRejected(error: unknown): error is AgentAdmissionRejected {
return error instanceof Error && 'code' in error && error.code === ADMISSION_REJECTED
}
/**
* Map a failed prompt admission onto the Session Remote's failures: Remote
* failures and the client's own cancellation pass unchanged, and an admission
* refusal keeps its reason and whether its plugin presents it.
*/
function promptFailure(agent: Agent, error: unknown, signal: AbortSignal): unknown {
if (remoteErrorOf(error) !== undefined || signal.aborted) return error
if (isAdmissionRejected(error)) {
if (error.reason === AGENT_GONE) {
return new RemoteError('session/not-found', `session "${agent.id}" was disposed during prompt admission`, {
sessionId: agent.id,
})
}
return new RemoteError('session/prompt-rejected', error.message, { reason: error.reason, presented: error.presented })
}
if (error instanceof AttachmentError) {
return new RemoteError('session/attachment-invalid', error.message, { reason: error.code })
}
return new RemoteError('session/agent-busy', 'prompt rejected', { reason: String(error) })
}
function resolvePromptFileReceipts(
content: SessionPromptRequest['content'],
stagedFile: (receiptId: FileUploadReceiptId) => FileAttachmentRef | undefined,
packages/api/session-controller/src/index.ts
+3 -8
View File
@@ -118,7 +118,6 @@ export interface SessionControllerInternals {
/** Host service backing the generated `ctx.remote.session` namespace. */
export class SessionController extends TypertRemoteService {
static inject = [
'agentController',
'agentDefaultModel',
'agents',
'attachments',
@@ -452,14 +451,10 @@ export class SessionController extends TypertRemoteService {
}
/**
* Admit one prompt after explicitly resuming its Session. The prompt is
* delivered through `ctx.agentController`, whose `agent-controller/admission`
* stage runs before it enters the inbox, so the call can wait for a listener
* that holds it.
* Admit one prompt after explicitly resuming its Session.
* @param request - Session identity, prompt content, source metadata, and delivery mode.
* @param signal - caller cancellation; it also stops a prompt waiting for or held at the admission stage.
* @param signal - caller cancellation before prompt admission begins.
* @returns acknowledgement that the Agent accepted the prompt.
* @throws `session/prompt-rejected` when an admission listener refuses the prompt or fails.
*/
@Remote('prompt')
async prompt(request: SessionPromptRequest, signal: AbortSignal): Promise<SessionPromptValue> {
@@ -485,7 +480,7 @@ export class SessionController extends TypertRemoteService {
...(clientTimeZone === undefined ? {} : { clientTimeZone }),
...(principal === undefined ? {} : { principal }),
}
return this.commands.prompt(request, source, signal)
return this.commands.prompt(request, source)
}
/**
packages/api/session-controller/src/types.ts
-6
View File
@@ -185,12 +185,6 @@ declare module '@deepseek-ai/dsh-typert-protocol' {
'session/projections-unavailable': Record<string, never>
'session/writer-held': { readonly sessionId: SessionId }
'session/agent-busy': { readonly reason: string }
'session/prompt-rejected': {
/** Code of the refusal, owned by the plugin that refused, such as `forgePanel/worktree-missing`. */
readonly reason: string
/** Whether the refusing plugin shows its own explanation, so the client adds no notice. */
readonly presented: boolean
}
'session/invalid-time-zone': { readonly value: string }
'session/workspace-attach-failed': { readonly sessionId: SessionId; readonly workspaceId: string }
'agent-preset/conflict': {
packages/api/session-controller/tests/commands-upload-file.host.spec.ts
+16 -22
View File
@@ -1,8 +1,6 @@
import TypertRegistry from '@deepseek-ai/dsh-typert-registry'
import SessionProjectionRegistry from '@deepseek-ai/dsh-session-projection'
import { Context, FiberState } from '@deepseek-ai/cordis'
import AgentRegistry from '@deepseek-ai/dsh-agent'
import AgentController from '@deepseek-ai/dsh-agent-controller'
import { createInboxStub } from '@deepseek-ai/dsh-agent-loop-testkit'
import type { Agent } from '@deepseek-ai/dsh-agent'
import * as AgentModelPlugin from '@deepseek-ai/dsh-agent-model'
@@ -42,8 +40,6 @@ async function uploadHarness(origin?: 'subagent'): Promise<{
await ctx.plugin(SessionProjectionRegistry)
await ctx.plugin(AgentModelPlugin)
await ctx.plugin(AgentRegistry)
await ctx.plugin(TypertRegistry)
await ctx.plugin(AgentController)
await ctx.plugin(CommandRuntime)
const session = ctx.sessions.create(SESSION, {
meta: { cwd: '/workspace', ...(origin === undefined ? {} : { origin }) },
@@ -121,8 +117,6 @@ async function uploadHarness(origin?: 'subagent'): Promise<{
/** The owner source `SessionController.prompt()` builds for {@link promptRequest}. */
const USER_SOURCE: MessageSource = { kind: 'user', rpcId: 'req-1' as SessionRequestId }
/** A client that stays connected for the whole prompt. */
const SIGNAL = new AbortController().signal
function promptRequest(content: Parameters<SessionCommandController['prompt']>[0]['content']) {
return {
@@ -174,7 +168,7 @@ describe('Session file uploads', () => {
{ type: 'file', receiptId: receipt.receiptId },
{ type: 'image', mediaType: 'image/png', data: 'AAAA' },
{ type: 'text', text: 'read it' },
]), USER_SOURCE, SIGNAL)
]), USER_SOURCE)
expect(followup).toHaveBeenCalledTimes(1)
const message = followup.mock.calls[0]?.[0] as UserMessage
expect(message.content).toEqual([
@@ -201,7 +195,7 @@ describe('Session file uploads', () => {
name: 'huge.bin',
}))
expect(receipt.file).toMatchObject({ name: 'huge.bin', bytes: 5 })
await controller.prompt(promptRequest([{ type: 'file', receiptId: receipt.receiptId }]), USER_SOURCE, SIGNAL)
await controller.prompt(promptRequest([{ type: 'file', receiptId: receipt.receiptId }]), USER_SOURCE)
expect((followup.mock.calls[0]?.[0] as UserMessage).content).toEqual([
{ type: 'file', attachment: receipt.file },
])
@@ -228,7 +222,7 @@ describe('Session file uploads', () => {
await expect(controller.prompt(promptRequest([
{ type: 'image', mediaType: 'image/png', data: 'AAAA' },
{ type: 'file', receiptId: 'missing-receipt' as FileUploadReceiptId },
]), USER_SOURCE, SIGNAL)).rejects.toMatchObject({ code: 'session/attachment-invalid', details: { reason: 'FILE_NOT_STAGED' } })
]), USER_SOURCE)).rejects.toMatchObject({ code: 'session/attachment-invalid', details: { reason: 'FILE_NOT_STAGED' } })
expect(followup).not.toHaveBeenCalled()
expect(saveImages).not.toHaveBeenCalled()
})
@@ -326,7 +320,7 @@ describe('Session file uploads', () => {
const { controller, uploads, agent } = await uploadHarness()
uploads.retirePrompt(agent, 'not-staged')
const receipt = await uploads.upload(agent, { data: 'AAAA' }, new AbortController().signal)
await controller.prompt(promptRequest([{ type: 'file', receiptId: receipt.receiptId }]), USER_SOURCE, SIGNAL)
await controller.prompt(promptRequest([{ type: 'file', receiptId: receipt.receiptId }]), USER_SOURCE)
expect(uploads.resolve(agent, receipt.receiptId)).toEqual(receipt.file)
uploads.retirePrompt(agent, 'other-request')
expect(uploads.resolve(agent, receipt.receiptId)).toEqual(receipt.file)
@@ -341,7 +335,7 @@ describe('Session file uploads', () => {
{ data: 'AAAA' },
new AbortController().signal,
)
await first.controller.prompt(promptRequest([{ type: 'file', receiptId: observed.receiptId }]), USER_SOURCE, SIGNAL)
await first.controller.prompt(promptRequest([{ type: 'file', receiptId: observed.receiptId }]), USER_SOURCE)
first.agent.session.append('user/message', createUserMessage({
content: [{ type: 'text', text: 'extension event' }],
source: { kind: 'user', rpcId: 1 } as never,
@@ -366,9 +360,9 @@ describe('Session file uploads', () => {
it('deduplicates a retried rpcId already present in the Agent inbox', async () => {
const { controller, agent, followup } = await uploadHarness()
const request = promptRequest([{ type: 'text', text: 'once' }])
await controller.prompt(request, USER_SOURCE, SIGNAL)
await controller.prompt(request, USER_SOURCE)
agent.inbox.append('next-turn', followup.mock.calls[0]?.[0] as UserMessage)
await expect(controller.prompt(request, USER_SOURCE, SIGNAL)).resolves.toEqual({ accepted: true })
await expect(controller.prompt(request, USER_SOURCE)).resolves.toEqual({ accepted: true })
expect(followup).toHaveBeenCalledOnce()
})
@@ -385,7 +379,7 @@ describe('Session file uploads', () => {
source: { kind: 'user', rpcId: request.requestId },
}), { surfaceOp: 'append' })
await expect(controller.prompt(request, USER_SOURCE, SIGNAL)).resolves.toEqual({ accepted: true })
await expect(controller.prompt(request, USER_SOURCE)).resolves.toEqual({ accepted: true })
expect(followup).not.toHaveBeenCalled()
})
@@ -395,7 +389,7 @@ describe('Session file uploads', () => {
saveImages.mockReturnValueOnce(admitted.promise)
const prompting = controller.prompt(promptRequest([
{ type: 'image', mediaType: 'image/png', data: 'AAAA' },
]), USER_SOURCE, SIGNAL)
]), USER_SOURCE)
await vi.waitFor(() => { expect(saveImages).toHaveBeenCalledOnce() })
await disposeAgent()
admitted.resolve([{
@@ -407,7 +401,7 @@ describe('Session file uploads', () => {
it('rejects when a previously bound receipt retires during image admission', async () => {
const { controller, uploads, agent, saveImages } = await uploadHarness()
const receipt = await uploads.upload(agent, { data: 'AAAA' }, new AbortController().signal)
await controller.prompt(promptRequest([{ type: 'file', receiptId: receipt.receiptId }]), USER_SOURCE, SIGNAL)
await controller.prompt(promptRequest([{ type: 'file', receiptId: receipt.receiptId }]), USER_SOURCE)
const admitted = Promise.withResolvers<readonly ImageAttachmentRef[]>()
saveImages.mockReturnValueOnce(admitted.promise)
const prompting = controller.prompt({
@@ -416,7 +410,7 @@ describe('Session file uploads', () => {
{ type: 'image', mediaType: 'image/png', data: 'AAAA' },
]),
requestId: 'req-2' as SessionRequestId,
}, USER_SOURCE, SIGNAL)
}, USER_SOURCE)
await vi.waitFor(() => { expect(saveImages).toHaveBeenCalledOnce() })
uploads.retirePrompt(agent, 'req-1')
admitted.resolve([{
@@ -430,12 +424,12 @@ describe('Session file uploads', () => {
it('keeps the prior receipt binding when a later prompt attempt fails', async () => {
const { controller, uploads, agent, followup } = await uploadHarness()
const receipt = await uploads.upload(agent, { data: 'AAAA' }, new AbortController().signal)
await controller.prompt(promptRequest([{ type: 'file', receiptId: receipt.receiptId }]), USER_SOURCE, SIGNAL)
await controller.prompt(promptRequest([{ type: 'file', receiptId: receipt.receiptId }]), USER_SOURCE)
followup.mockImplementationOnce(() => { throw new Error('busy') })
await expect(controller.prompt({
...promptRequest([{ type: 'file', receiptId: receipt.receiptId }]),
requestId: 'req-2' as SessionRequestId,
}, USER_SOURCE, SIGNAL)).rejects.toMatchObject({ code: 'session/agent-busy' })
}, USER_SOURCE)).rejects.toMatchObject({ code: 'session/agent-busy' })
uploads.retirePrompt(agent, 'req-1')
expect(uploads.resolve(agent, receipt.receiptId)).toBeUndefined()
})
@@ -446,7 +440,7 @@ describe('Session file uploads', () => {
followup.mockImplementationOnce(() => { throw new Error('busy') })
await expect(controller.prompt(promptRequest([
{ type: 'file', receiptId: receipt.receiptId },
]), USER_SOURCE, SIGNAL)).rejects.toMatchObject({ code: 'session/agent-busy' })
]), USER_SOURCE)).rejects.toMatchObject({ code: 'session/agent-busy' })
uploads.retirePrompt(agent, 'req-1')
expect(uploads.resolve(agent, receipt.receiptId)).toEqual(receipt.file)
})
@@ -454,7 +448,7 @@ describe('Session file uploads', () => {
it('retires a prompt-bound receipt when its queued occurrence is removed', async () => {
const { controller, uploads, agent, followup } = await uploadHarness()
const receipt = await uploads.upload(agent, { data: 'AAAA' }, new AbortController().signal)
await controller.prompt(promptRequest([{ type: 'file', receiptId: receipt.receiptId }]), USER_SOURCE, SIGNAL)
await controller.prompt(promptRequest([{ type: 'file', receiptId: receipt.receiptId }]), USER_SOURCE)
const queued = followup.mock.calls[0]?.[0] as UserMessage
agent.inbox.append('next-turn', queued)
expect(await controller.updateQueue({
@@ -475,7 +469,7 @@ describe('Session file uploads', () => {
await controller.prompt(promptRequest([
{ type: 'file', receiptId: first.receiptId },
{ type: 'file', receiptId: second.receiptId },
]), USER_SOURCE, SIGNAL)
]), USER_SOURCE)
const message = followup.mock.calls[0]?.[0] as UserMessage
expect(message.content).toEqual([
packages/api/session-controller/tests/prompt-admission.host.spec.ts
-217
View File
@@ -1,217 +0,0 @@
/**
* Session prompts through the Agent controller's admission stage: how the
* `session/prompt` Remote reports a held, refused, or failed admission to the
* client.
*/
import { describe, expect, it, onTestFinished, vi } from 'vitest'
import { Context } from '@deepseek-ai/cordis'
import AgentRegistry from '@deepseek-ai/dsh-agent'
import type { Agent } from '@deepseek-ai/dsh-agent'
import { createInboxStub } from '@deepseek-ai/dsh-agent-loop-testkit'
import SessionStore from '@deepseek-ai/dsh-session'
import type { UserMessage } from '@deepseek-ai/dsh-session'
import SessionProjectionRegistry from '@deepseek-ai/dsh-session-projection'
import type { SessionPromptRequest, SessionRequestId } from '../src/types.ts'
import { createSessionTestRemote } from './test-remote.ts'
let nextRequestId = 1
async function harness(options: { readonly refusal?: Error } = {}) {
const ctx = new Context()
onTestFinished(() => ctx.fiber.dispose())
await ctx.plugin(SessionStore)
await ctx.plugin(SessionProjectionRegistry)
await ctx.plugin(AgentRegistry)
const session = ctx.sessions.create()
const inbox = createInboxStub()
const deliver = (target: 'next-turn' | 'next-step', message: UserMessage): void => {
if (options.refusal !== undefined) throw options.refusal
inbox.append(target, message)
}
const agent: Agent = {
id: session.id,
options: {},
session,
inbox,
status: 'idle',
ctx,
send: () => {},
followup: (message) => { deliver('next-turn', message) },
steer: (message) => { deliver('next-step', message) },
inject: () => {},
cancel: () => {},
runMaintenance: task => task(new AbortController().signal),
whenIdle: () => Promise.resolve(),
}
const unregister = await ctx.agents.register(agent)
const remote = await createSessionTestRemote(ctx, {
defaultModelSelection: () => ({ provider: 'fixture', model: 'fixture-model' }),
cwd: '/tmp',
})
const prompt = (text: string, mode: SessionPromptRequest['mode'] = 'queue', requestId?: SessionRequestId) => ({
requestId: requestId ?? `admission-${String(nextRequestId++)}` as SessionRequestId,
sessionId: session.id,
mode,
content: [{ type: 'text' as const, text }],
})
return { ctx, agent, inbox, remote, prompt, unregister }
}
function textsOf(messages: readonly UserMessage[]): string[] {
return messages.map(message => message.content.map(block => block.type === 'text' ? block.text : '').join(''))
}
describe('Session prompt admission', () => {
it('admits a queued prompt into the next turn and a steered one into the next step', async () => {
const { ctx, agent, inbox, remote, prompt } = await harness()
const admitted: Agent[] = []
ctx.on('agent-controller/admission', async ({ agent: subject }, next) => {
admitted.push(subject)
return await next()
})
expect(await remote.prompt(prompt('later'))).toEqual({ ok: true, value: { accepted: true } })
expect(await remote.prompt(prompt('now', 'steer'))).toEqual({ ok: true, value: { accepted: true } })
expect(admitted).toEqual([agent, agent])
expect(textsOf(inbox.nextTurn)).toEqual(['later'])
expect(textsOf(inbox.nextStep)).toEqual(['now'])
})
it('keeps the prompt call open while a listener holds it', async () => {
const { ctx, inbox, remote, prompt } = await harness()
const release = Promise.withResolvers<undefined>()
ctx.on('agent-controller/admission', async (_admission, next) => {
await release.promise
return await next()
})
let settled = false
const pending = remote.prompt(prompt('held')).finally(() => { settled = true })
await new Promise<void>((resolve) => { setImmediate(resolve) })
expect(settled).toBe(false)
expect(inbox.nextTurn).toEqual([])
release.resolve(undefined)
expect(await pending).toEqual({ ok: true, value: { accepted: true } })
expect(textsOf(inbox.nextTurn)).toEqual(['held'])
})
it('returns a refusal with its reason and whether its plugin presents it, and keeps the inbox untouched', async () => {
const { ctx, inbox, remote, prompt } = await harness()
let presented = true
ctx.on('agent-controller/admission', () => Promise.resolve({
kind: 'reject' as const, reason: 'fixture/worktree-missing', message: 'The worktree was removed.', presented,
}))
expect(await remote.prompt(prompt('refused'))).toMatchObject({
ok: false,
error: {
code: 'session/prompt-rejected',
message: 'The worktree was removed.',
details: { reason: 'fixture/worktree-missing', presented: true },
},
})
presented = false
expect(await remote.prompt(prompt('refused again'))).toMatchObject({
ok: false,
error: { code: 'session/prompt-rejected', details: { reason: 'fixture/worktree-missing', presented: false } },
})
expect(inbox.nextTurn).toEqual([])
})
it('reports a failed listener as a refusal the client explains', async () => {
const { ctx, inbox, remote, prompt } = await harness()
ctx.on('agent-controller/admission', () => Promise.reject(new Error('listener broke')))
const result = await remote.prompt(prompt('broken'))
expect(result).toMatchObject({
ok: false,
error: { code: 'session/prompt-rejected', details: { reason: 'agent-controller/admission-failed', presented: false } },
})
expect(result.ok ? '' : result.error.message).toContain('listener broke')
expect(inbox.nextTurn).toEqual([])
})
it('cancels a held prompt whose client went away', async () => {
const { ctx, inbox, remote, prompt } = await harness()
ctx.on('agent-controller/admission', async ({ signal }, next) => {
await new Promise<never>((_resolve, reject) => {
signal.addEventListener('abort', () => { reject(signal.reason as Error) }, { once: true })
})
return await next()
})
const client = new AbortController()
const pending = remote.prompt(prompt('abandoned'), client.signal)
await new Promise<void>((resolve) => { setImmediate(resolve) })
client.abort(new Error('client left'))
expect(await pending).toMatchObject({ ok: false, error: { code: 'gateway/cancelled' } })
expect(inbox.nextTurn).toEqual([])
})
it('reports a Session whose Agent was disposed while its prompt was held as not found', async () => {
const { ctx, agent, remote, prompt, unregister } = await harness()
const release = Promise.withResolvers<undefined>()
ctx.on('agent-controller/admission', async (_admission, next) => {
await release.promise
return await next()
})
const pending = remote.prompt(prompt('orphaned'))
await new Promise<void>((resolve) => { setImmediate(resolve) })
await unregister()
release.resolve(undefined)
expect(await pending).toMatchObject({
ok: false,
error: { code: 'session/not-found', details: { sessionId: agent.id } },
})
})
it('maps a refusal from another copy of the agent-controller module by its code', async () => {
const { ctx, agent, remote, prompt } = await harness()
// Another copy of the module raises the same fields from its own class.
const refusal = (reason: string): Error => Object.assign(new Error('The worktree was removed.'), {
name: 'AgentAdmissionRejected', code: 'AGENT_ADMISSION_REJECTED', reason, presented: true,
})
vi.spyOn(ctx.agentController, 'deliver')
.mockRejectedValueOnce(refusal('fixture/worktree-missing'))
.mockRejectedValueOnce(refusal('agent-controller/agent-gone'))
expect(await remote.prompt(prompt('refused'))).toMatchObject({
ok: false,
error: {
code: 'session/prompt-rejected',
message: 'The worktree was removed.',
details: { reason: 'fixture/worktree-missing', presented: true },
},
})
expect(await remote.prompt(prompt('orphaned'))).toMatchObject({
ok: false,
error: { code: 'session/not-found', details: { sessionId: agent.id } },
})
})
it('acknowledges a repeated request without admitting it again', async () => {
const { ctx, inbox, remote, prompt } = await harness()
let admissions = 0
ctx.on('agent-controller/admission', async (_admission, next) => {
admissions += 1
return await next()
})
const first = prompt('once')
expect(await remote.prompt(first)).toEqual({ ok: true, value: { accepted: true } })
expect(await remote.prompt(prompt('once', 'queue', first.requestId))).toEqual({ ok: true, value: { accepted: true } })
expect(admissions).toBe(1)
expect(textsOf(inbox.nextTurn)).toEqual(['once'])
})
it('reports the Agent refusing to queue a prompt as busy', async () => {
const { remote, prompt } = await harness({ refusal: new Error('inbox closed') })
expect(await remote.prompt(prompt('rejected'))).toMatchObject({
ok: false,
error: { code: 'session/agent-busy', details: { reason: 'Error: inbox closed' } },
})
})
})
packages/api/session-controller/tests/session-models.host.spec.ts
-144
View File
@@ -151,46 +151,6 @@ async function harness(logged?: {
return { ctx, agent, sessionId: session.id }
}
/** An image store whose saved attachments are one-pixel PNGs. */
function provideImageStore(ctx: Context): void {
const attachments = {
imageLimits: {
maxImageBytes: 4,
maxImagesPerMessage: 2,
maxMessageImageBytes: 4,
maxImagePixels: 4,
maxImageDimension: 2000,
mediaTypes: ['image/png'],
},
validateImage: () => Promise.resolve(),
saveImage: (input: { data: Uint8Array; mediaType: 'image/png' }) => Promise.resolve({
attachmentId: `att-${String(input.data[0])}`, mediaType: input.mediaType, bytes: input.data.byteLength, width: 1, height: 1,
}),
}
ctx.provide('attachments', Object.setPrototypeOf(attachments, AttachmentStore.prototype) as never)
}
/** A queued prompt with its text, after a one-pixel image when `withImage` is set. */
function imagePrompt(sessionId: SessionId, text: string, withImage: boolean): SessionPromptRequest {
return promptRequest({
sessionId,
mode: 'queue',
content: [
...withImage ? [{ type: 'image' as const, mediaType: 'image/png' as const, data: 'AQ==' }] : [],
{ type: 'text' as const, text },
],
})
}
function textOf(message: UserMessage): string {
return message.content.map(block => block.type === 'text' ? block.text : '').join('')
}
/** Let pending prompts run as far as they can. */
async function settle(): Promise<void> {
await new Promise<void>((resolve) => { setImmediate(resolve) })
}
function expectValue<T>(result: { ok: true; value: T } | { ok: false }): T {
if (!result.ok) throw new Error('expected successful response')
return result.value
@@ -367,110 +327,6 @@ describe('Web session model selection', () => {
await ctx.fiber.dispose()
})
it('admits image prompts in arrival order behind an image prompt that a listener holds', async () => {
const { ctx, agent, sessionId } = await harness()
provideImageStore(ctx)
const followed: string[] = []
Object.assign(agent, { followup: (message: UserMessage) => { followed.push(textOf(message)) } })
const release = Promise.withResolvers<undefined>()
ctx.on('agent-controller/admission', async ({ message }, next) => {
if (textOf(message) === 'held') await release.promise
return await next()
})
const remote = await createSessionTestRemote(ctx, {
defaultModelSelection: () => ({ provider: 'deepseek-official', model: 'deepseek-chat' }),
cwd: '/tmp',
})
const prompts = [
remote.prompt(imagePrompt(sessionId, 'held', true)),
remote.prompt(imagePrompt(sessionId, 'second image', true)),
remote.prompt(imagePrompt(sessionId, 'third text', false)),
]
await settle()
expect(followed).toEqual([])
release.resolve(undefined)
expect((await Promise.all(prompts)).map(result => result.ok)).toEqual([true, true, true])
expect(followed).toEqual(['held', 'second image', 'third text'])
await ctx.fiber.dispose()
})
it('keeps a model selection behind the image prompt being admitted but not behind one still queued', async () => {
const { ctx, agent, sessionId } = await harness()
registerTextOnly(ctx)
provideImageStore(ctx)
const events: string[] = []
Object.assign(agent, { followup: (message: UserMessage) => { events.push(`admitted ${textOf(message)}`) } })
const releases = new Map([['held image', Promise.withResolvers<undefined>()], ['held text', Promise.withResolvers<undefined>()]])
ctx.on('agent-controller/admission', async ({ message }, next) => {
await releases.get(textOf(message))?.promise
return await next()
})
const remote = await createSessionTestRemote(ctx, {
defaultModelSelection: () => ({ provider: 'deepseek-official', model: 'deepseek-chat' }),
cwd: '/tmp',
})
const select = () => remote.selectModel(request({ sessionId, provider: 'text-only', model: 'plain' }))
.then((result) => { events.push('selected'); return result })
// A held image prompt was checked against the current model, so the selection waits for its append.
const heldImage = remote.prompt(imagePrompt(sessionId, 'held image', true))
await settle()
const first = select()
await settle()
expect(events).toEqual([])
releases.get('held image')?.resolve(undefined)
expect((await heldImage).ok).toBe(true)
expect((await first).ok).toBe(true)
expect(events).toEqual(['admitted held image', 'selected'])
// An image prompt queued behind a held text prompt has not been checked yet, so the selection does not wait for it.
const heldText = remote.prompt(imagePrompt(sessionId, 'held text', false))
const queuedImage = remote.prompt(imagePrompt(sessionId, 'queued image', true))
await settle()
expect((await select()).ok).toBe(true)
expect(events).toEqual(['admitted held image', 'selected', 'selected'])
releases.get('held text')?.resolve(undefined)
expect((await heldText).ok).toBe(true)
expect(await queuedImage).toMatchObject({
ok: false, error: { code: 'session/attachment-invalid', details: { reason: 'MODEL_DOES_NOT_SUPPORT_IMAGES' } },
})
expect(events).toEqual(['admitted held image', 'selected', 'selected', 'admitted held text'])
await ctx.fiber.dispose()
})
it('passes the image admission turn on when the sender of a waiting image prompt leaves', async () => {
const { ctx, agent, sessionId } = await harness()
provideImageStore(ctx)
const followed: string[] = []
Object.assign(agent, { followup: (message: UserMessage) => { followed.push(textOf(message)) } })
const listing = Promise.withResolvers<readonly LlmModelInfo[]>()
ctx.llm.registerAdapter(['gated'], new class extends CatalogAdapter {
override listModels(): Promise<readonly LlmModelInfo[]> {
return listing.promise
}
}('Gated', []))
const remote = await createSessionTestRemote(ctx, {
defaultModelSelection: () => ({ provider: 'deepseek-official', model: 'deepseek-chat' }),
cwd: '/tmp',
})
// The selection holds the image admission turn while the provider lists its models.
const selection = remote.selectModel(request({ sessionId, provider: 'gated', model: 'slow' }))
await settle()
const sender = new AbortController()
const leaving = remote.prompt(imagePrompt(sessionId, 'left', true), sender.signal)
await settle()
sender.abort(new Error('client left'))
expect(await leaving).toMatchObject({ ok: false, error: { code: 'gateway/cancelled' } })
listing.resolve([{ provider: 'gated', id: 'slow', name: 'Slow' }])
expect((await selection).ok).toBe(true)
expect((await remote.prompt(imagePrompt(sessionId, 'after', true))).ok).toBe(true)
expect(followed).toEqual(['after'])
await ctx.fiber.dispose()
})
it('authorizes attachment bytes only when the session event stream references the id', async () => {
const { ctx, agent, sessionId } = await harness()
const ref = {
packages/api/session-controller/tests/test-remote.ts
-2
View File
@@ -31,7 +31,6 @@ import {
remoteErrorOf,
type RemoteResult,
} from '@deepseek-ai/dsh-typert-protocol'
import AgentController from '@deepseek-ai/dsh-agent-controller'
import RemoteAccessService from '@deepseek-ai/dsh-api-remote-access'
import SessionController from '../src/index.ts'
import type {
@@ -294,7 +293,6 @@ async function installControllers(
} as never)
}
installSessionReadTestServices(ctx)
if (ctx.get('agentController') === undefined) new AgentController(ctx)
if (ctx.get('remoteAccess') === undefined) new RemoteAccessService(ctx)
const cwd = vi.spyOn(process, 'cwd').mockReturnValue(defaults.cwd)
class TestController extends SessionController {
packages/api/session-controller/tsconfig.host.json
-3
View File
@@ -31,9 +31,6 @@
{
"path": "../../core/agent"
},
{
"path": "../../core/agent-controller"
},
{
"path": "../../core/agent-default-model"
},
packages/bundle/base/cordis.patch.yml
-4
View File
@@ -84,10 +84,6 @@
- id: agent-model
name: '@deepseek-ai/dsh-agent-model'
# Delivers messages from outside an Agent into its inbox after the admission
# stage; entry points such as the Web Session Remote deliver through it.
- id: agent-controller
name: '@deepseek-ai/dsh-agent-controller'
- id: plugin-package-inventory-deepseek
name: '@deepseek-ai/dsh-plugin-package-inventory-deepseek'
packages/bundle/base/package.json
-1
View File
@@ -37,7 +37,6 @@
"@deepseek-ai/dsh-hmr": "workspace:*",
"@deepseek-ai/cordis-plugin-timer": "workspace:~",
"@deepseek-ai/dsh-agent": "workspace:*",
"@deepseek-ai/dsh-agent-controller": "workspace:*",
"@deepseek-ai/dsh-agent-default-model": "workspace:*",
"@deepseek-ai/dsh-agent-loop": "workspace:*",
"@deepseek-ai/dsh-attachment-local": "workspace:*",
packages/client/AGENTS.md
+1 -1
View File
@@ -115,7 +115,7 @@ One UI feature = one plugin package (`src/client/` browser half). A multi-domain
## Styling and localization
[docs/web-styling.md](../../docs/web-styling.md) is authoritative. Shared `--dsw-*` tokens and global sheets live in `ui-theme/src/styles/`; feature components consume semantic aliases through CSS Modules and `clsx`, with no literal colors, component library, or Tailwind. Code comments are English. Design judgment beyond those rules — feedback surfaces, loading states, overlay safety, platform window adaptation — follows [dsh-client-ui-ux](../../.agents/skills/dsh-client-ui-ux/SKILL.md).
[docs/web-styling.md](../../docs/web-styling.md) is authoritative. Shared `--dsw-*` tokens and global sheets live in `ui-theme/src/styles/`; feature components consume semantic aliases through CSS Modules and `clsx`, with no literal colors, component library, or Tailwind. Code comments are English. Design judgment beyond those rules — feedback surfaces, loading states, overlay safety, platform window adaptation — follows [dsh-client-ui-ux](../../.agents/skills/dsh-client-ui-ux/SKILL.md); responsive layout and telling a mobile or touch client from a desktop one follow [dsh-responsive-ui](../../.agents/skills/dsh-responsive-ui/SKILL.md).
Menus use `Menu` or `MenuSurface`, including custom listboxes, except the explicitly listed schedule containers. Preserve the shared transparency and blur ([rules](../../docs/web-styling.md#component-rules)).
packages/client/locale/src/client/LanguageRow.module.css
+4 -2
View File
@@ -42,8 +42,10 @@
cursor: pointer;
}
.selector:hover {
background: var(--dsw-alias-interactive-bg-hover);
@media (hover: hover) {
.selector:hover {
background: var(--dsw-alias-interactive-bg-hover);
}
}
.chevron {
packages/client/ui-agent-preset/README.i18n.yaml
+2 -2
View File
@@ -15,8 +15,8 @@
en: d152484eb41ac6b4
zh: b3ad7867f23dad1c
/deepseek-ai-dsh-client-ui-agent-preset/use-this-package:
en: e015215c32d78b52
zh: e8a8029903de7515
en: 475fae2164ced556
zh: 4d0c729ffc210a83
/deepseek-ai-dsh-client-ui-agent-preset/understand-the-implementation:
en: d7a38ad9e5497789
zh: 5f5595ddd90fa6bd
packages/client/ui-agent-preset/README.md
+1 -1
View File
@@ -25,7 +25,7 @@ Choose Agent presets and the new-task default in Web, read what each mode does a
<a id="use-this-package"></a>
## Use this package
Settings shows the built-in and custom card groups with default highlighting and card-body selection; a group without presets is omitted, except the custom group, which keeps its Creator entry on screen. Every card offers “View configuration”, which opens the preset's declared plugin list as read-only YAML in the Loader's own dialect (`!!js` conditions included); a failed preset stays readable because its diagnostic points into that YAML. Escape closes only the viewer and returns focus to its card; leaving Settings clears the viewer, and a late read does not reopen it. The page edits nothing: the Creator entry starts a Creator-mode task that authors or overrides a preset as a bundle, offered while the `cordis` preset is on the roster and a conversation flow exists.
Settings shows the built-in and custom card groups with default highlighting and card-body selection; a group without presets is omitted, except the custom group, which keeps its Creator entry on screen. Coarse-pointer devices show complete card descriptions; other devices clamp descriptions to four lines with the full text available on hover. Every card offers “View configuration”, which opens the preset's declared plugin list as read-only YAML in the Loader's own dialect (`!!js` conditions included); a failed preset stays readable because its diagnostic points into that YAML. Escape closes only the viewer and returns focus to its card; leaving Settings clears the viewer, and a late read does not reopen it. The page edits nothing: the Creator entry starts a Creator-mode task that authors or overrides a preset as a bundle, offered while the `cordis` preset is on the roster and a conversation flow exists.
With Coding Tools off in General Settings, the new-session picker and Settings roster offer Standard, Creator, and custom presets; turning them on also offers PTC and Minimal. Default selection and the Creator entry remain available in both states. Choosing a healthy default also synchronizes the blank session on the current new-task surface. Creator starts a new task using the `cordis` preset.
packages/client/ui-agent-preset/README.zh.md
+1 -1
View File
@@ -25,7 +25,7 @@ kind: "package-reference"
<a id="use-this-package"></a>
## 使用此包
设置页显示内置与自定义卡片分组、默认项高亮和点击卡片选择；没有 preset 的分组不显示，但自定义分组会保留其创造入口。每张卡片都提供「查看配置」，以只读 YAML 打开该 preset 声明的插件列表，使用 Loader 自己的方言（含 `!!js` 条件）；加载失败的 preset 同样可读，因为其诊断信息正指向这份 YAML。Escape 只关闭查看器并将焦点还给卡片；离开设置页会清空查看器，迟到的读取结果不会重新打开它。本页不编辑任何内容：创造入口启动一个创造模式任务，以 bundle 形式创建或覆盖 preset；当 `cordis` preset 在列表中且存在会话流程时提供。
设置页显示内置与自定义卡片分组、默认项高亮和点击卡片选择；没有 preset 的分组不显示，但自定义分组会保留其创造入口。粗指针设备完整显示卡片描述；其他设备最多显示四行描述，悬停可查看全文。每张卡片都提供「查看配置」，以只读 YAML 打开该 preset 声明的插件列表，使用 Loader 自己的方言（含 `!!js` 条件）；加载失败的 preset 同样可读，因为其诊断信息正指向这份 YAML。Escape 只关闭查看器并将焦点还给卡片；离开设置页会清空查看器，迟到的读取结果不会重新打开它。本页不编辑任何内容：创造入口启动一个创造模式任务，以 bundle 形式创建或覆盖 preset；当 `cordis` preset 在列表中且存在会话流程时提供。
关闭通用设置中的代码工作工具后，新会话选择器和设置页列表提供标准、创造和自定义预设；开启后还提供 PTC 和极简模式。两种状态下均保留默认值选择和创造入口。选择健康定义作为默认值也会同步当前新任务页面的空白会话。Creator 入口开启一个使用 `cordis` preset 的新任务。
packages/client/ui-agent-preset/src/client/AgentPresetSeat.module.css
+6 -1
View File
@@ -34,11 +34,16 @@
cursor: pointer;
}
.seat:not(:disabled):hover,
.seat[aria-expanded='true'] {
background: var(--dsw-alias-interactive-bg-hover);
}
@media (hover: hover) {
.seat:not(:disabled):hover {
background: var(--dsw-alias-interactive-bg-hover);
}
}
.seat:disabled {
cursor: default;
color: var(--dsw-alias-label-quaternary);
packages/client/ui-agent-preset/src/client/AgentPresetSection.module.css
+53 -12
View File
@@ -61,8 +61,10 @@
}
.card:hover:not(.cardActive) {
background: var(--dsw-alias-interactive-bg-hover);
@media (hover: hover) {
.card:hover:not(.cardActive) {
background: var(--dsw-alias-interactive-bg-hover);
}
}
/* The default preset reads as selected exactly like the Appearance cubes:
@@ -81,8 +83,10 @@
border-color: var(--dsw-alias-state-error-primary);
}
.cardBroken:hover {
border-color: var(--dsw-alias-state-error-primary);
@media (hover: hover) {
.cardBroken:hover {
border-color: var(--dsw-alias-state-error-primary);
}
}
.brokenBadge {
@@ -127,11 +131,16 @@
/* Focusing the card reveals it too: a broken card keeps its place in the tab
order (`aria-disabled`, not `disabled`), so this is the keyboard path to a
reason the card face no longer shows. */
.brokenBadge:hover .brokenTip,
.cardMain:focus-visible .brokenTip {
opacity: 1;
}
@media (hover: hover) {
.brokenBadge:hover .brokenTip {
opacity: 1;
}
}
.cardMain[aria-disabled='true'] {
cursor: default;
}
@@ -155,6 +164,18 @@
white-space: nowrap;
}
@media (hover: none) {
.cardBrokenReason {
position: static;
display: block;
width: auto;
height: auto;
overflow: visible;
clip: auto;
white-space: normal;
}
}
/* The card body is the control that picks the preset. */
.cardMain {
flex: 1;
@@ -230,6 +251,12 @@
overflow-wrap: anywhere;
}
@media (pointer: coarse) {
.cardDesc {
-webkit-line-clamp: unset;
}
}
.cardId {
flex-shrink: 0;
max-width: 35%;
@@ -267,11 +294,16 @@
color: var(--dsw-alias-label-tertiary);
}
.helpButton:hover,
.helpButton:focus-visible {
color: var(--dsw-alias-label-primary);
}
@media (hover: hover) {
.helpButton:hover {
color: var(--dsw-alias-label-primary);
}
}
/* Icon-only actions: the label rides `data-tip` so the row stays quiet until
someone reaches for it. */
.iconButton {
@@ -287,9 +319,11 @@
align-items: center;
}
.iconButton:hover {
background: var(--dsw-alias-bg-layer-1);
color: var(--dsw-alias-label-primary);
@media (hover: hover) {
.iconButton:hover {
background: var(--dsw-alias-bg-layer-1);
color: var(--dsw-alias-label-primary);
}
}
.iconButton:focus-visible {
@@ -315,11 +349,16 @@
transition: opacity .12s;
}
.iconButton:hover::after,
.iconButton:focus-visible::after {
opacity: 1;
}
@media (hover: hover) {
.iconButton:hover::after {
opacity: 1;
}
}
.dialog {
width: min(720px, 100%);
}
@@ -370,8 +409,10 @@
cursor: pointer;
}
.creatorButton:hover:not(:disabled) {
background: var(--dsw-alias-interactive-bg-hover);
@media (hover: hover) {
.creatorButton:hover:not(:disabled) {
background: var(--dsw-alias-interactive-bg-hover);
}
}
.creatorButton:disabled {
packages/client/ui-agent-preset/tests/section.client.spec.tsx
+1 -1
View File
@@ -287,7 +287,7 @@ it.each([false, true])('only offers a description tooltip when the card clips it
const disconnect = vi.spyOn(ResizeObserver.prototype, 'disconnect')
vi.useFakeTimers()
view({ rows: [{ id: 'mine', isDefault: false, description: 'Preset description' }] })
fireEvent.mouseEnter(screen.getByText('Preset description'))
fireEvent.pointerEnter(screen.getByText('Preset description'))
act(() => { vi.advanceTimersByTime(400) })
expect(screen.queryByRole('tooltip')?.textContent ?? null).toBe(overflow ? 'Preset description' : null)
cleanup()
packages/client/ui-approval/src/client/ApprovalPanel.module.css
+6 -4
View File
@@ -60,8 +60,10 @@
padding: 14px 16px;
}
.reject:hover:not(:disabled) {
background: var(--dsw-alias-interactive-bg-hover-danger);
color: var(--dsw-alias-state-error-primary);
--dsh-button-outline-color: transparent;
@media (hover: hover) {
.reject:hover:not(:disabled) {
background: var(--dsw-alias-interactive-bg-hover-danger);
color: var(--dsw-alias-state-error-primary);
--dsh-button-outline-color: transparent;
}
}
Loaded 100 of 339 files, more files were not shown because too many files have changed in this diff. Show more