Commit Graph
2217 Commits
Author SHA1 Message Date
Turtle ce6ab42635 Merge pull request #8225 from deepseek-harness/turtle/add-desktop-ua-suffix-e911e659413d
feat(llm): report the booted profile in the provider User-Agent
2026-11-11 09:28:02 +08:00
creatixchu 0874eb2217 fix(cua): retain macOS permission diagnostic process IDs 2026-11-10 22:22:09 +08:00
creatixchu 1a62850eec Merge master into Windows CUA branch 2026-11-10 20:21:17 +08:00
imccyu 793ce790a6 feat(ptc): add scoped REPL execution with copied top-level bindings
With replScoped: true, persistent-in-memory runs each call as a separate
VM Script in one retained context. Each call starts with copies of the
top-level bindings left by earlier calls, so a later call may declare any
earlier name again with any keyword while earlier functions and
still-running code keep their values. Results, top-level await, errors
and asynchronous failures follow the existing REPL evaluator; modules
load only through dynamic import(), resolved from the environment's first
working directory.

The control protocol moves to version 3; the upgrade guide covers
preinstalled bootstraps and the new PtcProgramFormat member. The Agent
Note records the decision and its comparison with node:repl and
vm-script, and the ptc-scoped-repl snapshot pins the model guidance.
2026-11-10 18:52:19 +08:00
creatixchu 1aefd4de11 Merge master and migrate Windows CUA callers to explicit inspection 2026-11-10 18:24:02 +08:00
creatixchu 1991fa2b22 fix(cua): address shared API, packaging and migration review 2026-11-10 18:10:43 +08:00
Turtleandsjw1231 51951b1593 feat(llm): report the booted profile in the provider User-Agent
Co-authored-by: sjw1231 <[email protected]>
2026-11-10 16:19:32 +08:00
_Kerman 4e9adccec7 Merge branch 'master' into xtr/ptc-inspect-output 2026-11-10 16:18:34 +08:00
_Kerman 5fd443618f fix(ptc): preserve raw text and cross-realm image inspection 2026-11-10 16:07:14 +08:00
Turtle d79851e05d Merge pull request #8180 from deepseek-harness/feat/cua-run-code-standard
feat(cua): 为标准模式提供持久 cua_run_code 工具
2026-11-10 07:59:40 +00:00
_Kerman 86984d8358 Merge branch 'feat/cua-run-code-standard' into xtr/ptc-inspect-output 2026-11-10 14:24:26 +08:00
turtle1999 26427a26e1 refactor(ptc): initialize each environment through a registered hook 2026-11-09 22:19:35 +08:00
Yichen Jiang ef22fbe7d3 fix(forge): report unconfirmed writes and harden the overview actions
- ForgeActionFailure gains `unknown` for an answer that does not confirm
  the outcome; the GitHub provider passes it on instead of `failed`, the
  panel Remote refuses with `forgePanel/action-unknown`, a viewed mark
  stays, and the other actions ask the reader to check the browser.
- Reruns skip workflow runs with a pending check, try every finished run,
  and report the refused ones together; the card offers a rerun only when
  a failed check's run has finished.
- Review requests resolve users and teams to node ids and call
  `requestReviews`, which GitHub Enterprise Server 3.19 and 3.20 have;
  bots still go through `requestReviewsByLogin`.
- Viewed marks are kept per change request and head, so a reread at the
  same head keeps writes in flight and their locks, and paths named after
  object properties read only their own entries.
- The reviewer search never chooses from the listing of an earlier query.
- Documentation states the unknown code, the branch update evidence, and
  the `canBeRebased` approximation, and drops stray blank lines.
2026-11-09 20:56:46 +08:00
turtle1999 f0e9dc1c92 fix(cua): support clean source runs and preserve per-mode guidance 2026-11-09 18:17:56 +08:00
_Kerman a2ea3970fd refactor(ptc): return ordered text and image output 2026-11-09 17:44:12 +08:00
_Kerman 1fed265b3b fix(ptc): bound binary inspection and preserve failure output 2026-11-09 17:19:48 +08:00
Yichen Jiang b1fae5b326 docs(forge): document the overview and viewed marks
The forge subsystem page gains the overview and viewed mark types and
contracts, and the generated Cordis, configuration, and event catalogs
follow the new service methods and the maxViewedFiles option.
2026-11-09 17:13:17 +08:00
_Kerman 83e3d0ebc0 Merge remote-tracking branch 'origin/master' into xtr/ptc-inspect-output
# Conflicts:
#	packages/experimental/cua-macos/README.i18n.yaml
#	packages/experimental/cua-macos/README.md
#	packages/experimental/cua-macos/README.zh.md
#	packages/experimental/cua/src/api.ts
#	packages/experimental/cua/src/guidance.ts
#	packages/experimental/cua/src/types.ts
#	packages/experimental/cua/tests/api.spec.ts
#	snapshots/session/cua-native/system-prompt.expected.md
#	snapshots/session/cua-ptc/system-prompt.expected.md
#	snapshots/session/ptc-script-retained/system-prompt.expected.md
2026-11-09 16:17:18 +08:00
turtle1999 b9966a12a6 feat(cua): expose persistent code tool outside PTC mode 2026-11-09 15:53:22 +08:00
_Kerman 4e72e43037 chore(ptc): align fixtures and docs with simplified image handling 2026-11-09 15:42:49 +08:00
_Kerman 724e56b286 feat(ptc): unify explicit text and image inspection 2026-11-09 15:06:09 +08:00
Yichen Jiang 28df6c3117 Merge remote-tracking branch 'origin/feat/forge-review-threads' into feat/forge-review-writes
# Conflicts:
#	docs/config-catalog.md
#	docs/config-catalog.zh.md
#	packages/experimental/client-ui-forge/README.i18n.yaml
#	packages/experimental/client-ui-forge/src/client/ChangesTab.tsx
#	packages/experimental/forge-github/README.i18n.yaml
#	packages/experimental/forge-github/README.md
#	packages/experimental/forge-github/README.zh.md
#	packages/experimental/forge/README.i18n.yaml
#	packages/experimental/forge/src/index.ts
#	packages/extensions/tool-cordis/src/api-catalog.ts
2026-11-09 14:53:18 +08:00
Yichen Jiang 2479e6ddab Merge remote-tracking branch 'origin/master' into feat/forge-review-threads
# Conflicts:
#	docs/event-producer-consumer.md
#	docs/event-producer-consumer.zh.md
2026-11-09 14:43:20 +08:00
Yichen Jiang dacbc37209 Merge remote-tracking branch 'origin/master' into feat/forge-review-threads
# Conflicts:
#	docs/config-catalog.md
#	docs/config-catalog.zh.md
#	packages/experimental/client-ui-forge/README.i18n.yaml
#	packages/experimental/client-ui-forge/README.md
#	packages/experimental/client-ui-forge/README.zh.md
#	packages/experimental/client-ui-forge/src/client/ChangesTab.tsx
#	packages/experimental/client-ui-forge/src/client/DiffFile.tsx
#	packages/experimental/forge-github/README.i18n.yaml
#	packages/experimental/forge-github/README.md
#	packages/experimental/forge-github/README.zh.md
#	packages/experimental/forge/README.i18n.yaml
#	packages/experimental/forge/src/index.ts
#	packages/extensions/tool-cordis/src/api-catalog.ts
2026-11-09 14:28:39 +08:00
Yichen Jiang 6e882cd6e3 fix(forge): confine review writes to the bound pull request and reconcile unknown submissions
A write to a thread or comment now reads, in the head check, the pull
request that holds it and is refused without a mutation when that is
another pull request. Submitting the pending review keeps the
submission to reconcile, so an unknown outcome blocks a second review.
GraphQL errors count as refusals only when every type names one, so a
GitHub timeout reports an unknown outcome. Pending reviews are found
by viewerDidAuthor among up to 100, without the unused commit field. A
resolved thread stays open while the reader writes in it, unsupported
writes no longer poll, and the docs state the partial two-step write.
2026-11-09 14:08:22 +08:00
Jiawei Song b43fcce60e Merge pull request #8006 from deepseek-harness/refactor/tool-errors-fs
feat(fs,lsp,present): give file, search, LSP, and present tool failures error codes
2026-11-09 13:59:49 +08:00
Jiawei Song 74ae5f0da9 Merge pull request #8053 from deepseek-harness/refactor/tool-errors-shell
feat(shell,sandbox,jobs,terminal): give shell, sandbox, job, and terminal tool failures error codes
2026-11-09 13:56:40 +08:00
Yichen Jiang 9f875cf44d Merge remote-tracking branch 'origin/feat/forge-review-threads' into feat/forge-review-writes
# Conflicts:
#	apps/web/tests/forge-panel.e2e.ts
#	docs/config-catalog.md
#	docs/config-catalog.zh.md
#	docs/subsystems/forge.i18n.yaml
#	docs/subsystems/forge.md
#	docs/subsystems/forge.zh.md
#	packages/experimental/client-ui-forge/README.i18n.yaml
#	packages/experimental/client-ui-forge/README.md
#	packages/experimental/client-ui-forge/README.zh.md
#	packages/experimental/client-ui-forge/src/client/ChangesTab.tsx
#	packages/experimental/client-ui-forge/src/client/mount.ts
#	packages/experimental/client-ui-forge/src/client/thread-state.ts
#	packages/experimental/client-ui-forge/src/index.ts
#	packages/experimental/client-ui-forge/tests/changes-tab.client.spec.tsx
#	packages/experimental/client-ui-forge/tests/mount.client.spec.ts
#	packages/experimental/client-ui-forge/tests/review-cache.host.spec.ts
#	packages/experimental/client-ui-forge/tests/review.host.spec.ts
#	packages/experimental/forge-github/README.i18n.yaml
#	packages/experimental/forge-github/src/threads.ts
#	packages/experimental/forge-github/tests/threads.spec.ts
#	packages/experimental/forge/src/types.ts
#	packages/experimental/forge/tests/registry.spec.ts
#	packages/extensions/tool-cordis/src/api-catalog.ts
2026-11-09 13:03:22 +08:00
_Kerman 790991a6c0 docs(ptc): align inspection guidance and references 2026-11-09 12:56:55 +08:00
Yichen Jiang 1ba5977691 fix(forge-panel): address the review thread findings
- A jump to a thread centered its card on every later jump to the same file, since the jump target outlived the jump.
  The scroll target now carries the card to center and is cleared when the jump ends; unfolding still follows the
  latest thread jump.
- In a commit range view, a thread on a file the chosen commits do not change opened on the host instead of choosing
  all commits; a thread the chosen commits do not hold now always chooses all commits first.
- A quote of an old-side thread named the head as the commit its lines number. It now says the lines number the file
  before the change and names the head only as where the diff ends, and ForgeAnchor documents the same.
- Refresh reads review threads from the host past the panel's cache (`changeRequestReview` takes `fresh`), so a thread
  resolved since shows without waiting for `reviewRefreshMs`; `maxCachedReviewBytes` defaults to 16 MiB, at least
  one read under forge-github's default `maxOutputBytes`.
- Picking a thread from a list that covers the files moves the focus to the thread's card.
- `ForgeReviewData.viewer`, its GraphQL field, `ThreadActions.viewer`, and `ThreadReadState.unsupported` had no
  consumer and are removed.
- A quoted outdated thread is named by the lines it was written on; a missing pull request or repository makes a
  review read `not-found`; a truncated read that kept no thread still offers the thread list.
2026-11-09 12:50:33 +08:00
_Kerman c1783e9dff feat(ptc): select program output with Harness.inspect 2026-11-09 12:00:07 +08:00
Yichen Jiang f03981630c Merge remote-tracking branch 'origin/feat/forge-review-threads' into feat/forge-review-writes
# Conflicts:
#	apps/web/tests/forge-panel.e2e.ts
#	packages/experimental/client-ui-forge/README.i18n.yaml
2026-11-09 11:50:38 +08:00
Yichen Jiang ce1e1f2681 Merge remote-tracking branch 'origin/master' into perf/forge-changes-view 2026-11-09 11:47:49 +08:00
Yichen Jiang c04c40b68a docs(forge): document review writes and record pending review drafts
The forge subsystem page documents ctx.forge.writeReview, its write
and result types, and the panel's changeRequestReviewWrite; the package
READMEs describe the GitHub mutations and their measured behavior, the
thread actions, and the pending review flow. A new Agent Note records
moving pull request comment drafts to the code host's pending review,
and the forge panel note links it. The Web e2e replies to, edits,
deletes, and resolves threads, and starts, submits, and discards a
pending review through the scripted gh double.
2026-11-09 11:38:56 +08:00
Yichen Jiang e7e5ed638d Merge remote-tracking branch 'origin/master' into feat/forge-review-threads
# Conflicts:
#	apps/web/tests/forge-panel.e2e.ts
#	packages/experimental/client-ui-forge/README.i18n.yaml
#	packages/experimental/client-ui-forge/src/client/DiffFile.tsx
#	scripts/type-equiv.manifest.json
2026-11-09 11:38:53 +08:00
Kexin Liu 6a20990684 Merge pull request #7933 from deepseek-harness/feat/session-append-events
feat(session): accept multiple events through append
2026-11-08 21:52:12 +08:00
Kexin Liu 2c37d513ed Merge pull request #8085 from deepseek-harness/agent/answer-concurrent-tools-2-fda34beac313
feat(tools): declare read-only tools concurrency-safe
2026-11-08 17:13:38 +08:00
liukx0205 7f6b7ba4b6 refactor(session): narrow batch publication metadata and validation 2026-11-07 17:02:19 +08:00
liukx0205 1e7f265cfd fix(session): complete batch API review coverage and declarations 2026-11-07 17:02:19 +08:00
liukx0205 143fabbc34 fix(session): share live append validation and expose batch publication 2026-11-07 17:02:19 +08:00
liukx0205 9d18b68095 fix(session): assign batch sequences after payload snapshots 2026-11-07 17:02:19 +08:00
liukx0205 b3d570cdf8 feat(session): accept positional event tuples for batch append 2026-11-07 17:02:19 +08:00
liukx0205 748c53b865 feat(session): accept single event objects through append 2026-11-07 17:02:19 +08:00
liukx0205 b5969f9e8d feat(session): accept multiple events through append 2026-11-07 17:02:19 +08:00
Kexin Liu e49a941d65 Merge pull request #7910 from deepseek-harness/fix/agent-input-lifecycle
fix(agent): drain work after cancellation failures
2026-11-07 17:02:15 +08:00
Yichen Jiang 9b456b4e93 perf(forge): compare from the recorded base while confirming the head
Status snapshots carry the base commit the host recorded with the head
(baseSha). ForgeService.diff passes it to the provider when the latest
status names the requested head, and forge-github starts the compare read
from it while its GraphQL read confirms the head; a stale head or another
base cancels and joins the early read.
2026-11-07 16:22:24 +08:00
Yichen Jiang f11daf3ad3 docs(forge): document review thread reads, the thread overlay, and the unsupported failure
The forge subsystem page documents ctx.forge.readReview, its thread
types, and the panel's changeRequestReview cache; the package READMEs
describe the GitHub thread mapping, the overlay and thread list, and the
new panel Config fields. An upgrade guide covers the unsupported
ForgeFailure code. The Web e2e overlays scripted threads, opens them from
the thread list, and sends a quoted thread through a replayed turn.
2026-11-07 14:31:30 +08:00
Yichen Jiang ef34a844de fix(forge-panel): refuse commit runs off the first-parent line and harden review state
- Shift-chosen commits must follow each other as first parents; the picker
  says why it keeps a run that leaves the line, and the Host refuses one with
  forgePanel/commits-not-linear for the Branch scope and change requests.
- The Branch scope lists its newest maxCommits commits with
  --no-show-signature, so log.showSignature cannot corrupt the listing.
- Viewed marks merge with the stored entry on each write and follow the
  storage event, so tabs of one Session keep each other's marks.
- .gitattributes patterns follow Git's wildmatch rules, checked against
  git check-attr; change digests include file mode lines.
- GitHub commit pagination refuses a repeated cursor or a changed count;
  commits validates the repository before its host like the other reads.
- The commit list is read once per head, base, or refresh; the lens menu
  hides the count of the disabled Tests lens.
2026-11-07 12:56:23 +08:00
Yichen Jiang fb637c2bfd Merge remote-tracking branch 'origin/master' into feat/forge-review-viewed-lenses
# Conflicts:
#	apps/web/tests/forge-panel.e2e.ts
#	docs/config-catalog.md
#	docs/config-catalog.zh.md
#	docs/subsystems/forge.i18n.yaml
#	docs/subsystems/forge.md
#	docs/subsystems/forge.zh.md
#	packages/experimental/client-ui-forge/README.i18n.yaml
#	packages/experimental/client-ui-forge/README.md
#	packages/experimental/client-ui-forge/README.zh.md
#	packages/experimental/client-ui-forge/src/client/ChangesTab.tsx
#	packages/experimental/client-ui-forge/src/index.ts
#	packages/experimental/client-ui-forge/tests/controller.host.spec.ts
#	packages/experimental/forge-github/README.i18n.yaml
#	packages/experimental/forge-github/README.md
#	packages/experimental/forge-github/README.zh.md
#	packages/experimental/forge-github/src/index.ts
#	packages/experimental/forge-github/src/provider.ts
#	packages/experimental/forge-github/tests/plugin.spec.ts
#	packages/experimental/forge-github/tests/provider.spec.ts
#	packages/experimental/forge-github/tests/review.spec.ts
#	packages/experimental/forge/README.i18n.yaml
#	packages/extensions/tool-cordis/src/api-catalog.ts
2026-11-07 12:13:56 +08:00
Yichen Jiang 0a10b6fd16 Merge pull request #8049 from deepseek-harness/fix/forge-pinned-diff
fix(forge): pin change request diffs to the polled head and count every review thread
2026-11-07 11:22:43 +08:00