Commit Graph
1308 Commits
Author SHA1 Message Date
Wenlu Wang 61e46f077e Merge pull request #1362 from deepseek-harness/fix/landlock-runner-failure-classification
Fix false sandbox failures from Landlock partial-enforcement notices
2026-08-05 13:59:47 +08:00
kingwl 75bd2b8aaa Merge remote-tracking branch 'origin/master' into fix/landlock-runner-failure-classification
# Conflicts:
#	.agents/notes/implemented/feature/2026-07-06-sandbox.i18n.yaml
#	.agents/notes/implemented/feature/2026-07-06-sandbox.zh.md
#	docs/core-data-structures/sandbox.i18n.yaml
#	docs/core-data-structures/sandbox.zh.md
#	docs/postmortem/README.i18n.yaml
#	docs/postmortem/README.zh.md
#	examples/acp-agent/tests/acp.snapshot.ts
#	native/landlock-run/README.i18n.yaml
#	packages/sandbox/sandbox-local/README.i18n.yaml
#	packages/sandbox/sandbox-local/README.zh.md
#	packages/sandbox/sandbox/README.i18n.yaml
2026-08-05 13:17:38 +08:00
imccyu e61839a6ab fix(session): refresh Session.create artifacts 2026-08-05 11:57:17 +08:00
xjt 7bd111b2af Merge origin/master into xjt/proofreading-active-docs-2-apply 2026-08-05 10:57:53 +08:00
Huanqi Cao 39b0695d48 Merge remote-tracking branch 'origin/master' into feat/pwsh-tool
# Conflicts:
#	apps/cli/composition.md
#	docs/testing.i18n.yaml
#	docs/testing.md
#	docs/testing.zh.md
2026-08-04 17:54:45 +08:00
xjt 2db712eec7 docs(i18n): proofread active Chinese documentation 2026-08-04 17:36:14 +08:00
Hypatia May 293fac9421 fix(sandbox): round 5 harden spawn attribution 2026-08-04 16:06:54 +08:00
Hypatia May ec4a070f0e Merge remote-tracking branch 'origin/master' into fix/landlock-runner-failure-classification 2026-08-04 14:47:58 +08:00
Hypatia May 96ba98c99e fix(sandbox): require runner-specific spawn evidence 2026-08-04 14:47:46 +08:00
Turtle 64f9a83bd6 Merge pull request #1369 from deepseek-harness/codex/remove-dsh-tui-entrypoint
cleanup: remove TUI package and legacy dsh entrypoints
2026-08-04 14:11:37 +08:00
Hypatia May 0cd1611023 Merge remote-tracking branch 'origin/master' into fix/landlock-runner-failure-classification 2026-08-04 13:54:37 +08:00
Hypatia May b0f33f13f4 docs(sandbox): cross-link Landlock snapshot rule 2026-08-04 13:49:34 +08:00
Hypatia May f91aedd074 refactor(sandbox): defer native protocol exports (round 3) 2026-08-04 13:37:33 +08:00
Turtle 5fc420c5c0 test(acp-snapshot): wait for the durable goal pause before disposal
The goal scenario asserted a paused revision-2 goal, but the pause is
appended only after cancellation reaches idle — after turn/end. Under
parallel snapshot files the subprocess could dispose before the pause
record persisted, folding the log to an active revision-1 goal.

Add a waitForEventAfterTurnEnd input step (the turn-end/title waiters'
shape, parameterized by event type) and use it in the goal scenario to
hold the subprocess open until the goal-state record lands.
2026-08-04 13:25:39 +08:00
Turtle 10bb9cbf4a cleanup: remove TUI package and legacy dsh entrypoints 2026-08-04 13:20:28 +08:00
Hypatia May e36d040d0a fix(sandbox): spawn confined argv directly (round 1) 2026-08-04 12:04:48 +08:00
Hypatia May 2eedb54849 fix(examples): declare sandbox fixture dependency 2026-08-04 09:37:50 +08:00
Hypatia May 71aafeb858 Merge origin/master into fix/landlock-runner-failure-classification 2026-08-04 09:14:13 +08:00
Huanqi Cao 3d1166fcdd feat(bash): shell tools reject a mismatched executor dialect at load
The seam gains ShellDialect ('bash' | 'powershell' - concrete shells, not
families: zsh or fish would be their own values, never 'bash'); bash-local
declares bash (bash-sandbox inherits), pwsh-local declares powershell, and
both tools throw at load when the mounted executor speaks another dialect -
previously tool-pwsh over bash-local handed PowerShell text to bash -c and
the deployment error surfaced as ordinary nonzero exits. Pinned by mismatch
tests on both tools; the parity note records the contract (both languages).

Also from the review round: the tool-bash README's managed-environment
section becomes a summary linking the owning dsh-bash-env contract (the
duplicated prose carried a stale owner in its example import), the
pwshOnly JSDoc drops the stale 'on PATH' phrasing, and the task-tools
contract comment in the two pwsh compositions is indented into its block.
2026-08-03 22:47:53 +08:00
Huanqi Cao ffeccc138b Merge branch 'feat/ripgrep-packaged-binary' into feat/pwsh-tool
# Conflicts:
#	docs/module-graph.md
2026-08-03 20:22:16 +08:00
Huanqi Cao 0ee198d70e Merge remote-tracking branch 'origin/master' into feat/ripgrep-packaged-binary 2026-08-03 20:19:49 +08:00
Hypatia May 6343d8f6e6 fix(sandbox): evidence-gate runner failures (round 1) 2026-08-03 17:31:08 +08:00
Huanqi Cao 37bceb7dad Merge branch 'feat/ripgrep-packaged-binary' into feat/pwsh-tool 2026-08-02 22:29:44 +08:00
ZiyaZhang e72978ba98 fix(tool-goal): scope the wrap-up no-more-tools clause to the current run
The injected instruction persists as a durable user message, so an
unscoped 'Do not call any more tools.' stays in every later request's
history. Scope it: '...in this run; further work waits for the user's
next instruction.' A/B probes on deepseek-v4-pro show the scoped wording
is non-inferior in-turn (4/4 zero tool calls, closing quality unchanged)
and next-turn tool use is unaffected under both wordings; the scoped
form states the instruction's actual lifetime.
2026-08-02 07:29:18 -07:00
Huanqi Cao ffb7d681f4 Merge remote-tracking branch 'origin/master' into feat/ripgrep-packaged-binary 2026-08-02 22:28:21 +08:00
Huanqi Cao b2838fb2be fix(pwsh): close the remaining review threads
- scrubbedParentEnv folds case before the DSH_ prefix check (Windows env
  names are case-insensitive; a parent dsh_* entry read back as \*
  in the child) and the service spec pins the lowercase probe.
- The acp.snapshot.ts pwsh probe follows resolvePwshPath() like the package
  suites, so a Windows host with only an install-location pwsh still runs
  the scenario.
- pwsh-tool-turn is re-recorded around [Console]::Out.Write('PWSH_OK'):
  the fixture carries no platform newline, so one recording replays on
  Windows and POSIX alike (record + refresh; replay-verified keyless).
- The pwsh-local Known Limitations bullet drops the self-defeating no-op
  advice: & { } is scoped to param(...), using/#requires scripts run from a
  file (both languages, pairing re-recorded).
- The capability-seams graph moves ctx.bashEnv ownership to bash-env and
  lists pwsh-local/tool-pwsh on the ctx.bash seam (source updated,
  docs regenerated).
- The tool-bash presenter fixture retires the stale 'command aborted'
  literal for the shipped 'tool call aborted' message.
2026-08-02 21:46:53 +08:00
Huanqi Cao 7b7525c091 fix(fixtures): repair the cancel-fixture abort error placement; exempt pwsh-local coverage on pwsh-less hosts
The hand-edit in 348ab41151 put the seq-17 abort error inside the tool-result
message and dropped a closing brace, breaking every JSONL consumer; tool-calls.ts
appends 'error' as a data-level sibling of 'message' (the seq-19 shape), so the
fixture now matches the emitter.

Coverage: pwsh-local's executor suites self-skip without a real pwsh, which left
per-file 100% unreachable on pwsh-less contributor hosts (mirror of the existing
windowsCoverageExclusions contract). A PATH-only probe exempts only
pwsh-local/src/index.ts; CI runners ship pwsh and still enforce the full bar.
docs/testing.md (+zh, pairing re-recorded) names the prerequisite; the testing.md
budget rises 1120->1150 because the coverage-gate contract genuinely grew.
2026-08-02 21:28:22 +08:00
ZiyaZhang 666ef95f81 fix(tool-goal): let the model deliver a wrap-up message after a goal-round complete/blocked
A goal round reporting complete or blocked used to conclude the turn at
the tool result, so the model never spoke after the call and sessions
ended on a bare update_goal card. The terminal update now defers one
plugin-sourced <goal_complete>/<goal_blocked> instruction onto its
result asking for a grounded closing message without further tool
calls; the turn then ends through the ordinary no-tool-calls stop.
Direct-human mutations stay uninstructed. Wording chosen by A/B
sampling on deepseek-v4-pro; one extra request per goal lifecycle.

New keyless ACP snapshot goal-wrapup drives the shipped app through
create -> round one -> autonomous complete and pins the injection, the
same-turn closing message, and the completed turn end.
2026-08-02 05:58:45 -07:00
Huanqi Cao 6ae0f78c2a fix(tools): pin the abort error info in the cancel fixture, close the jscpd mirror block, and restore the pwsh binary ignore 2026-08-02 20:27:06 +08:00
Huanqi Cao ed39ff096f test(fixtures): pin the TOOL_ABORTED message and keep the pwsh fixture replayable; restore trimmed testing.md clauses 2026-08-02 20:07:26 +08:00
Huanqi Cao c376802f44 fix(acp-snapshot): skip pwshOnly scenarios without pwsh and mount task tools in the pwsh composition 2026-08-02 19:37:09 +08:00
imccyu 53531b9af2 test(snapshot): refresh Cordis steering API fixture 2026-08-02 17:53:07 +08:00
Huanqi Cao eeffd2cfce fix(headless): mount bash-env in the code-mode harnesses 2026-08-02 17:35:46 +08:00
Huanqi Cao 7201a417c2 fix(subagent): mount bash-env in the spawn and headless e2e harnesses 2026-08-02 17:25:47 +08:00
Huanqi Cao a79d7c896c fix(tool-bash): mount bash-env in the sandboxed and HMR-safety harnesses; normalize the pwsh-tool-turn fixture line endings for Linux replay 2026-08-02 16:53:29 +08:00
Huanqi Cao 1361e9718e Merge remote-tracking branch 'origin/feat/ripgrep-packaged-binary' into feat/pwsh-tool
# Conflicts:
#	apps/cli/package.json
#	packages/core/tools/tests/gen-tool-catalog.spec.ts
2026-08-02 16:22:24 +08:00
Huanqi Cao 782894ff8c Merge remote-tracking branch 'origin/master' into feat/ripgrep-packaged-binary 2026-08-02 16:18:46 +08:00
Huanqi Cao 64f9e68bd9 test(acp): register and record the keyed pwsh-tool-turn snapshot scenario 2026-08-02 15:31:52 +08:00
Huanqi Cao 96cf8a2fbc chore(compositions): mount bash-env and the pwsh tool in shipped and demo compositions 2026-08-02 14:18:42 +08:00
Dudu-0223 431fb4b035 feat(subagent): add explicit child reports 2026-08-02 12:51:10 +08:00
imccyu d950150cdc fix(persistence): preserve subagent session origin 2026-08-02 12:51:09 +08:00
imccyu b8be3583e0 test(cordis): refresh session origin API snapshot 2026-08-02 12:51:09 +08:00
Dudu-0223 a977ef30ee fix(subagent): preserve published run failures 2026-08-02 12:51:09 +08:00
Dudu-0223 8b0a7a5d8d fix(subagent): close stacked catalog integration gaps 2026-08-02 12:51:09 +08:00
Dudu-0223 a967f5174a test(snapshot): update unlabeled workflow children 2026-08-02 12:51:09 +08:00
Dudu-0223 774ee34b9a feat(subagent): catalog one-shot child sessions 2026-08-02 12:51:09 +08:00
Dudu-0223 4bd98407a9 fix: address codex review round 1
- listChildren() takes an optional AbortSignal and rechecks it after every
  un-signalled session-query await (the cold-resume cooperative-cancellation
  boundary); list_agents forwards exec.signal so the registry's drain of
  started tool bodies cannot block on a slow or large catalog.
- The list_agents description now presents running/complete as a stored-record
  snapshot and defers deliverability to send_message, matching the ownership-
  conflict semantics the service tests pin.
2026-08-02 12:51:09 +08:00
Dudu-0223 4240c7dd7b feat(subagent): durable child catalog and list_agents
Implements the durable-subagent-catalog RFC: SubagentControlService.listChildren()
enumerates a parent's direct continuable children from one sessionQuery trace,
validates each child's sole subagent/descriptor event (now carrying the durable
creation label), and returns one ordered SubagentListEntry[] with per-child
corrupt/unsupported/unavailable diagnostics. The list_agents tool ships as a
separately loadable plugin of dsh-tool-subagent-control requiring sessionQuery
at load; send_message stays usable without it.
2026-08-02 12:51:09 +08:00
imccyu 4d0a24d8ed test(subagent): preserve durability failure after rebase 2026-08-02 12:51:08 +08:00
Dudu-0223 7428cdf41e fix(subagent): address codex review round 3
- Make host-user authority unforgeable. `{ kind: 'user' }` was a bare
  discriminant, so any plugin holding `ctx.subagents` — including
  model-generated cordis_mount code, which the advanced ACP composition ships
  alongside continuable subagents — could construct it and skip the
  direct-parent check for any known child id. It now carries an opaque grant
  that only SubagentService.userAuthority() mints, which composition hands to
  trusted host adapters; a model-facing tool uses parent authority from its own
  execution context.
- Reconcile a delivery discarded inside its own admission window. An enqueue
  listener that cancels fires the discard before followup() returns, so the
  discard listener could not clear an id it had not seen; submit() retained it
  and residency stayed `running` until an explicit drain.
- Recheck the caller signal after materialization. An abort landing between
  publication and inbox acceptance still submitted the prompt and returned both
  ids; it now rolls the child back.
- Stop promising the model transcript access that no shipped continuable config
  mounts. The tools now state only that a background child does not report back.
- Restate the implemented note as shipped state rather than a proposal, so it
  works as current authority.
2026-08-02 12:51:08 +08:00