Commit Graph
3310 Commits
Author SHA1 Message Date
Tianyi Cui 38c2b5a6ec docs(rfc): preserve SDK telemetry scope 2026-07-19 21:49:41 +08:00
Yichen Jiang 4c0bdc7180 Merge remote-tracking branch 'origin/master' into worktree/acp-default-fs-suite 2026-07-19 21:49:06 +08:00
Yichen Jiang 299059665d Merge branch 'master' into worktree/pi-ai-manual-e2e 2026-07-19 21:47:04 +08:00
Yichen Jiang 43d036ca41 Merge remote-tracking branch 'origin/master' into worktree/dsbench-patch 2026-07-19 21:46:17 +08:00
Yichen Jiang 8a5792187b Merge branch 'master' into worktree/contain-result-observer-rejections 2026-07-19 21:40:22 +08:00
Tianyi Cui 012cd0469e Merge remote-tracking branch 'origin/master' into codex/reject-credentialed-search-redirects 2026-07-19 21:33:56 +08:00
Tianyi Cui f2d708511a Merge pull request #397 from deepseek-harness/codex/rename-golden-files-to-expected
Rename golden snapshot filenames to expected
2026-07-19 21:33:30 +08:00
Tianyi Cui 5e0e4b2401 fix(web): reject credentialed search redirects 2026-07-19 21:32:23 +08:00
Tianyi Cui b7cdad6360 ci: reject golden filenames on matching PRs 2026-07-19 21:24:42 +08:00
Tianyi Cui 2bc4791f84 Merge remote-tracking branch 'origin/master' into codex/pr370-review-20260719 2026-07-19 21:21:53 +08:00
Tianyi Cui fe5c3e2a7a docs(rfc): formalize SDK follow-up design 2026-07-19 21:21:39 +08:00
Tianyi Cui 546f2ae478 Merge remote-tracking branch 'origin/master' into codex/rename-golden-files-to-expected 2026-07-19 21:09:20 +08:00
Tianyi Cui f84092045c Merge pull request #396 from deepseek-harness/codex/package-readme-kv-cache-effect
docs: standardize package Model Experience fields
2026-07-19 21:01:30 +08:00
Tianyi Cui 9c28960b7f Merge remote-tracking branch 'origin/master' into codex/package-readme-kv-cache-effect 2026-07-19 20:52:56 +08:00
Tianyi Cui 27cb50ad46 Merge pull request #398 from deepseek-harness/codex/model-experience-field-headings
docs: structure Model Experience fields
2026-07-19 20:49:46 +08:00
Yichen Jiang d49808c6dd fix(tools): avoid retaining result executions 2026-07-19 20:14:32 +08:00
Yichen Jiang 55a66024fa fix: contain result observer rejections and correct bundle attribution 2026-07-19 18:38:35 +08:00
Yichen Jiang 07b15f014a Merge remote-tracking branch 'origin/master' into worktree/acp-default-fs-suite 2026-07-19 18:28:07 +08:00
Yichen Jiang 1cb63c7a18 fix(acp-agent): enable default filesystem suite 2026-07-19 18:27:44 +08:00
Tianyi Cui 1cd999b3fe Merge remote-tracking branch 'origin/master' into codex/pr353-retarget-master-20260719 2026-07-19 18:12:09 +08:00
Tianyi Cui 6e790b95f2 fix(compact): complete master retarget integration
Declare the tool-result pruning plugin in the examples workspace so the repl Cordis configuration resolves through plain Node and the Loader metadata gate.

Express the validated single-node surface rewrite without a non-null assertion or an unreachable defensive branch, preserving both the runtime contract and per-file 100% coverage.
2026-07-19 18:11:48 +08:00
Tianyi Cui a278374e8f docs: structure model experience fields 2026-07-19 18:08:42 +08:00
Tianyi Cui f756fcb7d3 Merge pull request #395 from deepseek-harness/codex/code-review-prevention
docs(agents): promote review principles
2026-07-19 18:00:09 +08:00
Tianyi Cui 6500de8402 Merge remote-tracking branch 'origin/master' into codex/pr353-retarget-master-20260719 2026-07-19 17:56:03 +08:00
Tianyi Cui 9c5b86051b Merge master into tool-result pruning
Retarget the pruning feature onto the merged compaction foundation.

Adapt content-only tool-result rewrite validation to the session-owned surface manager, migrate the demo wiring to repl-agent, and refresh generated type, catalog, and website contracts while preserving pressure and overflow pruning behavior.
2026-07-19 17:54:55 +08:00
Tianyi Cui 8733f365a4 Merge remote-tracking branch 'origin/master' into codex/code-review-prevention 2026-07-19 17:54:35 +08:00
Tianyi Cui 38b7275eb1 Merge pull request #393 from deepseek-harness/codex/code-review-human-principles
docs(review): distill human review principles
2026-07-19 17:53:10 +08:00
Tianyi Cui 2db4061861 Merge remote-tracking branch 'origin/master' into codex/code-review-human-principles 2026-07-19 17:47:06 +08:00
Tianyi Cui 0893aaa7cb test: rename golden snapshot files to expected 2026-07-19 17:45:49 +08:00
Yichen Jiang b0dc299508 Merge remote-tracking branch 'origin/master' into fix/subagent-depth-budget
# Conflicts:
#	examples/acp-agent/tests/acp.snapshot.ts
2026-07-19 17:42:52 +08:00
Tianyi Cui bcc920369c docs: document package KV cache effects 2026-07-19 17:39:50 +08:00
Tianyi Cui 4ab5a76317 docs(agents): promote review principles 2026-07-19 17:35:43 +08:00
Yichen Jiang 4a26ccb6f5 Merge branch 'fix/subagent-depth-budget' of github.com:deepseek-harness/deepseek-harness into fix/subagent-depth-budget 2026-07-19 17:34:40 +08:00
Yichen Jiang d231f30024 test(acp-snapshot): cover the child-omission branches synthetically
The childToolOmissions comparison arms ran only through the examples
snapshot suites, which the unit coverage gate does not count. Add an
authored child-omission scenario to the synthetic replay suite: one
scripted child omits the declared tool (header pin minus the omission,
prompt pin skipped), one keeps the full set (pin and prompt compared
verbatim), restoring 100% branch coverage on suite.ts.
2026-07-19 17:34:10 +08:00
Tianyi Cui b3c4b732d8 Merge pull request #332 from deepseek-harness/compact-post-step-overflow-recovery
feat(agent): add post-step compaction and bounded overflow recovery
2026-07-19 17:26:50 +08:00
Yichen Jiang d614374ac6 Merge branch 'master' into fix/subagent-depth-budget 2026-07-19 17:24:24 +08:00
Yichen Jiang cb74477d42 feat(tool-subagent): default maxDepth 1 with at-cap schema hiding
An omitted maxDepth meant unbounded recursion, and the shipped examples
shipped that default. maxDepth now defaults to 1; a numeric cap requires
the provider's depthLimit capability (the mount fails loud and points to
the explicit 'provider-managed' opt-out for out-of-process providers),
and a child AT the cap loses the delegating tool from its own schema via
the child toolFilter — prompt-face hiding on top of the execution-face
depth check. Examples pin maxDepth explicitly. The ACP snapshot harness
gains Scenario.childToolOmissions so a child session may legitimately
omit declared delegation tools from its pinned header and prompt;
affected subagent/workflow goldens are re-recorded.
2026-07-19 17:20:49 +08:00
Yichen Jiang 0d00106fa8 fix(session): persist the delegation depth in the session header
A subagent child's recursion depth lived only in runtime AgentOptions,
so a persisted child came back from resume counted as top-level and
maxDepth stopped binding after every restart. Add
SessionHeader.delegationDepth, round-trip it through the JSONL and
SQLite backends (SQLite schema v5), restore it on agent-loop resume,
and write it when the in-process backends create a child. The seam now
owns the shared depth vocabulary (delegationDepthOf): the persisted
header is authoritative and monotone — runtime options may deepen it
but never lower it.
2026-07-19 17:20:36 +08:00
Tianyi Cui d5350cd42e docs(review): distill human review principles 2026-07-19 17:08:00 +08:00
Tianyi Cui 20fd7133a0 Merge latest master into post-step recovery
Adopt initiator-derived private loop and session plumbing while preserving post-step recovery and balanced synthetic tool-call results after cancellation.

Keep the architecture map within its existing budget while documenting ambient Agent ownership and explicit cross-boundary identities.
2026-07-19 17:02:55 +08:00
Tianyi Cui b6858a5ce1 Merge pull request #384 from deepseek-harness/codex/simplify-agent-identity-plumbing
refactor(core): simplify AgentLoop identity plumbing with ctx.agents
2026-07-19 16:58:10 +08:00
Tianyi Cui a1de4d0071 docs: codify initiator-owned identity plumbing 2026-07-19 16:52:40 +08:00
Tianyi Cui f0d6637972 Merge latest origin/master into codex/simplify-agent-identity-plumbing 2026-07-19 16:52:40 +08:00
Tianyi Cui fd54b5c848 Merge latest master into post-step recovery
Combine agent initiator scoping and exact Cordis JSDoc inspection with post-step compaction and bounded request recovery.

Regenerate the Cordis and website API catalogs, and classify the recovery and compaction types required by the current catalog link-coverage gate.
2026-07-19 16:46:01 +08:00
Yichen Jiang 5f9fe14156 docs(bash-local): clarify spill cleanup semantics 2026-07-19 16:39:58 +08:00
Tianyi Cui bebbbbb7f0 Merge pull request #387 from deepseek-harness/codex/cordis-inspect-jsdoc
feat: inspect exact Cordis API JSDoc
2026-07-19 16:39:16 +08:00
Yichen Jiang e1bf278266 Merge branch 'master' into worktree/dsbench-patch 2026-07-19 16:36:01 +08:00
Tianyi Cui 5e2607df8c refactor(core): capture session inside loop operations 2026-07-19 16:28:31 +08:00
Tianyi Cui a72b86f82c Merge pull request #391 from deepseek-harness/codex/pr332-adapter-failure-provenance
fix(llm): scope adapter failures to model calls
2026-07-19 16:26:28 +08:00
Tianyi Cui 36dbe68631 test(cordis): snapshot targeted JSDoc inspection 2026-07-19 16:22:09 +08:00