Repository navigation
knowledge: drop default sidecar fields, keep case digests in records - #176
Merged
Merged
Conversation
…n records Skill sidecars no longer store empty sources/references or the shared applicability placeholder; readers already apply those defaults. Case documents link their typed record instead of repeating source digests, and the request metadata digest moves into the record. Contract tests guard both. Agent-assisted-by: Claude Code / claude-opus-5-5 [effort=max] (implementation, tests and documentation) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: David Chen <530634352@qq.com>
Generated with tools/build_skill_index.py from the preceding source commit. Agent-assisted-by: Claude Code / claude-opus-5-5 [effort=max] (generation and verification) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: David Chen <530634352@qq.com>
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
Skill evaluationPLANNED: execution is not authorized by knowing this hash. Selected comparisons: 1 · Estimated task calls: 2. Applicable only while these revisions and inputs match the current PR. Evidence and logs · Retention: 30 days. Pipeline evidence only; offline fixed outputs do not establish skill efficacy. Plan bindings, budget and frozen examPlan SHA-256: Model/settings:
Frozen exam hashes: Uncovered / unsupported paths (123)
|
…ieval-sidecars Signed-off-by: David Chen <530634352@qq.com> # Conflicts: # src/inferrsi/data/repo_packs/inferrsi/knowledge/README.md # src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-aclgraph-capture-replay.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-device-health.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-distributed-parallel.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-graph-mode.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-model-adaptation.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-operator-profiling.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-operator-tuning.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-optimization-casebook.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-performance-analysis.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascendc-tiling-design.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/cann-attribution-overlap.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/torchnpu-device-memory-lifecycle.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/torchnpu-pinned-host-memory-lifecycle.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/assess-serving-preemption.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/assess-split-backward-schedule.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-h3-runtime-residency.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-llama-runtime-state.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-sd-cpp-residency.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-sglang-diffusion-lifetimes.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-vllm-kv-lifetimes.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/bound-windowed-attention-replay.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/budget-speculative-verification.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/calibrate-serving-simulation.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/cgroup-resource-controls.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/checkpoint-recovery-consistency.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/choose-parallel-training-plan.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/cite-check-downstream-pr.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/claim-contract.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/comfy-ui.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/competitor-scan.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/dcp-async-save-performance.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/dense-performance-evidence.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/diagnose-vllm-admission-pressure.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/diffusers-invocation-state.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/draft-acceptance-issue.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/evaluate-elastic-serving-placement.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/generation-cache-lifecycle.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/host-memory-management.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/host-overhead-optimization.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/inference-benchmark-design.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/inference-optimization-casebook.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/kernel-writing-guide.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/kv-tier-management.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/lazy-evaluation-stream-lifetimes.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/npu-skill-factory.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/optimization-search-policy.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/os-process-placement.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/performance-ceiling-analysis.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/plan-activation-residency.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/repository-evolution-analysis.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/rsi-evidence.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/skill-evolution-playbook.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/storage-io-measurement.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/training-input-pipeline.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/training-resource-budget.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/weight-loading-optimization.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/audit-afd-p2p-exchanges.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/diagnose-attention-bottlenecks.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/host-overhead-profiling.json # src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/reason-hopper-kernels.json
Generated with tools/build_skill_index.py from the preceding merge commit. Agent-assisted-by: Claude Code / claude-opus-5-5 [effort=max] (generation and verification) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: David Chen <530634352@qq.com>
Dong1017
reviewed
Oct 27, 2026
Dong1017
left a comment
Dong1017
left a comment
Collaborator
There was a problem hiding this comment.
Reviewed head 9e8996c88ed277912ff2c42076738fe4dc0d2884, base 38d5883b8e72e352865e214194cb22dc99253dbd. One reader-facing P2 remains inline; no unconditional LGTM.
The 53 sidecar edits are limited to the stated empty/default removals. Unchanged reader fallbacks preserve the intended defaults; the 40 applicability placeholders deliberately become {} and therefore change lexical input as disclosed. The moved request-metadata digest matches its former Markdown value and retains reported/not-rehashed limitations. No source-efficacy claim is inferred.
All four Ready jobs, lint, declaration policy and plan/publish pass; execute is skipped. The green source-tree contract test does not cover the standalone case-bundle link failure below.
Two smaller consistency notes: (P3) qwen-ascend-rejected-directions.md still repeats the standalone 40-character upstream head from its JSON record; the new digest regex catches only 64-character values. Remove the redundant prose identity or explicitly narrow the ownership rule, while retaining required pinned source URLs. Also the PR body cites generation source ad90f01, but actual README/projections pin cd01523c030c5ea5a44326be61b63b8cec502b7a; inspected source trees match, so this is evidence wording, not stale generation.
Static source/diff/reader/test and remote-CI review only; no local PR-code or model/hardware execution. #175 remains Proposed without accepted scope/acceptance ownership; this review neither grants that decision nor merges/closes anything.
Case documents now link their typed record, but read bundles stored the record only as case.json, so that link failed through resource(). Each case declares its own JSON basename in required_resources, and a test resolves every local case-document link through KnowledgeReader at the returned revision. The Ascend case no longer repeats its upstream head outside the record. Agent-assisted-by: Claude Code / claude-opus-5-5 [effort=max] (implementation and tests) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: David Chen <530634352@qq.com>
Generated with tools/build_skill_index.py from f321699. Agent-assisted-by: Claude Code / claude-opus-5-5 [effort=max] (generation and verification) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: David Chen <530634352@qq.com>
10 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Governing Issue and delivery slice
Refs #175
limitationsorprovenance_readsets.Knowledge authorship declaration
Changed artifacts and evidence
skill_library/**/metadata.json: 32 emptysources, 49 emptyreferences, 40 placeholderapplicabilityremoved (deletions only)skill_metadatafor all 60 skills compared against6b13cf0: only those 40applicabilityvalues change, placeholder →{}. Newtest_local_sidecars_omit_fields_that_equal_reader_defaultspassesadmission-drain-occupancy.jsongainsrequest_metadata_sha256; its MD drops both hashes and the Kind/Origin/State line; the other three case MDs link their recordtest_case_documents_cite_their_record_without_copying_digestspasses; the old admission MD contains 2 digests and would fail itrepo_packs/inferrsi/knowledge/**, last generated fromf321699b42a490276ffa4ea52350591683b575cc(earlier fromad90f01, then mergecd01523)python tools/build_skill_index.py --source-revision f321699 --check: current, 60 recordsdocs/knowledge-base.mdfield-ownership table and normalization paragraphrequired_resourcestest_case_document_links_resolve_inside_the_delivered_bundlereads every case viaKnowledgeReaderand resolves each local link withresource()at the returned revision; fails before the fix withUnknown resourceupstream_head; check covers 40- and 64-hex outside URLsBundle content revisions of the touched skills and cases change, as with any content edit. Collect lexical scoring no longer sees the shared placeholder text for those 40 skills.
Checks and dependencies
main(38d5883) and regenerating the index from the merge commit:ruffpass,build_skill_index.py --checkcurrent, all knowledge/skill/collect test modules pass.ruff check src tests tools: pass.pytest tests/test_knowledge_contract.py: pass. Fullpytestlocally (Windows, Python 3.12): 16 failures, the identical set to an unmodified6b13cf0baseline run in the same environment (test_knowledge_install,test_knowledge_policy_*,test_harness_improvementCRLF case,test_module_boundaries,test_qwen21_*,test_server_processes), so none is introduced here.f321699b42a490276ffa4ea52350591683b575cc, consistent with the existing generator practice.Remaining work and Issue closure
🤖 Generated with Claude Code