[Dong1017]

Scope

Refs #253 and the accepted #215 planning/advisory boundary. Make nonpaid PLANNED feedback useful to PR authors without changing task selection or execution policy.

• Lead with a short next-action/blocker/evidence checklist. Keep commands and task-specific instructions in Author verification steps, with raw bindings and paths separately collapsed; unexecuted plans no longer show zero-valued pseudo-results.
• Retain actual path-map triggers, existing task purpose/scope and required outputs. Explain that a path match is not a task-relevance verdict, and expose unsupported snapshot constraints without weakening them.
• For applicable canonical/index changes, derive the existing read-only index check's explicit source pin from bounded Git metadata and verify canonical tree consistency before displaying it. Never substitute HEAD, execute PR code, invent a mapping or present a paid command as a nonpaid fallback.
• Preserve every raw coverage path and existing evidence/head/admission boundaries; path counts are not claims. Ask for command/input/exit/output/failure receipts and independent verified/refuted/unverified disposition.

Verification and limits

Final local verification: 249 affected tests passed; whole-tree Ruff and diff checks passed. A prior fixed snapshot before the final safe-reason display projection passed 3,257 tests, 12 skipped, 284 subtests; that aggregate is not claimed for the final head. Independent review covered the display boundary and the final history-stability correction, including 38 focused tests. Normal current-head repository CI remains required.

Data-only previews reuse the exact retained #90/#263 plans plus independently fetched public source metadata: both gain an actual pinned index-check command; #263 also names its snapshot rejection and the diagnostic task's purpose/required outputs. The compact front screen is about 1.1k characters, with the full audit retained. These previews are not executed index checks, a refresh of historical bot comments, re-bound trial evidence, a real author-agent loop or measured skill benefit. No model/provider/hardware execution; no mapping, grader, budget, credential or acceptance-gate changes. The only workflow change aligns plan checkout to the existing publisher with fetch-depth: 0; execute, permissions and credentials are unchanged. #253's remaining actual author evidence/disposition requirements stay open.

Native-review correction

The native P2 identified that shallow planning and full-history publication could derive different source-pin metadata/fingerprints. The correction aligns both history depths and restricts automatic command suggestions to the frozen head’s ancestor closure; unrelated cached/missing objects now give the same explicit unavailable result. This is display eligibility, not a failed source check or a new acceptance gate. Real local file-transport regressions cover both shallow/full parity and deletion of an unrelated source branch. Original evidence remains unchanged; current-head CI is still required.

Authorship

Actual implementation and independent review: Codex / gpt-6-astra / xhigh.

{"schema_version":1,"head_sha":"297835b34d22add47a1a458cdb883f0d67f2c939","human_only":false,"material_ai_authors":[{"tool":"Codex (nonpaid SkillEval author guidance)","model_id":"gpt-6-astra","effort":"xhigh","role":"author"}]}

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-31T14:51:25.503917Z	297835b	New commits

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[github-actions]

Skill evaluation

Automatic execution / cumulative PR budget: {"reason":"Trusted configuration or pricing missing, invalid or expired","remaining_usd":"unknown","reserved_or_unresolved_usd":"unknown","spent_or_legacy_usd":"unknown","state":"blocked"}

Advisory: obtain missing evidence

Retained evidence availability: planned / unexecuted. No approval or merge authorization is issued.

• Retained evidence is planned / unexecuted.
• No supported comparison cases were selected.
• Uncovered/unsupported scope needs a case or explicit reviewer disposition.

PLANNED: execution is not authorized by knowing this hash.
Outcomes: {"fix":0,"inconclusive":0,"missing_pair":0,"regression":0,"unchanged_correct":0,"unchanged_incorrect":0}
Earlier revision results are stale for this plan.

Selected comparisons: 0 · Estimated task calls: 0.
Uncovered/unsupported changes: 6. Gaps never count as passes.

Applicable only while these revisions and inputs match the current PR.
Paid execution requires trusted label authorization and cumulative PR reservation. Unknown accounting blocks new paid calls; planning is not execution.
Pipeline evidence only; offline fixed outputs do not establish skill efficacy.
Evidence and logs · Retention: 30 days.
Outcomes are historical; verifiable only before Unix time 1796050213 while the linked artifact is available. Expired evidence is unverified.

Plan bindings, budget, frozen exam and advisory details

Unchanged-correct is valid neutral evidence, not incremental benefit.
Content/source review and authorship admission remain separate.
Trusted regrading checks retained answers, not model/hardware provenance or a hostile executor; this is not an efficacy certification.

Per-case outcomes (discovery excluded; regressions and gaps first):

[]

Showing at most 20 of 0 pairs; the evidence link contains all outcomes, including omitted regressions/gaps.

Plan SHA-256: 5af427703adbbf55ab6f180ec097bff14e78a532f9127391bdc856ba9f2d1829
Base: 33de003e2b613b25a06b22d61797ddf06b07afe3 · Head: 297835b34d22add47a1a458cdb883f0d67f2c939
Trusted implementation: 33de003e2b613b25a06b22d61797ddf06b07afe3

Model/settings: {"effort":"not_applicable","name":"claude-opus-5-5"}
Budget: {"comparisons":0,"estimated_calls":0,"limits_per_comparison":{"campaign_seconds":600,"max_task_calls":2,"repeats":1,"timeout_seconds":180,"token_stop_new_trials":200000},"provider_config_sha256":"b28d80f8194fbe0f5a9f01515efb24b218979fd45f702776560da276f4b30d04","reservation":null}

Skill	Comparison	Cases	Estimated calls

Provider/settings: {"client":{"api_version":"2023-06-01","endpoint":"https://yibuapi.com/v1/messages","kind":"anthropic-messages-full-context-v1","max_output_tokens":4096,"max_read_bytes":32768,"max_request_bytes":262144,"max_requests":1,"max_tool_calls":12,"provider":"yibuapi"},"limits":{"token_stop_new_trials":200000},"model":{"effort":"not_applicable","name":"claude-opus-5-5"},"pricing":{"currency":"USD","expires_at":1793271228,"group":"default","max_input_tokens":1000000,"model":"claude-opus-5-5","rate_evidence":"https://github.com/InferMatrixLab/InferRSI/issues/215#issuecomment-6467531341","rates":{"cache_read":"0.4","cache_write":"10","input":"8","output":"40"},"request_fee":"0","unit":"per_million_tokens"},"spend_limit":{"amount":"200","currency":"USD"}}

Uncovered / unsupported paths (6)

• uncovered: ".github/workflows/skill-eval-pr.yml"
• uncovered: "docs/skill-eval-pr.md"
• uncovered: "tests/test_skill_eval_planned_guidance.py"
• uncovered: "tests/test_skill_eval_pr.py"
• uncovered: "tests/test_skill_eval_pr_summary.py"
• uncovered: "tools/skill_eval_pr.py"

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 52971cc818

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Make projection-check metadata independent of checkout depth

When the declared source pin predates the base checkout, this lookup produces different plans in the existing workflow's jobs: .github/workflows/skill-eval-pr.yml uses a default shallow checkout for plan, but fetch-depth: 0 for publish. The fetch in main() does not remove the shallow boundary, so the planner records projection_check.status = unavailable while the publisher rebuilds it as ready for identical PR revisions. I reproduced the missing ancestor with the workflow-equivalent fetch. Because this metadata enters the plan fingerprint, source_plan_expiry() rejects the original execution plan and retained results become invalid rather than publishable. Ensure both jobs have the same source objects before fingerprinting, or keep availability-dependent display metadata outside the execution binding; git fetch -h documents --unshallow as “convert to a complete repository.”

Useful? React with 👍 / 👎.

[Dong1017]

Scoped software LGTM for 297835b34d22add47a1a458cdb883f0d67f2c939 against 33de003e; no remaining substantive finding in the six-file change.

The guidance now gives a pinned, read-only index-check command where justified, identifies mapped task purpose/trigger/required outputs, and asks for explicit evidence and unverified-claim dispositions. Generated path counts are not presented as independent claims; unsupported snapshots and absent metadata remain explicit.

Prior P2 disposition: 4432063862 is addressed by matching full-history plan/publish checkouts and restricting automatic source-pin suggestions to the frozen head’s ancestor closure. I independently reproduced both the shallow/full mismatch and an unrelated-ref availability variant; the new regressions establish equal plans/fingerprints after correction. Missing/unrelated pins remain unavailable—not silently ready. The historical thread is outdated but still unresolved in the UI; this is a substantive disposition, not a claim that its UI state changed.

Evidence and limits

• Published tree 12daa787 and all six blobs match the independently reviewed local source. Final independent focused checks: 38 passed; author’s affected suite: 249 passed.
• Ready 41211932585: all five jobs succeeded. Latest declarations passed; current-head native review reports no findings. Corrected body/head declaration was read back.
• Reviewed bounded Git-data reads, command construction, escaped metadata and fixed public error categories; raw historical plans remain unchanged. No mapping, grader, paid gate or snapshot-safety relaxation.
• The #90/#263 examples are data-only previews, not executed checks or rebinding of historical results. The current bot comment still uses trusted base 33de003e, not this undeployed renderer. This clears software guidance—not an actual author/agent loop, task applicability for every claim, or efficacy; #253’s remaining evidence/disposition acceptance stays open.

COMMENT review only; no formal approval or experiment authorization.