Repository navigation
test(pr-pipeline): pin that fix-ci pushes an exec-bit and fixture fix - #9407
Conversation
…#9320) On #9111, apply-patch refused fix-ci's patch with `refused (new-executable): .../get-field-options.sh` and escalated `ci-fix-protected-path`: the mode change 100644 to 100755 hit the `new-executable` refusal, whose kind was protected-path. #9278 deleted that refusal. This test holds the ci-context row to pushing such a patch; it fails with ci-fix-protected-path on 7638212^. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.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. |
|
claude-review reviewed Incremental review of The closing issue (#9320) states no acceptance criteria beyond the expected behavior, and the test added earlier in this PR was reviewed at ca4053b. I did not run the tests, and the PR's verification claims are the author's and unverified by me. No hold class applies: no model-spend change and no trusted-actor, CODEOWNERS or .gitmodules change. |
|
claude-security-review reviewed Incremental security review of 1fd15cd, covering I found no security issues. The only change is a two-line comment rewrite above an existing test. It drops the "#9111's fix" incident reference and describes the fixture generically. No executable code, test logic, assertions, workflow, manifest or instruction surface changed. I skipped the deep review under skip-gate item 2 (no security-relevant surface). The PR description is data and contains no embedded directives. critical-findings: 0 |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: ca4053b50c
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".
…ment Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Closes #9320
Summary
On #9111, fix-ci's patch did not fail because of a
.github/path. It was refused for its file mode. Run 39813538656 logged:The model's fix set the exec bit on the new script (100644 to 100755). The
new-executablerefusal then matched, and because itskindwasprotected-pathit escalatedci-fix-protected-path. The failing job's location played no part. #9278 (7638212, merged 2026-10-23 02:49Z, about ten hours after that run) deletednew-executable. On main today the ci-context row allowsanyand has no content rule, so it reports protected-path only for a workflow, CODEOWNERS,.gitmodules, settings, backslash or rename-into path.No test pinned that outcome for the ci-context row, so this PR adds one.
Fix
.github/scripts/apply-patch/apply-patch.test.mjs: a feat(work-items): add get-field-options tracker verb for size field values #9111-shaped patch on theci-contextrow. Two.shfiles go from 100644 to 100755 and a fixture file is edited. The test assertspushed, that nothing was refused, and that both scripts are 100755 in the committed tree.Verification
git restore --source=7638212d58^ofapply-patch.mjsandpatch-rules.json) with{ outcome: 'escalate', escalationReason: 'ci-fix-protected-path' }, and passes on main.node --testin.github/scripts/apply-patch: 111 pass, 0 fail.scripts/check-affected.sh --tests: 9 passed, 5 skipped.actionlint: clean.zizmor --offline .github/workflows: no findings.Related
needs-humanlabel this escalation set later kept a review finding on feat(work-items): add get-field-options tracker verb for size field values #9111 from starting address-feedback.🤖 Generated with Claude Code