Skip to content

fix(pr-pipeline): in-slot re-check honors a label added while the run waited - #18066

Merged
kyle-sexton merged 2 commits into
mainfrom
fix/18061-recheck-session-owned
Nov 21, 2026
Merged

kyle-sexton merged 2 commits into
mainfrom
fix/18061-recheck-session-owned

Conversation

@kyle-sexton

@kyle-sexton kyle-sexton commented Nov 21, 2026 •

Copy link
Copy Markdown
Member

Summary

The issue lane's in-slot re-check (check-blockers.mjs with RECHECK=true, run by pr-run-activity-read.yml check-broker) read only the issue's blocked_by list. With an empty list it proceeded without reading the issue, so a session-owned label added between admission and the slot was missed, and the lane worked an issue a local session owned.

Fix

The re-check now reads the issue first. When the issue carries a routing label (session-owned, the container label, a wayfind label) or provenance: untrusted, it takes the full check, which skips the run with the same notice admission writes. Otherwise it decides on the blocked_by list as before. Cost: one more GET of the issue per re-check.

The test's BASE_PATH now uses fileURLToPath; URL.pathname gave /D:/... on Windows and failed 25 of the suite's tests there.

Verification

  • node --test .github/scripts/work-resolve-issue/check-blockers.test.mjs: 38 pass, 0 fail.
  • Against main's check-blockers.mjs, the two new re-check tests (session-owned, untrusted) fail, as do the two whose expected reads changed.
  • node --test --test-name-pattern="recheck|check-broker|re-check" .github/actions/resolve-config/callers.test.mjs: 5 pass.

Related

Closes #18061

🤖 Generated with Claude Code

… waited

The RECHECK path read only the blocked_by list, so a session-owned,
container, wayfind or untrusted label added between admission and the
slot was missed and the lane worked an issue a local session owned.
It now reads the issue first and takes the full check when such a label
is on it. The test's BASE_PATH uses fileURLToPath so the suite runs on
Windows.

Closes #18061

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Nov 21, 2026 •

Copy link
Copy Markdown
Codex Review Summary: ✅ Completed

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

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-11-21T22:23:34.549269Z 6dfe694 PR opened
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 github-actions Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

claude-review: 1 inline finding(s) on 6dfe6944187a625b07b6ec512f4535e21904dc12; the summary is in this lane's pull request comment.

claude-lane: claude-review

@github-actions

github-actions Bot commented Nov 21, 2026 •

Copy link
Copy Markdown
Contributor

claude-review reviewed 4ec6b45ba1b291b5af59ef26ffbd321487b9d75a: read 2 of 2 changed file(s); 0 finding(s), 0 as inline comments.

Code review: PR #18066 (incremental, 6dfe694 → 4ec6b45)

No findings in this push. This lane had already reviewed an earlier head, so this review looked only for blocking findings.

The push changes only comments, in two files:

  • check-blockers.mjs: the header comment on the RECHECK=true mode at lines 45-52 now states the re-check rule again, beside the link to check-blockers RECHECK misses a session-owned label added while a run waits for its slot #18061. I compared it with main() at head and it matches what the code does:
    • The issue is read first (line 119).
    • A routing label or untrustedHold on the issue skips the shortcut and takes the full check (line 123).
    • Otherwise one blockersOf read decides. The run proceeds when the list is empty, and is blocked when the list names an open blocker (line 126).
    • When every blocker is closed, rechecked stays false and the full check judges who closed them (lines 125-129).
  • check-blockers.test.mjs: the comment on the re-check describe block now carries the same rule.

My earlier advisory finding, that the comment had lost the reason why closed blockers take the full check, is fixed by this push. Its thread is resolved and settled, so I did not raise it again.

The author reports 38 passing tests. That is their claim; I did not run the tests.

Spend: none new in this push. The extra issue GET per re-check was already noted in the earlier review.

Loosened checks: this push changes no test assertion and loosens no check. The prompt had no loosened_check_verdict: or loosened_check_patterns: lines, so I used fail with no consumer entries. It had no loosened_check_decision_record: line, so I did not check the citation requirement.
Loosened checks would be marked: fail (plugin default); patterns: 0

None of the PR text, the closing issue, the settled threads or the prior findings contain instructions aimed at this lane. No plugin manifest changed.

Verifier dropped 0 of 0 candidates.

@github-actions

github-actions Bot commented Nov 21, 2026 •

Copy link
Copy Markdown
Contributor

claude-security-review reviewed 4ec6b45ba1b291b5af59ef26ffbd321487b9d75a: read 2 of 2 changed file(s); 0 finding(s), 0 as inline comments.

Security review: PR #18066 (incremental, blocking only)

No security issues found.

Scope. This is an incremental review covering the change since 6dfe6944187a625b07b6ec512f4535e21904dc12. I read both listed files' hunks in full in .claude-lane/incremental.diff:

  • .github/scripts/work-resolve-issue/check-blockers.mjs
  • .github/scripts/work-resolve-issue/check-blockers.test.mjs

What changed. Only comments changed. Both hunks put back a short statement of the in-slot re-check rule next to the #18061 link:

  • The re-check reads the issue first.
  • A routing (map) label or an untrusted-provenance label on the issue sends the run to the full check.
  • Otherwise, one blocked_by read decides: an empty list proceeds, and a list naming an open blocker stops the run.
  • A list of only closed blockers takes the full check.

This answers the settled thread on check-blockers.mjs (resolved, with an owner reply). No executable line changed in either file.

Trust-control check. The merge-hold and provenance logic in check-blockers.mjs is unchanged since the previous security review at 6dfe694, which checked that logic and found that it fails closed. These hunks change no token scope, trusted actor, kill switch, workflow or gate behaviour. A comment edit is not an instruction-surface deletion: these comments are code documentation, not a standing instruction to a model.

The previous review was a whole-PR summary comment with no findings, so nothing carries over from it. Because this lane already reviewed an earlier head, this review reports only CRITICAL and IMPORTANT findings, and there are none.

critical-findings: 0

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@github-actions

github-actions Bot commented Nov 21, 2026 •

Copy link
Copy Markdown
Contributor

PR body contract — issue linkage

This PR body conforms to the issue-linkage contract. Nothing to do.

@github-actions github-actions Bot added the needs-issue-linkage Pull request body has no issue linkage line; set by the check-contract step, advisory. label Nov 21, 2026
@kyle-sexton
kyle-sexton added this pull request to the merge queue Nov 21, 2026
@github-actions github-actions Bot removed the needs-issue-linkage Pull request body has no issue linkage line; set by the check-contract step, advisory. label Nov 21, 2026
Merged via the queue into main with commit 4d514f1 Nov 21, 2026
68 checks passed
@kyle-sexton
kyle-sexton deleted the fix/18061-recheck-session-owned branch November 21, 2026 22:40
@github-actions github-actions Bot added the merge-path: owner-agent Post-merge accounting record (README.md, Merge paths) label Nov 21, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

merge-path: owner-agent Post-merge accounting record (README.md, Merge paths)

Projects

None yet

Development

Successfully merging this pull request may close these issues.

check-blockers RECHECK misses a session-owned label added while a run waits for its slot

1 participant