Repository navigation
Session Handoff [auto-3610]: #3610 Count Copilot Findings Under the Unresolved Bullet and File Table #3641
Description
Next steps, in priority order
Re-read the issue and the code rather than trusting this summary.
- Fix Count the Findings Under Copilot's "Review Findings Remain Unresolved" Bullet in pr_review.py #3610. In the Copilot
ccr-overview-v2reader (scripts/pr_review_copilot.py, with its digest inscripts/pr_review_digest.py), count each finding that the nested "Review findings remain unresolved:" list or the file table's Findings column names. Where the headline ("Approval recommended", "No open issues",no_active_issues) and those lists disagree, the lists win. Done looks like this: the constructed body in Count the Findings Under Copilot's "Review Findings Remain Unresolved" Bullet in pr_review.py #3610 counts 2 findings rather than 0,waitdoes not report that round clean, and a regression fixture plus unit test intests/test_pr_review_copilot.pycover it. Run the whole test suite, andtests/test_pr_review_size.pyamong it. Drive the pull request intodevelop, withCloses on promotion: #3610in its body.
External blockers
- None known at pick time.
Internal dependencies
- Module size ceilings in
tests/test_pr_review_size.pybind the change.pr_review_copilot.pyhas room under its ceiling.pr_review_checks.pyis close to its own, so put the change outside it, or split along module boundaries rather than raising a ceiling.
State
Re-derive with git worktree list, git log origin/main..origin/develop, and gh pr list.
- No branch or worktree exists yet for this lane.
- Open pull requests at pick time: none.
The parked decision queue
Zero open decision issues in this repository at pick time, read live.
What the last round did
- This handoff was created by the unattended picker from the open backlog. No round has worked it yet.
What not to repeat
- Nothing yet.
New learnings
- None yet.
Activity
ptr727 commentedon Oct 30, 2026
Outcome
#3642 merged into
developas squash commit 8beff32, under the unattended run'sdevelopscope. It fixes #3610, which closes when the promotion that carries 8beff32 merges, through the PR'sCloses on promotion: #3610line.Reading.listedfield.unthreadedincludes them, soFINDINGS WITH NO THREADprints.Reading.openandoverview=T/Mkeep the count the round states.CLEAN VERDICT HEADLINE. A flagging round in that shape still exits 50, and its lines now name the headline beside the list or table.Local strict review ran over every push. Its findings were fixed in the branch, and the last pass, over fc8eb49, found none. Copilot's six threads from its round on an earlier head were each answered with the fixing SHA and resolved. Checks passed 13/13.
Cleanup exited 0. It deleted the remote branch, the worktree, and the local branch, and confirmed each.
Nothing else was filed along the way.