Repository navigation
docs(prepare-pr): say the --reviewers pin no longer hides a whole-design CONCERNS - #22198
Conversation
fdee9a8 to
c54cccc
Compare
UX Review (Opus 5.5, fork) — ⏭️ skippedRevision |
First Principles Review (Opus 5.5, fork) — ✅ PASSPremise-level review of First-Principles-Verdict: PASS Check that the rationale's "Until #21808…" history sentence is wanted. Everything else matches what What this change shipsInventory (2 items) — 2 justifiedIntent: correct two prepare-pr skill passages so an agent learns the exit code, not printed text, stops on an unanswered whole-design CONCERNS under the
[FIRST-PRINCIPLES-REVIEWED] c54cccc Description read[DESCRIPTION-READ] 65eaefecd2d31a9f920d0fff1747e2c095bb4515f8c7f823a73696e74e53c900 That sha256 names the PR description this verdict read. Recompute it with |
Design Review (Opus 5.5, fork) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing to check. [DESIGN-REVIEWED] c54cccc Description read[DESCRIPTION-READ] b6dad86a1e66bb6486bb6e8585e0eee8087a6311ae38a031ebf162cbaf97b428 That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
GPT 6.1 Review (fork) — ✅ no blocking findingsReviewed Review detailsNo findings. |
Opus 5.5 Review (fork) — ✅ no blocking findingsReviewed |
mbriones98
left a comment
There was a problem hiding this comment.
Checks green and readiness passed at c54cccc; no behavior change.
iamwhatever
left a comment
There was a problem hiding this comment.
Approve at c54cccc.
The two corrected passages now match the code: pr_status.py L2397-L2409 merges design_lane_verdicts() into the pinned verdicts before unanswered_concern_lanes(), so an unanswered whole-design CONCERNS exits 20 under --reviewers gpt,opus, while marker verdicts are still pin-filtered at L1415. On the First Principles punchline about the "Until #21808" sentence: it explains why the second read exists, and rationale.md already records history the same way at L314 (#15200), so I see no reason to drop it.
All AI lanes pass or were skipped, there is no change request, CI is green (83 pass, 17 skipped) and the branch merges cleanly into current main.
— crew-pr-reviewer (r113)
Problem / Motivation
Goal: The prepare-pr skill tells an agent that an unanswered whole-design CONCERNS stops the loop through
pr_status.py's exit code under the--reviewerspin, now that #21808 made it do so.Two passages still describe the bug #21808 fixed. SKILL.md says that under the
--reviewerspin, a CONCERNS in theWhole-design punchlinesblock is the stop. The rationale section "Why a pinned reviewer fleet hides a whole-design CONCERNS" says the pin keeps those lanes' CONCERNS out of the exit code. Since #21808,pr_status.pyexits20on an unanswered Design, First Principles or UX CONCERNS whatever the pin says.Why it matters
An agent reading the skill is told to watch printed text for a stop the exit code now gives. The rationale tells a reader the opposite of what the code does, so the next person to touch the check starts from a false premise.
Not a goal
pr_status.pyor its tests. fix(prepare-pr): keep unanswered CONCERNS stop under --reviewers pin #21808 already carries the fix and its test.What changed (motivation → approach → change)
--reviewerspin does not hide the exit-20stop, instead of naming the punchline block as the stop.design_lane_verdicts(), which ignores the pin. The punchline block still changes no exit code.Backwards compatibility
Compatible: documentation only; no code, key or exit code changes.
Removes nothing: the removed sentences described behaviour #21808 already removed from
pr_status.py.Tests
N/A, documentation only. The behaviour the text now describes is pinned by #21808's test in
test/test_prepare_pr_status_concerns.py. The tests that read the prepare-pr skill files pass.Manual verification
Read
pr_status.pyon main after #21808: the unanswered-CONCERNS check mergesdesign_lane_verdicts(bot_comments, head_sha, marker_bindings)into the pinned verdicts beforeunanswered_concern_lanes(). A grep of main finds no other passage that describes the old behaviour.Related Issues
Part of #19652
Checklist
feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)