Repository navigation
Merge picks offer "Explain it first" and "Is this the best approach?" - #768
Conversation
Quest-Claim: 3533a183ca04f20c
… picks Skip and delete become options only as the recommendation. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
|
Warning Review limit reachedYou've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository. Next included review available in 27 minutes. View limit details
No actionable comments were generated in the recent review. WalkthroughThe PR question guidance now offers “Explain it first” and “Is this the best approach?” options, then requires asking again without the selected option. Skip and delete are options only when recommended. The planning document about proposed merge-pick options and follow-up flows was deleted. Merge Risk:
|
kixelated
left a comment
There was a problem hiding this comment.
Automated review by review (OpenAI)
Reviewed head 72cbd0f64882db2f39916eab0aae6d244898a5f9.
Go for the scoped prompt change. The option budget works even when the recommendation is skip or delete, and neither explanation choice grants a merge pick. One worthwhile guard: when the independent assessment returns, refresh the head and readiness before re-asking. If a fixer pushed while the agent was working, the verdict is about the old diff and should be rerun or clearly marked stale, not presented as the verdict for the new head. No CLI or wire concern here; static review.
There was a problem hiding this comment.
✅ No new issues found.
Reviewed changes: the skill wording edit in assets/skills/run.md and the deletion of the completed quest file.
- Merge pick options: the per-PR question now offers the recommended action,
/quest-mergewhen something else is recommended, "Explain it first", and "Is this the best approach?". That is at most 4 options, which fits the harness limit. - Follow-up behavior: "Explain it first" re-asks without that option. "Is this the best approach?" dispatches a background sub-agent that isn't given the recommendation, then re-asks with its verdict beside the recommendation.
- Quest cleanup:
quest/m0/pick-options.mdis removed, since this PR finishes it.
claude-sonnet-5-5 | 𝕏
|
Maintainer decision (from the maintainer's
(Written by Claude Opus 5.5 with high effort) |
… options with /quest-review Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
|
Applied the decision above on head 7acd753: the best-approach sub-agent now also gets the implementing quest's Goal, Plan, and evidence (read from the PR's base or git history, since this PR deletes the file), and the pick question is written once in (Written by Claude Sonnet 5.5) |
kixelated
left a comment
There was a problem hiding this comment.
Automated review by review (OpenAI)
Reviewed head 7acd753e975e00d2ae7cd2ed65bb95f8068db745.
No new blocker in this round. Having /quest-run and /quest-review use the same pick instructions in merge.md prevents their option sets drifting. Giving the independent reviewer the quest at the base supplies the missing context without giving it the coordinator's recommendation. Keep both the quest and the assessed PR revision explicit when dispatching, so a later push can't make an old assessment look current. Static review.
There was a problem hiding this comment.
✅ No new issues found.
Reviewed changes: the commits since 72cbd0f move the merge pick options into one shared place.
- Shared pick options:
assets/skills/merge.mdnow states the pick question once: the recommended action,/quest-mergewhen another is recommended, "Explain it first", and "Is this the best approach?". It keeps the 4-option cap. - Quest context for the sub-agent: the best-approach agent gets the quest's Goal, Plan, and evidence, read from the deleted file on the base, but not the recommendation.
- Callers point to it:
assets/skills/run.mdandassets/skills/review.mdnow refer to/quest-merge's pick options instead of restating them, so the option sets can't drift.
claude-sonnet-5-5 | 𝕏
|
Maintainer decision (2026-11-17,
(Written by Claude Opus 5.5 with high effort) |
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
There was a problem hiding this comment.
✅ No new issues found.
Reviewed changes
Since the prior review, one commit reworded the merge pick's second option in assets/skills/merge.md.
- Renamed option — "Is this the best approach?" is now "Rethink it".
- Clarified sub-agent task — the fresh sub-agent starts from scratch, brainstorms alternatives, and recommends merging, iterating toward a named alternative, or dropping the PR.
- Re-ask wording — the follow-up shows its recommendation beside the coordinator's, with where they differ.
claude-sonnet-5-5 | 𝕏

Finishes quest/m0/pick-options.md (deleted here).
Summary
The merge pick question, now written once in
assets/skills/merge.md(/quest-run,/quest-complete, and/quest-review's closing question point to it), asks each merge pick with these options:/quest-merge,/quest-iterate, skip, and/quest-delete, plus/quest-mergewhen something else is recommended (so the common 84% pick is never typed under Other);Skip and delete are options only as the recommendation. That is at most 4 options (merge, recommended, explain, rethink), within the harness's limit.
Public API
No CLI or wire change; skill wording only.
Validation
No behavior to exercise: a shipped skill's wording.
just checkandjust testpass (1424 tests passed, 5 skipped);quest checkok.Decisions
Settled by the maintainer in this comment:
/quest-review's closing question offer the same options? ✅ yes, said once inassets/skills/merge.md, which/quest-runand/quest-reviewbuild on, and/quest-reviewpoints to it (rejected: restating the options inreview.md, askills: rundependency, which pushesreviewpast the 30,000-character inline limit).Settled by the maintainer in this comment:
Models
Size: S
Implementer: Claude Code, claude-sonnet-5-5, high effort (arm B)
Implementer: Claude Code, claude-sonnet-5-5, low effort (iterate, XS, arm B)
Implementer: Claude Code, claude-sonnet-5-5, low effort (iterate, XS, arm B)
(Written by Claude Sonnet 5.5 with high effort)
Review
Rewritten by
quest reviewafter each round.Head: 32a8a87
Base: origin/main at 4249857
Risk: high,
Changes to shared agent skills alter how agents obtain and interpret merge picks across consuming repositories, potentially causing unintended merges or releases whose effects persist after reverting the skills.Classified: high by Codex, gpt-6.1-sol, low effort, session
01a1e8ef-29a9-73b2-918f-fd29aa8710d6Reviewer: Codex, gpt-6.1-sol, medium effort
Session:
01a1e8d1-8816-73a0-b4a3-30a0549b1923, round 3Score: 10/10,
The incremental change clarifies the independent reassessment option and its expected recommendation without introducing actionable issues.Rounds: 3; first round 0; findings 0: fixed 0, refuted 0, open 0
Findings: none
Reviewer: Codex, gpt-6.1-sol, low effort, shadow
Session:
01a1e8d1-8816-7352-b308-d6213253a46eRounds: 1; first round 0; findings 0: fixed 0, refuted 0, open 0