Repository navigation
🐛 fix(review): require sign-off on review-fix follow-up commits - #14628
Conversation
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com> Co-authored-by: clubanderson <407614+clubanderson@users.noreply.github.com>
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
Open PR overlap checkBase branch: Conflicts are checked with every PR merged onto the current tip of
|
|
hive/awaiting-ci: CI pending — sweep will merge when green. 🐝 Hive Agent: — hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88 |
There was a problem hiding this comment.
Decision needed: @clubanderson should decide whether this PR may close #14605 with only the prompt-text requirement landed, or should cite Refs #14605 and leave the issue open for the push-guard refusal and the no-dco → fix-attempt routing it also asks for.
The code change is correct and minimal; the open question is the issue link, not the diff.
What I verified. src/pkg/review/dispatch.go:1368 appends the sign-off requirement unconditionally inside BuildFixPrompt, after the Hive-Fix-Attempt trailer line and before the human-change-request section, so every review-fix kick carries it. bin/contributor-relay.js:4023-4025 appends FIX_PR_SIGNOFF_REQUIREMENT only when task.kind === FIX_PR_KIND, after the HIVE_WORKSPACE_DIR substitution, so issue/other task prompts are untouched; resolveTaskPrompt is already exported (the existing #7908 test at contributor-relay.test.js:14233 calls it), so the new test at :14246 exercises the real function. Both tests fail without the change. Nothing else in the tree is affected.
Issue-link scope. #14605's Expected section asks for three things: (1) prompt-level git commit -s / git rebase --signoff requirement, (2) a gh-wrapper / push-guard refusal for hive-owned heads whose new commits lack the trailer, (3) routing a hive-owned no-dco head straight back into a bounded "add sign-off" fix attempt instead of 072. This PR delivers (1) and the body explicitly says (2) and (3) are "not included" because push_branch_request_watcher.go has no commit-inspection seam. That reasoning is sound — but the body then writes Closes #14605, which would close the issue on merge with two of its three asks unimplemented and untracked. If the deferral stands, Refs #14605 (push-guard refusal and no-dco → fix-attempt routing still open: push watcher has no commit-inspection seam) keeps the remainder visible; otherwise a follow-up issue for (2)+(3) should exist before this closes the parent.
Not a defect in the change itself — I'd merge the diff as-is once the link wording is settled.
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
There was a problem hiding this comment.
Decision needed: @clubanderson should decide whether #14628 may keep Closes #14605 (closing the issue with the push-guard refusal and the no-dco → fix-attempt routing still undone) or should become Refs #14605 with a follow-up tracking the remaining two expectations.
The code is fine; the issue link is the question.
Verified on this head: src/pkg/review/dispatch.go:1368 appends the sign-off requirement to every BuildFixPrompt output (the only caller is dispatch.go:1207), and bin/contributor-relay.js:4021-4026 appends FIX_PR_SIGNOFF_REQUIREMENT only when task.kind === 'fix-pr', so issue/PR tasks are untouched. Both tests fail without the change.
What I cannot settle: #14605 "Expected" lists three things — (1) the prompt wording, (2) a push guard that refuses a hive-owned head whose new commits lack the trailer, (3) routing a hive-owned no-dco head straight back into a bounded fix attempt. This PR does (1) and the body explicitly says (2) and (3) are not included because push_branch_request_watcher.go has no commit-inspection hook. Yet it says Closes #14605, so merging will close the issue with the two mechanical safeguards unimplemented and untracked — the symptom (unsigned fix commits bouncing to 072) recurs whenever a model ignores the prompt text.
Options: keep Closes and accept (2)/(3) as out of scope, or switch to Refs #14605 and open a follow-up for the guard/routing. Either is defensible; it is a scope call, not a defect, so I have not marked it as one.
— hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
Needs a decision, not a merge:
Head: verdict requires_human @3b4a2b5, 66/66 checks green, DCO yes, soak 0s of 30m0s. What to do: approve this head ( How to release it:
Clear the cause and the hive re-evaluates on its next pass; the stage moves to |
|
/hive decision approve Land the prompt-text change now — it is a correct, cheap first step. But it does not close #14605:
|
|
Decision recorded by @clubanderson for head |
|
Applied the decision: body changed from 🐝 Hive Agent: — hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88 |
|
Approval recorded by @clubanderson for head Recorded from the hive dashboard. |
|
Thank you for your contribution! Your PR has been merged. We'd love to hear how your experience was: share feedback |
Follow-up commits from hive/relay agents were unsigned, bouncing PRs to no-dco / 072 (and the hive nudged itself). The review-fix kick prompt (BuildFixPrompt) and relay fix-pr prompts now state that every follow-up commit must use git commit -s / git rebase --signoff.
Not included: the push-guard refusal and the no-dco → fix-attempt routing. The only push hook (push_branch_request_watcher.go) has no commit-inspection logic or hive-owned-head notion, so that needs a separate design; I didn't invent a new guard.
Refs #14605 (deliberately partial — this PR does not close #14605; #14605 stays open: push-guard
rebase --signoffon hive-owned heads and hive-ownedno-dco→ one bounded 070 fix attempt remain open on the issue and get their own PR(s), per the maintainer decision on this PR)— hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88 requested_by=@clubanderson