Repository navigation
#1894 AC #4: give maintainers a dev-lead-independent way to disposition a blocking PR comment #1910
Description
Summary
Split from #1894 (PR #1902 delivers AC #1–#3 only). Raised on PR #1902 review threads on scripts/lib/approval-diagnostic.sh.
The approval diagnostic correctly reports that an undispositioned non-agent PR comment is the blocking gate, and names what satisfies it: the comment must be minimized with classifier RESOLVED (or an @mention FORCE_REVIEW bypass). maintainer-comment-gate.sh already honours a RESOLVED-minimized comment as addressed. That is AC #1 and it works.
What is missing (AC #4): the only thing that ever performs the minimize is dev-lead. So a comment on a PR where dev-lead is suppressed, rate-limited, cancelled, or never dispatched stays undispositioned forever, and no approval can stand. There is no path a human maintainer can invoke to satisfy the gate against their own comment.
This is the acceptance criterion that actually unblocks the queue of stuck PRs — reporting the blocker is necessary but not sufficient.
Proposed direction
A small script or a documented gh api graphql minimizeComment invocation a maintainer can run against their own comment (minimize with classifier RESOLVED after a verified disposition), so the gate can be satisfied without dev-lead running. Any implementation must preserve the existing gate semantics in maintainer-comment-gate.sh.
Acceptance criteria
- A maintainer can satisfy the undispositioned-comment gate on their own comment without dev-lead executing.
- The path is documented (AGENTS.md or script
--help) and does not weaken theRESOLVED-minimized requirement the gate already enforces. - A test covers the maintainer-invocable path.
Activity
don-petry commentedon Sep 23, 2026
Dev-Lead Implementation Plan
Issue: #1910 — #1894 AC #4: give maintainers a dev-lead-independent way to disposition a blocking PR comment
Scope
Add a small, maintainer-invocable script that minimizes a maintainer's OWN PR issue comment with GraphQL
minimizeCommentclassifierRESOLVED— the exact signalmaintainer-comment-gate.shalready treats as addressed — so the undispositioned-comment gate can be satisfied without dev-lead ever running.Implementation Checklist
scripts/maintainer-resolve-comment.sh— new executable with pure, testable helpers (comment-ref parsing, self-authorization, the fixedclassifier:RESOLVEDmutation) + agh-backedmainguarded byBASH_SOURCE, plus--helpscripts/lib/consumer-manifest.json— register the new script on the pr-review + dev-lead surfaces.github/workflows/lint.yml— add the new bats file to the test listAGENTS.md— document the maintainer path and that it does not weaken theRESOLVED-minimized requirementTests to Write
tests/dev-lead/unit/test_maintainer_resolve_comment.bats— structural (shebang,set -euo pipefail, shellcheck),--helpdocuments the RESOLVED requirement, mutation hardcodesclassifier:RESOLVED, self-authorization (own comment passes, other author fails closed), comment-ref parsing (node id vs URL), idempotency on already-RESOLVED, and a round-trip that the resolved comment then passescheck_maintainer_commentsTest Command
scripts/run-bats.sh tests/dev-lead/unit/test_maintainer_resolve_comment.bats(full suite via thebatsjob inlint.yml)don-petry commentedon Sep 23, 2026
Dev-Lead: Progress (in progress — not yet durable)
Plan Execution (intended)
minimizeCommentclassifierRESOLVED— the exact signalmaintainer-comment-gate.shalready honours — so the undispositioned-comment gate can be satisfied without dev-lead running.RESOLVEDhardcoded (requirement not weakened), idempotent, fails closed when it cannot confirm the comment author or the viewer.--helpand a new AGENTS.md subsection; test added and wired into thebatsjob inlint.yml.Files touched (this workspace, not yet pushed)
scripts/maintainer-resolve-comment.sh: new maintainer CLI (puremrc_*helpers +gh-backedmain).tests/dev-lead/unit/test_maintainer_resolve_comment.bats: structural,--help, mutation-classifier, self-authz, ref parsing, idempotency, and round-trip-through-the-real-gate tests..github/workflows/lint.yml: registered the new bats file.AGENTS.md: documented the dev-lead-independent path.Notes
scripts/lib/consumer-manifest.json: this is a human-invoked CLI, not a reusable-workflow runtime surface (the manifest validator only acceptsscripts/lib/*.shorprompts/*values).maintainer-comment-gate.sh) is unchanged — the maintainer path produces the identicalRESOLVED-minimized signal rather than a new/weaker one.approval-diagnostic.sh(AC test issue from agent #1–Add @claude delegation, auto-merge, and rebase handling #3 / PR feat(#1894): answer "why isn't this PR approved?" from the real pr-review gate chain (+ why-not-approved agent) #1902) is not on this base branch; this change depends only on the existing gate, so no coupling to that unmerged work.don-petry commentedon Sep 23, 2026
Dev-Lead: Implementation Complete — PR #1911
The implementation for issue #1910 is durable: commits are pushed and a pull request is open. This record is backed by verifiable artifacts you can confirm in one click.
b7b805b00fdf75b56cb500aa7eea38d5e2ed8816main: 4 file(s), +399/-0 linesReview happens on the PR. Acceptance criteria and test results are evidenced by the PR's diff and CI checks — not asserted here. If a later stage fails, this claim is retracted in place.
feat: implement issue #1910 — #1894 AC #4: give maintainers a dev-lea…