Repository navigation
Closed
Labels
enhancementFeature requestsFeature requestsneeds-human-reviewFlagged by automated PR review agentFlagged by automated PR review agent
Description
Triage of the "genuine feature/bug, no-PR" backlog (2026-07-13 overnight session). Grouped by recommended action. Complex agent-engine + product-direction items were not implemented unattended — they're flagged here for a go/no-go.
✅ Already resolved — closed
- .github-private#324 (yamllint extglob SC2046) —
ci.ymlalready usesshopt -s extglob+!(*.lock).yml; dev-lead confirmed "already applied in main". Closed.
🟢 Likely already fixed — recommend verify + close
- .github#437 (compliance false-positive on transient API reads) —
compliance-audit.shnow has a 3-retrygh_api()wrapper with backoff (L113-123) + explicit 403-stdout handling. Core concern appears addressed. Verify no remaining single-read paths, then close. - .github-private#145 (PR Review Agent run docs(AGENTS): record dev-lead/pr-review automation rules (hands-off label; pr-review is CODEOWNER approver) #715 aborted early, 2026-05-12) — 62-day-old incident post-mortem referencing
opus-4-7(now 4-8); the agent's been reworked since (advisory gate, timeout fixes dev-lead engine repeatedly times out (engine-error) on larger issues — exhausts retries without delivering (e.g. .github#518) #901). Likely stale — confirm the 3 failure modes no longer apply, then close.
🔴 Live bugs — need implementation (higher risk, want a go-ahead)
- .github-private#463 (PR Review Agent never reviews at green CI → clean PRs stuck at
REVIEW_REQUIRED) — still live: this session's own PRs are sitting atREVIEW_REQUIRED. High value, but touches the pr-review trigger architecture (check_suite/workflow_rundon't fire) — needs careful design + validation. Recommend prioritizing. - .github#443 (multi-label issue creation cancels dev-lead dispatch — per-issue
cancel-in-progressrace) — real concurrency bug indev-lead.yml. Fix needs care not to break the labeled-trigger flow. - .github-private#655 (stale-base merge feat: implement issue #574 — pr-review: PRs with >300 changed files hard-fail the cascade (gh pr diff 406); define org standard for oversized PRs #606 silently reverted the initiative-planner feat: implement issue #591 — bug(initiative-planner): claude-code-action rejects
discussionevents — theidea:approvedauto-trigger never plans #618 fix →idea:approvedstopped planning; "Unsupported event type: discussion") — verify current planner state; re-apply the reverted fix if still broken. - .github#203 (add
dependabot-automerge-petrybypass toprotect-branches+ TalkTerm rulesets) — config change to ruleset JSON + apply. Moderate risk (merge-gate change); straightforward once greenlit.
🟣 Need product/design direction (not started)
- .github-private#617 (agent-authored infra PRs hard to merge — approval-identity + bot-review) — this is the meta-problem hit all session; overlaps PR Review Agent never reviews at green CI — skips on ci-pending with no CI-completion re-trigger (clean PRs stuck at REVIEW_REQUIRED) #463. Wants a design decision on the approval path.
- .github-private#619 (self-host deadlock:
force_reviewbypasses advisory gate but not CI) — agent-gate design. - .github#74 (pr-quality ruleset support in
apply-rulesets.sh, opened 2026-04-06) — old feature; confirm still wanted. - bmad-bgreat-suite#60 (multi-agent "party mode" cross-domain review, 2026-04-06) — old feature; confirm still wanted.
- Safe Release Strategy epic — Initiative: Safe Release Strategy for Agentic Workflows (versioning · rings · canary) #495 (initiative) + Phase-2 stories [Phase 2] Add next + per-ring channel tags (blue/green green channel) #499 (next + per-ring channel tags), [Phase 2] One-action rollback + per-version observability report #502 (one-action rollback + observability), [Phase 2] SC2 game-day: prove a broken next cannot block its own fix PR #503 (SC2 game-day). Large; needs prioritization/sequencing before any implementation.
Recommended order when you're back
- PR Review Agent never reviews at green CI — skips on ci-pending with no CI-completion re-trigger (clean PRs stuck at REVIEW_REQUIRED) #463 (unblocks the whole review pipeline, incl. this session's PRs) → 2. fix: support fine-grained tokens in pr-review scope validation #443 (dispatch race) → 3. bug: stale-base merge (#606) silently reverted the initiative-planner discussion→dispatch fix (#618);
idea:approvedstopped planning #655 (planner regression) → 4. ci: add pr-review-mention.yml caller stub #203 (rulesets) → 5. verify+close petry-projects — workflow failures detected 2026-06-05 #437/PR Review Agent: run #715 aborted early — 3 failure modes identified #145 → 6. decide on Agent-authored infra PRs are hard to merge — approval-identity + bot-review friction #617/Self-host deadlock: force_review bypasses the advisory gate but not the CI gate, so CI-gate fixes can't merge through ring-0 #619/fix: pin actions/checkout to SHA in pr-review.yml #74/Compliance: missing-code-quality #60 and the Safe Release epic.
Metadata
Metadata
Assignees
Labels
enhancementFeature requestsFeature requestsneeds-human-reviewFlagged by automated PR review agentFlagged by automated PR review agent
Activity
github-actions commentedon Jul 14, 2026
Thanks for the thorough triage write-up — this is a great consolidated view of the backlog. Before moving forward on the live bugs, could you confirm whether you'd like to proceed with the recommended priority order (starting with #463 to unblock the review pipeline), or if any of the 🔴 items should be re-sequenced or held pending the 🟣 design decisions (particularly #617/#619, which overlap with #463's trigger architecture)?
don-petry commentedon Jul 14, 2026
Live-bugs outcome (2026-07-14) — all 4 resolved
All four turned out to be already fixed by intervening infrastructure work (verified, not re-implemented — the fleet self-healed via the sweep / concurrency / canary / ruleset-standardization programs):
pr-review-sweep.yml— scheduled ≤15-min backstop +workflow_runfast path, PAT-dispatched (#573/#898). Verified running today.cancel-in-progress: false+ per-issue/per-PR lanes (#450, landed 2026-06-06 — before #443 was filed). Added regression guard #1221.initiative-planner-canary.yml.code-quality+pr-qualityuniformly carry the bypass fleet-wide (0 missing in a full scan).One new PR: #1221 (dev-lead concurrency regression guard).
Residual non-blocking notes:
standards/ci-standards.mdstill has a legacyclaude-issueexample (~L417) showing the oldlabel.name=='claude'+cancel-in-progress: trueanti-pattern — doc-hygiene cleanup for the claude-deprecation (feat(token-report): org-wide weekly Token Cost Observatory report #456).dev-lead/dispatch,add-to-project) makescompute_ci_statussee CI as failing, so the sweep/agent won't approve. Worth a decision if it recurs.idea:approvedstopped planning #655: prevention items (require-up-to-date-branches/merge-queue; flag test-deletions) remain process decisions.don-petry commentedon Jul 14, 2026
Verify-and-close + product-direction outcomes (2026-07-14)
Verify + close
check_codeownersnow distinguishes a deterministic 404 (missing) from a transient/auth error (inconclusive → warning, never filed as missing), mirroring #487. +4 bats.test_batch_fallback.Product-direction
FORCE_REVIEWnow bypasses the ci-failing gate (manual-only break-glass); #608 had handled the CANCELLED half. +2 bats.build_pr_quality_ruleset_json), applied fleet-wide.require_last_push_approval, fix.githubdispatch. Posted for your call; not changed unilaterally.pr-review/next/dev-lead/v1-next, ring topology, self-host break-glass via #1230). Needs PM reconciliation — several phase stories appear substantially delivered. Recommend a pass to close/rescope against reality rather than fresh implementation.New PRs this round: .github#721 (#437), .github-private#1230 (#619). Plus earlier: #1221 (#443 guard).
Everything with a clean, safe, code-level fix is done. What remains (#617 process decisions, #60 roadmap, #495 reconciliation) is genuinely your call — tell me the direction and I'll implement.
don-petry commentedon Sep 1, 2026
Closing — stale; the go/no-go it asked for was answered
This was a triage snapshot taken 2026-07-13 of the "genuine feature/bug, no-PR" backlog, held open for a maintainer go/no-go on the higher-risk items.
Every issue it tracked closed on 2026-07-14 — the day after the snapshot:
.github-private#324(yamllint extglob SC2046).github#437(compliance false-positive on transient API reads).github-private#145(PR Review run #715 aborted early).github-private#463(pr-review never reviews at green CI)The decision was made and the work landed; only the tracking artifact outlived it. Nothing actionable remains.