Repository navigation
docs(dev-lead): document + enable dev-lead-issue-retry dispatch type (#781) - #494
Conversation
…ombsmith/.github-private#781) Companion to combsmith/.github-private#810, which gives failed initial issue implementations a cause-bearing failure marker and a bounded auto-retry via the dev-lead-retry cron (new dev-lead-issue-retry repository_dispatch). For that retry to reach a consumer, the consumer's dev-lead.yml caller stub must list dev-lead-issue-retry in on.repository_dispatch.types. The canonical template had only [dev-lead-ci-failure] — it never gained dev-lead-reviews-retry either, so cron-driven retries only worked where a stub was manually ahead. This brings the template (and this repo's own stub) in line with the full set the cron emits: [dev-lead-ci-failure, dev-lead-reviews-retry, dev-lead-issue-retry] - standards/workflows/dev-lead.yml — canonical template gains both retry types. - .github/workflows/dev-lead.yml — this repo runs dev-lead on its own issues (where #478 stalled), so its stub gets the same types. - standards/ci-standards.md — new "Failed issue implementations" subsection (cause marker, bounded retry, dev-lead:needs-human escalation) + runbook update for the exit-124 timeout row. The dispatch-type list is not enforced by check_dev_lead_stub (it validates only the pin/agent_ref/concurrency/permissions), so this is non-breaking: consumers without the type simply don't receive issue retries until re-synced. Refs combsmith/.github-private#781 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
|
Caution Review failedPull request was closed or merged during review No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: ASSERTIVE Plan: Pro Run ID: 📒 Files selected for processing (3)
📝 WalkthroughWalkthroughTwo workflow files ( ChangesDev-lead issue-retry dispatch expansion
Estimated code review effort🎯 2 (Simple) | ⏱️ ~8 minutes Possibly related issues
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
There was a problem hiding this comment.
Code Review
This pull request introduces documentation and configuration for a new visibility and bounded auto-retry mechanism for failed issue implementations before a PR is created. It details how failures are classified, surfaced on the issue, and retried up to three times, and updates the dev-lead.yml workflow template to support the new dev-lead-reviews-retry and dev-lead-issue-retry repository dispatch types. The feedback suggests correcting a spelling inconsistency ("labelled" to "labeled") in the documentation for consistency.
Dev-Lead — fix-reviews (applied)Changes committed and pushed. |
Dev-Lead — review-changes (no-changes)No changes were needed for this PR. |
|
Advisory bots were rate-limited; auto-approval is withheld until they recover. pr-review-sweep will re-review this PR after 2026-06-19T23:15:21Z. |
Dev-Lead — review-changes (no-changes)No changes were needed for this PR. |
donpetry-bot
left a comment
There was a problem hiding this comment.
Automated review — APPROVED ✓
Risk: LOW
Reviewed commit: 5bbad6ec2f7340b9bd195f3203ffd805b7a9e2f2
Review mode: triage-approved (single reviewer)
Summary
Companion to .github-private#810/#781. Adds the missing repository_dispatch types dev-lead-reviews-retry and dev-lead-issue-retry to both the canonical template (standards/workflows/dev-lead.yml) and this repo's own dev-lead.yml stub, plus a new ci-standards.md subsection documenting the failed-issue visibility + bounded (3-attempt) auto-retry path and a runbook update on the exit-code-124 timeout row. 28 additions across 3 files.
Linked issue analysis
Closes combsmith/.github-private#781 (failed issue implementations should be non-silent + auto-retried). This PR supplies the trigger-plumbing half: a consumer stub must list dev-lead-issue-retry in on.repository_dispatch.types for the cron's re-dispatch to reach it. The template + docs changes substantively advance that goal; the engine/cron behavior itself lands in the companion .github-private#810.
Findings
No blocking findings.
- The dev-lead.yml stub header explicitly states it 'may be freely modified on PR branches to adjust triggers for repo-specific needs', so editing repository_dispatch.types is in-scope for a thin caller stub (unlike claude.yml's byte-for-byte OIDC constraint).
permissions: {}is unchanged — no permission escalation. repository_dispatch can only be fired by a write-credentialed API caller, so adding named types introduces no new untrusted-actor attack surface.- Non-breaking per PR rationale: compliance-audit.sh check_dev_lead_stub() validates pin/agent_ref/no-concurrency/statuses:read, not the trigger type list; consumers without the new type simply don't receive issue retries (graceful degradation, identical to today's dev-lead-reviews-retry behavior).
- Docs (ci-standards.md) are internally consistent: anchor link, marker format, 3-attempt ceiling, and dev-lead:needs-human escalation all align with the PR description.
- Secret-scanning MCP unavailable (repo lacks GitHub Advanced Security); gitleaks CI check passed and the diff contains only event-type identifiers and markdown — no credential material.
CI status
All required checks green: Lint, ShellCheck, CodeQL, SonarCloud, agent-shield/AgentShield, Agent Security Scan, Secret scan (gitleaks), dev-lead/dispatch, pr-auto-review, CodeRabbit. Language-specific audit jobs (npm/pnpm/pip/cargo/govulncheck) SKIPPED as expected (no matching ecosystem). mergeStateStatus BLOCKED only on REVIEW_REQUIRED.
Reviewed automatically by the PR-review agent (single-reviewer mode: fable 5). Reply if you need a human review.
|
Dev-Lead — waiting on PR blockers (intent: review-changes)PR: #494 |
|
Note @don-petry I reviewed this PR and no code changes were needed, but it still has blocking checks or reviews (failing or cancelled checks, or changes-requested reviews), so I cannot mark it done yet. I'll re-check automatically. |



Summary
Companion to combsmith/.github-private#810 (Closes combsmith/.github-private#781), which makes a failed initial issue implementation never-silent (cause marker + redacted snippet on the issue) and gives it a bounded auto-retry via the
dev-lead-retrycron — a newdev-lead-issue-retryrepository_dispatch.For that retry to actually reach a repo, the repo's
dev-lead.ymlcaller stub must listdev-lead-issue-retryinon.repository_dispatch.types. The canonical template (standards/workflows/dev-lead.yml) only had[dev-lead-ci-failure]— it never even gaineddev-lead-reviews-retry, so cron-driven retries only worked where a stub had been manually advanced (e.g..github-private's inline workflow). This PR brings the template — and this repo's own stub — in line with the full set the cron emits.Changes
standards/workflows/dev-lead.yml(canonical template):repository_dispatch.types→[dev-lead-ci-failure, dev-lead-reviews-retry, dev-lead-issue-retry]. Adding the missingdev-lead-reviews-retryalso closes a latent drift (review retries weren't reaching template-synced consumers)..github/workflows/dev-lead.yml(this repo's own stub): same type list — this repo runs dev-lead on its own issues, and deploy-standard-workflows.sh should deploy via PRs, not direct Contents-API pushes (breaks on ruleset-protected repos) #478 (the live stall that motivated Epic: Promote persona-mention to the fleet (reach incubator + auto-detect new repos) #781) was apetry-projects/.githubissue, so it needs to receive its own issue retries.standards/ci-standards.md: new "Failed issue implementations: visibility & bounded auto-retry" subsection (cause classification, thedev-lead-issue-retrymarker/cron path, 3-attempt ceiling, anddev-lead:needs-humanescalation), plus a runbook update on theexit code 124timeout row.Why this is non-breaking
check_dev_lead_stub()incompliance-audit.shvalidates only the stub's pin /agent_ref/ no-concurrency /statuses: read— not therepository_dispatch.typeslist (triggers are explicitly stub-editable per the stub header). So this adds no compliance findings: consumers without the new type simply don't receive issue retries until their stub is re-synced — graceful degradation, identical to today's behaviour fordev-lead-reviews-retry.Follow-up
Consumer
dev-lead.ymlstubs across the org should be re-synced from the updated template (via the standard stub-sync process) to pick up both retry types. That mass re-sync is intentionally not part of this PR.🤖 Generated with Claude Code
Summary by CodeRabbit
Documentation
Chores