fix(ops-ci): emit only currently-failing workflows; add pre-dispatch staleness check by auroracapital · Pull Request #196 · Lifecycle-Innovations-Limited/claude-ops

auroracapital · 2026-05-02T02:44:29Z

Summary

Stop wasting Sonnet quota dispatching /ops-fires agents to already-self-resolved CI failures.

bin/ops-ci previously pulled "last 3 failures" per repo regardless of whether those workflows are still red on branch HEAD. /ops-fires and /ops-go treated any 24h-old failure as actionable and dispatched fix agents — burning ~50–150k Sonnet tokens per agent before concluding "nothing to fix".

Real-world signal: during a /ops-fires session tonight, 4 of 7 dispatched fixers reported "self-resolved before I started" (PRs already merged: healify-agentcore #520, fiberwifi #471/#473, stagery-contracts #30, stagery-api Inngest Vercel redeploy). That's ~400k tokens spent confirming green builds.

Smoke test

Real portfolio (38 repos, ops-ci registry):

OPS_CI_MODE=legacy bash bin/ops-ci: 14 repos with failures
bash bin/ops-ci (new default): 6 repos with failures
8 stale entries filtered → 57% noise reduction

Changes

bin/ops-ci: survey workflows from last 30 runs, then for each workflow × tracked branch (main/dev/master) fetch latest run, emit only when conclusion=="failure". Behind OPS_CI_MODE=current default; OPS_CI_MODE=legacy reverts.
skills/ops-fires/SKILL.md: MANDATORY pre-dispatch staleness check — gh run list --workflow X --branch Y --limit 1 before spawning any fix agent. Defense in depth for cache races.
Bump plugin version 2.0.8 → 2.0.9.

Test plan

bash -n bin/ops-ci syntax check
Real-portfolio smoke (legacy 14 → current 6)
JSON output schema unchanged (consumers don't break)
Verify /ops-fires skips self-resolved failures on next invocation
Confirm /ops-go morning briefing surface count drops to actionable-only

🤖 Generated with Claude Code

Note

Medium Risk
Changes CI failure detection logic used by downstream automation, so false negatives/positives could alter what incidents get surfaced or fixed; also increases gh API calls per repo which may hit rate limits or slow runs.

Overview
CI failure reporting is switched from “recent failures” to “current state.” bin/ops-ci now surveys recent workflow names, then for each workflow and tracked branch (main/dev/master) fetches the latest run and emits only those whose latest conclusion is failure, filtering out stale failures; OPS_CI_MODE=legacy restores the prior behavior.

Ops-fires adds a defense-in-depth staleness check. skills/ops-fires/SKILL.md now requires re-checking the latest gh run result (and handling in_progress) before dispatching any fix agent, skipping self-resolved failures. Plugin version bumps to 2.0.9.

^{Reviewed by Cursor Bugbot for commit 09d8e62. Bugbot is set up for automated code reviews on this repo. Configure here.}

Summary by CodeRabbit

Release Notes

New Features
- Enhanced CI failure detection with a new default mode that surveys latest workflow runs per repository and branch for improved accuracy
- Added pre-dispatch staleness verification to ensure CI fires are still active before attempting fixes, with special handling for PR-scoped workflows
Chores
- Updated plugin version to 2.0.9

…staleness check `bin/ops-ci` previously pulled "last 3 failures" per repo regardless of whether those workflows are still red on branch HEAD. Downstream consumers (`ops-fires`, `ops-go`) treated any 24h failure as actionable and dispatched fix agents to already-self-resolved fires — burning ~50–150k Sonnet tokens per agent before it concluded "nothing to fix". Smoke test on a real portfolio: legacy mode emits 14 repos with failures, new mode emits 6 (8 stale entries filtered, 57% noise gone). Changes: - `bin/ops-ci`: per repo, survey workflows from last 30 runs, then for each workflow × tracked branch (main/dev/master) fetch latest run and emit only when conclusion=="failure". Behind `OPS_CI_MODE=current` default; `OPS_CI_MODE=legacy` reverts to old behaviour. - `skills/ops-fires/SKILL.md`: add MANDATORY pre-dispatch staleness check — `gh run list --workflow X --branch Y --limit 1` before spawning any fix agent. Defense in depth for cache races (fix landed in seconds since cache write). - Bump plugin version 2.0.8 → 2.0.9. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

coderabbitai · 2026-05-02T02:44:41Z

Caution

Review failed

Pull request was closed or merged during review

📝 Walkthrough

Walkthrough

This PR releases plugin version 2.0.9 with two behavioral improvements: refactoring the CI-failure detection script to support dual modes (current state vs. legacy 24h-based detection) and adding a pre-dispatch staleness validation check to the ops-fires skill that re-verifies CI failures before dispatching fix agents.

Changes

Plugin 2.0.9 Release: CI Failure Detection & Validation

Layer / File(s)	Summary
Version Manifest `claude-ops/.claude-plugin/plugin.json`	Version bumped from 2.0.8 to 2.0.9.
CI Failure Detection Refactor `claude-ops/bin/ops-ci`	Script refactored to support `OPS_CI_MODE`: new default "current" mode scans latest 30 workflow runs per repo, extracts distinct workflows, and emits only failures (conclusion=failure) for each workflow against tracked branches (main, dev, master); "legacy" mode retains prior status-filtered query (up to 3 results). Logic moved from inline computation into `scan_repo_current()` and `scan_repo_legacy()` functions; output format unchanged.
Pre-Dispatch Staleness Check `claude-ops/skills/ops-fires/SKILL.md`	Added mandatory validation before spawning fix agents: re-checks selected fire results at branch HEAD via `gh run list`; skips if latest conclusion is `success`, proceeds if `failure`, waits/rechecks if `in_progress`. Special handling for PR-scoped workflows using `gh pr checks`.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~28 minutes

Possibly related PRs

release: v1.8.0 — superpowers integration + wacli --follow protection #153: Modifies claude-ops/.claude-plugin/plugin.json with a version field bump, same manifest change pattern.
chore(release): promote to v1.7.0 #142: Updates the plugin manifest version field in claude-ops/.claude-plugin/plugin.json, directly related versioning change.

Poem

🐰 A rabbit hops with glee,
Failures caught in "current" mode,
Staleness checked before we flee,
CI workflows now bestowed,
Version bumped to 2.0.9! ✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately summarizes the two main changes: fixing ops-ci to emit only currently-failing workflows and adding a pre-dispatch staleness check.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create PR with unit tests
Commit unit tests in branch fix/ops-ci-stale-data

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.

❤️ Share

_{Review rate limit: 0/1 reviews remaining, refill in 60 minutes.}

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

Pins marketplace ops plugin to v2.0.9 (released via PR #196 — fix ops-ci stale data). Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

cursor

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{Bugbot Autofix is ON, but it could not run because the branch was deleted or merged before autofix could start.}

^{Reviewed by Cursor Bugbot for commit 09d8e62. Configure here.}

cursor · 2026-05-02T02:53:19Z

+
+- If `conclusion == "success"` → SKIP. Mark task completed with metadata `{resolution: "self-resolved-pre-dispatch"}`. Do NOT spawn agent.
+- If `conclusion == "failure"` → proceed to dispatch.
+- If `conclusion == null` (in_progress) → wait 30s, recheck once, then proceed if still null.


Staleness check dispatches during in-progress runs defeating PR goal

Medium Severity

The conclusion == null branch waits only 30 seconds then proceeds to dispatch a fix agent. Since most CI runs take 2–15 minutes, an in-progress run (likely triggered by a human pushing a fix) will almost always still be null after the recheck, causing dispatch anyway. This directly undermines the PR's goal of avoiding wasted token spend on self-resolving failures. Additionally, conclusion == null also matches the case where no runs exist at all (empty array from gh run list, with --jq '.[0]' yielding null), which is conflated with "in-progress" but really means the failure is stale/gone.

^{Reviewed by Cursor Bugbot for commit 09d8e62. Configure here.}

…release (#199) The v2.0.5 → v2.0.9 patch series shipped meaningful features (multi-workspace Slack #195, /ops:credentials audit #184, ops-ci current-state filter #196, telegram preflight #185, userConfig schema upgrades #182). Per semver these should have been a minor bump. This release retroactively rolls them up into v2.1.0 with a single coherent CHANGELOG entry. No code changes — only: - plugin.json: 2.0.9 → 2.1.0 - CHANGELOG.md: new [2.1.0] entry consolidating Added/Fixed/Notes for the patch series - README header + What's-new section: refer to v2.1.0 - 11 docs/*.md badges + agents-reference subtitle + migration latest-stable note: v2.0.9 → v2.1.0 Marketplace pin (.claude-plugin/marketplace.json) bumped in follow-up PR. Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

auroracapital merged commit 10c60c2 into main May 2, 2026
10 of 11 checks passed

auroracapital deleted the fix/ops-ci-stale-data branch May 2, 2026 02:47

auroracapital mentioned this pull request May 2, 2026

fix(marketplace): bump version to 2.0.9 to match plugin #197

Merged

cursor Bot reviewed May 2, 2026

View reviewed changes

auroracapital mentioned this pull request May 2, 2026

chore(release): v2.1.0 — consolidate v2.0.5–v2.0.9 features as minor release #199

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix(ops-ci): emit only currently-failing workflows; add pre-dispatch staleness check#196

fix(ops-ci): emit only currently-failing workflows; add pre-dispatch staleness check#196
auroracapital merged 1 commit intomainfrom
fix/ops-ci-stale-data

auroracapital commented May 2, 2026 •

edited by coderabbitai Bot

Loading

Uh oh!

coderabbitai Bot commented May 2, 2026 •

edited

Loading

Review failed

Walkthrough

Changes

Estimated code review effort

Possibly related PRs

Poem

Uh oh!

Uh oh!

cursor Bot left a comment

Uh oh!

cursor Bot May 2, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

auroracapital commented May 2, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Smoke test

Changes

Test plan

Summary by CodeRabbit

Release Notes

Uh oh!

coderabbitai Bot commented May 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Review failed

Walkthrough

Changes

Estimated code review effort

Possibly related PRs

Poem

Uh oh!

Uh oh!

cursor Bot left a comment

Choose a reason for hiding this comment

Uh oh!

cursor Bot May 2, 2026

Choose a reason for hiding this comment

Staleness check dispatches during in-progress runs defeating PR goal

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

auroracapital commented May 2, 2026 •

edited by coderabbitai Bot

Loading

coderabbitai Bot commented May 2, 2026 •

edited

Loading