ci-orchestration_skill

Overview

This skill provides CI orchestration for GitHub Actions and other CI systems, enabling monitoring, fail-fast waiting, retrying, and preview URL extraction. It helps surface actionable failures, extract deployment previews, and automate common CI management tasks for pull requests and branches.

How this skill works

The skill inspects CI run status, checks rollups, and workflow run metadata to determine pass/fail state and detect merge conflicts or branch divergence. It can wait for checks with fail-fast timeouts, parse CI output to extract Vercel/Netlify preview URLs, and trigger reruns for failed jobs. It also formats CI check summaries and links to failed-job logs for debugging.

When to use it

• Waiting for PR CI checks to complete before merging
• Extracting preview or staging URLs from CI output to share with reviewers
• Retrying flaky jobs or re-running only failed checks
• Debugging CI failure by collecting failed-job URLs and logs
• Blocking or gating workflow steps on CI success

Best practices

• Use a short fail-fast timeout (e.g., 10 minutes) to surface early, actionable failures
• Limit automatic retries to transient errors (max ~3 attempts) with backoff
• Always extract and share preview URLs for QA and reviewers
• Parse failed-job logs to find root causes before rerunning
• Update PR comments with CI status and links to debug artifacts

Example use cases

• Wait for CI on a newly created PR, then extract and echo preview URLs for reviewers
• Watch CI with fail-fast logic to detect merge conflicts or branch divergence early
• Collect failed checks and open their logs to debug failing tests or build steps
• Automatically rerun only failed jobs for a flaky test suite with a retry loop
• Format CI checks into a compact table and post it to a PR comment for maintainers