ci-monitor_skill

Overview

This skill continuously monitors GitHub pull request CI/CD checks until all checks complete, then reports results with clear summaries and optional audio alerts. It is tenacious by design: it polls regularly, handles transient errors, and keeps watching until a definitive pass or fail is reached or the user stops it. Use it after creating or updating a PR or whenever you want live tracking of check status.

How this skill works

The skill identifies the target PR either from a provided number or the current branch using gh. It polls gh pr checks every 60 seconds, watches individual check states (PENDING, IN_PROGRESS, COMPLETED) and conclusions (SUCCESS, FAILURE, CANCELLED, SKIPPED, NEUTRAL). When all checks complete it announces a concise success or failure summary, prints links to failing check details, and optionally plays a macOS audio notification. Transient network/API errors are retried and monitoring continues until the user explicitly stops it.

When to use it

• After creating or updating a pull request and you want to know when CI finishes
• When you ask to “monitor the PR”, “watch the CI”, or “let me know when CI finishes”
• Before merging a PR to ensure all checks pass
• When you need real-time visibility into which checks are still running or have failed
• If you want an audio alert for long-running pipelines

Best practices

• Ensure GitHub CLI (gh) is installed and authenticated before starting
• Provide the PR number when possible to avoid ambiguity; otherwise the current branch is used
• Expect long waits for large pipelines—this skill never gives up polling until completion
• Use the periodic status updates (every 5 minutes and on state changes) to stay informed without flooding output
• When failures occur, open the provided detailsUrl links for actionable logs and fixes

Example use cases

• Developer pushes a fix and asks “watch the checks” so they can continue other work until CI finishes
• After creating a PR with gh pr create, offer to monitor CI and automatically start tracking results
• Reviewer wants to know if a flaky test failed—monitor to capture the exact failing check and link
• Release engineer waits for a long matrix build and wants audio notification when the entire job set completes
• CI dashboard alternative for CLI users who prefer step-by-step console updates and failure links