What problem does it solve?
The Monitor Skill replaces wasteful polling loops with event-driven watchers that keep the agent idle until something meaningful happens, reducing token usage and manual checking.
Core Features & Use Cases
- Background Log Watching: Track service logs for errors, warnings, crashes, and tracebacks, then wake the agent only when a matching line appears.
- GitHub PR and CI Monitoring: Watch pull request state changes and GitHub Actions runs so review and build outcomes surface automatically.
- Deployment Health Tracking: Monitor deploy status through Railway or a health endpoint and notify the agent when a service becomes healthy or fails.
- Operational Cleanup: Start, list, and stop active watchers while keeping monitor state in a local .claude/monitors directory.
Quick Start
Ask the assistant to start a monitor for a service log, pull request, deployment, or CI branch and it will register the watcher, manage its state, and wake you only when an event occurs.