canary-watch

Run a canary watch against the deployed URL to check HTTP status, console errors, API health, static assets, and performance metrics against a baseline. Use quick mode for a single pass or sustained mode to check every few minutes over a set duration.

Use diff mode with the compare flag, passing both the staging and production URLs. The skill runs the same checks against both environments and reports differences in status codes, performance, assets, and SSE behavior.

It tracks Core Web Vitals including LCP, CLS, and INP, plus API response times and SSE heartbeat latency. Critical alerts fire when LCP exceeds 4 seconds, and warnings fire when LCP regresses more than 500ms from baseline.

Yes, critical threshold crossings trigger desktop notifications on macOS and Linux, plus optional Slack or Discord webhooks. All events are also logged to a local canary-watch log file for later review.

Yes, it verifies that event-stream endpoints connect successfully and receive an initial event or heartbeat. A critical alert fires if the SSE endpoint cannot connect or drops before the first heartbeat.

Single-pass checks only capture the moment they run and may miss intermittent or load-dependent failures. For deeper coverage, pair with pre-deploy browser QA testing and CI-integrated checks after each deploy step.