What problem does it solve?
This Skill closes the gap between “CI is green / deploy succeeded” and “the deploy actually works for users,” preventing slow-burn and coverage-miss incidents after production releases.
Core Features & Use Cases
- Post-deploy smoke tests: Design fast, fail-loud checks against real production paths with safe, idempotent behavior.
- Canary metric checks: Compare canary-vs-baseline for error rate, p99 latency, traffic shape, and resource consumption with calibrated rollback thresholds.
- Rollback and verification authority: Define who can trigger rollback (automated vs human-approved vs on-call) and avoid approval-chain delays.
- Manual verification checklist: Add timeboxed, evidence-backed critical user-flow checks for high-risk deploys.
- Monitoring window and all-clear: Keep verification responsibility during a defined window, then issue an explicit all-clear after watching the signals.
- Post-mortem driven improvement: Convert incident learnings into new checks, tests, or monitoring so the same gap doesn’t recur.
- Feature flag separation: Verify code in production with flags in the intended default state, then roll out behavior in a separate, staged operation.
Quick Start
Use the postdeploy-verification skill to generate a verification plan for your next release by providing the deploy risk level, rollout method (canary or full), critical user paths, and the monitoring duration you can support.
