$npx -y skills add wednesday-solutions/ai-agent-skills --skill deploy-checklistPre-deploy and post-deploy checklist skill. Ensures env vars, migrations, CI, rollback plan, smoke tests, and monitoring are verified before and after every deployment.
| 1 | # Deploy Checklist Skill |
| 2 | |
| 3 | ## Trigger |
| 4 | |
| 5 | Load this skill when a dev is about to deploy or has just deployed: |
| 6 | - "We're deploying to production" |
| 7 | - "Pre-deploy check" |
| 8 | - "Post-deploy verification" |
| 9 | - "Run the deploy checklist" |
| 10 | - "Is it safe to deploy?" |
| 11 | |
| 12 | **Do NOT use this skill for:** committing code (use `git-os`), creating a PR (use `pr-create`), or planning a project (use `greenfield`). This skill only applies at the deployment stage — code is already merged. |
| 13 | |
| 14 | --- |
| 15 | |
| 16 | Run this checklist before and after every production deployment. |
| 17 | |
| 18 | ## Pre-Deploy Checklist |
| 19 | |
| 20 | - [ ] All CI checks green on the deploy branch |
| 21 | - [ ] Environment variables verified in target environment (no missing keys) |
| 22 | - [ ] Database migrations reviewed — irreversible migrations documented |
| 23 | - [ ] Migrations have been dry-run or tested in staging |
| 24 | - [ ] Rollback plan documented: what to revert and how |
| 25 | - [ ] Feature flags set correctly for the release |
| 26 | - [ ] Downstream services notified if API contracts changed |
| 27 | - [ ] Changelog updated with this release's changes |
| 28 | - [ ] Deployment window confirmed (avoid peak traffic) |
| 29 | |
| 30 | ## Deploy |
| 31 | |
| 32 | - [ ] Deploy initiated with correct branch / tag |
| 33 | - [ ] Deployment logs monitored in real time |
| 34 | - [ ] No unexpected errors during startup |
| 35 | |
| 36 | ## Post-Deploy Checklist |
| 37 | |
| 38 | - [ ] Smoke test: critical user flows verified manually or via synthetic monitoring |
| 39 | - [ ] Health check endpoint returns 200 |
| 40 | - [ ] Error rate in monitoring (Datadog, Grafana, Sentry) is normal |
| 41 | - [ ] No spike in latency or DB query time |
| 42 | - [ ] Monitoring alerts reviewed — no new alerts triggered |
| 43 | - [ ] Changelog published / communicated to stakeholders |
| 44 | - [ ] Ticket status updated (closed / released) |
| 45 | |
| 46 | ## Rollback Trigger Criteria |
| 47 | |
| 48 | Initiate rollback immediately if: |
| 49 | - Error rate rises above 1% of requests |
| 50 | - P95 latency increases by more than 2x baseline |
| 51 | - Any data integrity issue detected |
| 52 | - Critical feature path returns 5xx |
| 53 | |
| 54 | ## Tools |
| 55 | |
| 56 | | Action | Tool | |
| 57 | |--------|------| |
| 58 | | Run lint, test, build scripts | `Bash` | |
| 59 | | Check health endpoint | `Bash` — `curl -s <url>/health` | |
| 60 | | Read config or env files | `Read` | |
| 61 | | Check CI status | `Bash` — `gh run list` or `gh pr checks` | |
| 62 | |
| 63 | ## Notes |
| 64 | |
| 65 | - Never deploy on Fridays unless it's a critical hotfix |
| 66 | - Always have a second engineer available during production deploys |
| 67 | - Document the actual deploy time and outcome in the ticket |