Write a deploy runbook for my app that I can actually follow on a rough day. Not a wall of prose; a checklist I run top to bottom. The runbook covers three sections: 1. Pre-flight (before merging to main): tests green, CHANGELOG updated with anything user-facing, any new env vars added to staging and production secrets, any migration reviewed for backwards-compatibility, feature flags configured, and (if applicable) support/marketing notified for anything user-visible. 2. Deploy: exact commands or button clicks, in order. Deploy to staging first, wait for /ready, run staging smoke tests (list the exact steps: sign in, run the critical happy path, check the admin dashboard, verify a webhook if relevant). Only after staging is verified, promote to production. If migrations are involved, run them before the new code deploys. 3. Post-deploy verification: run production smoke tests (same critical happy path against production, or a canary account), check error tracking for new spikes, check basic metrics (error rate, p95 latency) for five to ten minutes. Document what 'looks bad' means so I don't second-guess normal noise. Also include: 4. A rollback pointer at the top: 'If any post-deploy check fails, execute [rollback runbook link]'. Don't bury this. 5. Contact list: who to page for infrastructure, database, and third-party issues. Even if it's just me, write it down so future-me remembers. Don't write a novel; the whole thing should fit on one page. The deliverable is the runbook checked into the repo (or wherever I keep operational docs), one deploy run through it end to end to confirm the steps are correct and complete, and any gaps fixed based on that dry run.