Write a small production smoke test suite that runs against a live environment after every deploy. Not a full test suite; a handful of tests that answer 'is the app basically working right now'. Build it as follows: 1. Pick a tool that does real HTTP or browser interaction: Playwright, Cypress, or a simple k6/curl script depending on what I already use. The tests must hit the deployed URL (staging first, then production) exactly like a user would. 2. Cover the critical flows only: - Homepage loads and shows expected content. - Sign-in works with a canary account (stored securely in the CI secret store). - The main authenticated action succeeds (create the primary object, submit the primary form). - Checkout works against Stripe test mode (or the equivalent) if payments matter. - A representative API endpoint returns 200 with the right shape. 3. Use a canary account, not a real user's account. Its data resets are documented and its email address is one you control. 4. Run automatically at two points: (a) after a staging deploy, blocking production promotion if it fails; (b) after a production deploy, alerting me if it fails so I can roll back. 5. Keep them fast. The whole suite should finish in a couple minutes. If a test flakes, fix it or delete it; nobody trusts a flaky smoke test. 6. Also run them on a low-frequency schedule (every 15-30 minutes) against production so you catch drift and outages that didn't come from a deploy. Don't try to test everything; that's what your unit/integration tests are for. The deliverable is the smoke test file(s) checked in, running post-deploy and on schedule, with alerts wired to a channel I actually watch, and one intentional failure verified to confirm the alert fires.