Pre-Deployment Check

Walk through this checklist with the user before any production deploy. If a target is provided, scope the check to that feature or release. Present each category, confirm status, and flag anything unresolved.

Pre-Deploy Checklist

Work through each category. Mark items as pass, fail, or N/A.

Code

Check	Details
All tests pass	Run the full test suite. No skipped tests unless documented.
No console.logs / debuggers	Search for console.log, console.debug, debugger statements in changed files.
Linting clean	lint command exits with zero errors and zero warnings.
PR approved	At least one approval from a reviewer who understands the changed area.
Changes rebased	Branch is up to date with the target branch. No merge conflicts.

Database

Check	Details
Migrations tested locally	Run migrations on a fresh local database and verify the schema.
Migrations reversible	Every up has a corresponding down. Test the rollback.
No breaking schema changes	Dropping columns, renaming tables, or changing types requires a multi-step migration plan.
Seeds updated	If migrations change structure, seed files still work.

Environment

Check	Details
New env vars in production	Every new variable from .env.local is set in the production environment.
Secrets rotated if needed	If a secret was exposed or is being replaced, rotate it before deploy.
Feature flags configured	New flags exist in the flag service and are set to the correct default for production.

Dependencies

Check	Details
Lock file committed	pnpm-lock.yaml / package-lock.json / yarn.lock is checked in and up to date.
No known vulnerabilities	Run pnpm audit (or equivalent). No critical or high severity issues.
No untested major version bumps	Major dependency upgrades have been tested in staging or a preview environment.

Performance

Check	Details
No bundle size regressions	Compare bundle output before and after. Flag increases > 5%.
Lighthouse check on key pages	Run Lighthouse on landing page, main app page. No score drops > 5 points.
No N+1 queries	Review new database queries. Use eager loading or batching where needed.

Monitoring

Check	Details
Error tracking configured	New features report errors to Sentry or equivalent. Source maps uploaded.
Alerts set	Critical paths have alerting. New endpoints included in uptime checks.
Logging adequate	Key operations log enough to debug issues without leaking sensitive data.

Rollback Plan

Define the rollback strategy before deploying. Don't deploy without one.

Scenario	Action
Bug in application code	Revert the commit and redeploy the previous version.
Feature behaving unexpectedly	Toggle the feature flag off. No redeploy needed.
Database migration failure	Run the down migration. If irreversible, restore from the pre-deploy backup.
Third-party service outage	Activate fallback or circuit breaker. Communicate status to users.

Steps for every rollback:

1. Identify the issue and confirm it's deploy-related.
2. Execute the appropriate action from the table above.
3. Verify the rollback resolved the issue (check error rates, run smoke tests).
4. Notify the team with a brief incident summary.

Smoke Tests

Run these critical paths immediately after deploy. Fail any one and trigger the rollback plan.

Path	What to verify
Auth flow	Sign up, sign in, sign out, password reset all complete without errors.
Main CRUD operations	Create, read, update, delete on the primary resource. Confirm data persists.
Payment flow (if applicable)	Test checkout with a test card. Verify webhook delivery. Confirm no double charges.
API health	Hit /api/health. Expect 200 with all services reporting healthy.
Email / notifications	Trigger a transactional email or notification. Confirm delivery.
Key integrations	Verify third-party API calls succeed (OAuth, webhooks, external data).

Deploy Strategy

Choose the right strategy based on risk and infrastructure.

Strategy	How it works	When to use	Risk level
Rolling	Replace instances one at a time.	Low-risk changes, stateless services.	Low
Blue-green	Run two identical environments. Switch traffic after verification.	Database migrations, breaking API changes.	Medium
Canary	Route a small percentage of traffic to the new version. Monitor, then expand.	High-risk changes, large user base, uncertain performance impact.	Lowest
Recreate	Stop all old instances, start new ones. Brief downtime.	Dev/staging environments, or when downtime is acceptable.	Highest

Output Format

Present the checklist as a walkthrough. For each category:

1. List the checks with their current status.
2. Flag any items that fail or need attention.
3. Summarize at the end: ready to deploy, or blocked with specific items to resolve.