Deploy

2. Deployment Execution

Find deployment method:

• Look for script/deploy or similar
• Check Makefile for deploy target
• Look in package.json for deploy script
• Check for CI/CD configs (GitHub Actions, GitLab CI, etc.)
• Check for platform-specific (Heroku, AWS, Vercel, etc.)

Execute with confirmation:

• Show what will be executed
• Ask for confirmation (especially for production)
• Run deployment command
• Monitor output for errors

3. Post-Deployment

Verification

• Check deployment logs
• Verify health endpoints if available
• Suggest smoke tests
• Monitor error tracking (Sentry, etc.)

Rollback Plan

• Document the current version/commit
• Explain how to rollback if needed
• Keep previous version info available

Safety Checks

Critical Checks (Block Deployment)

• Uncommitted changes on production branch
• Failing tests
• Security vulnerabilities in dependencies
• Wrong branch for environment

Warning Checks (Warn but Allow)

• Missing tests
• Outdated dependencies (non-security)
• No CI/CD validation
• Direct push to production (suggest PR process)

Environment-Specific

Staging

• Less strict validation
• Allow faster iteration
• Auto-deploy from develop branch OK

Production

• Strict validation required
• Require manual confirmation
• Prefer tag/release over direct branch deploy
• Ensure rollback plan exists

Common Platforms

Heroku

git push heroku main
heroku logs --tail

AWS/Docker

docker build -t app:latest .
docker push registry/app:latest
kubectl rollout restart deployment/app

Vercel/Netlify

vercel --prod
# or
netlify deploy --prod

Traditional

ssh user@server 'cd /app && git pull && restart-service'

Rollback Procedures

If deployment fails:

1. Note the failing commit
2. Revert to previous working version:

   git revert <commit>
   # or
   git reset --hard <previous-good-commit>
   git push --force  # only if absolutely necessary
   

3. Re-deploy previous version
4. Investigate failure in staging
5. Fix and re-deploy

Best Practices

• Always deploy from clean state - commit or stash changes
• Tag releases - use semantic versioning
• Keep deployments small - easier to debug and rollback
• Automate what you can - reduce human error
• Monitor after deploy - catch issues early
• Have rollback plan - know how to undo quickly
• Use feature flags - deploy code, enable features separately
• Blue-green deployments - zero-downtime deploys

Checklist Template

## Deployment Checklist

- [ ] Tests passing
- [ ] No uncommitted changes
- [ ] Branch up-to-date
- [ ] Dependencies secure
- [ ] Deployment method identified
- [ ] Target environment confirmed
- [ ] Rollback plan documented
- [ ] Health checks defined
- [ ] Monitoring ready