E2e Production Ready

• Shortcut aggregated script:

  • npm run preflight:all (lint + tests + density)

• If ANY step fails, stop and triage with test output + logs. Do not deploy.

3) Database + schema validation

• Check Sequelize migrations up-to-date:
  • npx sequelize-cli db:migrate:status in src (or npm run migrate:status if available)
• Seed demo data if needed for E2E:
  • npx sequelize-cli db:seed:all
• Validate schema invariants (ledger double-entry, gig unique index)

4) Release hardening checks

• Bookkeeping invariants:
  • src/services/ledgerService.js automated tests pass
  • src/routes/nagrikRoutes.js booking retry path includes unique-constraint and 409 fallback
• Notification orchestrator path (no behavior change) check: services/notification-orchestrator/src/orchestrator.js
• Smart contract stubs remain non-production (for now).

5) Deploy to test/prod-like environment

• Deploy method (Railway): railway up / railway deploy with env set
• Or container method: docker-compose up --build + docker exec ...
• Confirm env config from .env.production.template is applied.

6) Post-deploy health checks

• API health endpoint: curl -f http://$HOST/health returns 200
• Auth endpoint quick check (OTP disabled in prod): POST /auth/login etc
• Gig lifecycle happy path (create gig, assign worker, complete gig)
• Ledger check via API/DB (balanced entries + expected commission/welfare account)

7) UAT scenario for user testing

1. Nagrik signs up, requests gig
2. Karmik accepts, completes gig
3. sanchalak can dispute + resolve
4. Settlement path charge + payout
5. Notification propagation (events->realtime->pusher/Socket.IO) validation

8) Monitoring + verification

• Check logs/metrics after smoke run:
  • services/realtime socket connects
  • services/insight-engine redis streams consuming
  • services/notification-orchestrator quiet hours behavior
• Validate stability for at least 10+ UAT sessions before sign-off

9) Completion criteria

• All tests green (jest, e2e, integration)
• Deployment done in target environment with no errors
• Production invariants maintained (ledger, atomic gig completion, booking index)
• UAT sign-off from product/QA with smoke-check checklist

Decision points + fallback

• If test failures involve DB (uuid type mismatch etc.), reproduce in isolated branch and fix model definitions before redeploy.
• If any API path returns 500 after deploy, stop and roll back to previous stable release.

Notes

• For local debugging, use npm run debug with NODE_ENV=development and inspect app logs in logs/.
• Keep this skill updated when architecture or deployment targets change (Railway vs Vercel vs Kubernetes).