Production readiness

Required launch gates

• Environment validation must pass before production startup.
• The public app URL and auth URL must use the production HTTPS origin.
• The health endpoint must report database, required environment, worker heartbeat, probe freshness, and pending alert handoff checks as healthy.
• Robots and sitemap responses must use the public origin and must not contain localhost URLs.
• Production releases should include test, typecheck, build, and smoke-test evidence.

Operational verification

• Use the production smoke-test script after deploy to check public pages, health, robots, sitemap, and manifest.
• Verify alert channels, incident routing, GitHub webhooks, Stripe webhooks, OTLP ingest, and agent telemetry in a staging organization.
• Keep a current NEXUS AI backup and restore drill record with observed RTO and RPO.
• Monitor the platform itself with a synthetic check against the public site and health endpoint.
• The dashboard shows No recent data when a probe is older than five minutes or three check intervals, whichever is longer. Paused monitors and monitors waiting for their first result are shown explicitly.
• Incident state, the incident record, and its pending alert are saved together. Pending alerts survive worker restarts and queue outages, and retries reuse the same delivery job.

Security evidence

• Confirm secrets are stored in the deployment secret store and not committed to the repository.
• Review tenant isolation, RBAC, webhook signatures, SSRF controls, rate limits, and audit log coverage.
• Rotate any credential that was pasted into logs, chat, screenshots, or untrusted storage.
• Document known disabled features, accepted risks, and remediation owners before launch.

Related documentation