Add low-cost backend alerting

[KostasCherv]

Summary

• provision an idempotent Cloud Monitoring baseline for backend uptime, 5xx bursts, and Cloud Run runtime/probe failures
• enable privacy-safe Sentry error capture for handled domain errors, SSE failures, and terminal outbox delivery failures
• wire Sentry production configuration through Secret Manager and Cloud Run
• document GCP, Sentry, and Inngest activation and cost guardrails

Why

The backend already exposed health/readiness signals and produced useful logs, but the production GCP project had no alert policies or uptime checks. Several SSE and outbox failures could also return HTTP 200 or be swallowed, making request-level 5xx monitoring insufficient.

Operational impact

The GCP alerting baseline has been applied successfully to cortex-502007 and the setup script was rerun to verify idempotency. Sentry remains inactive until a real sentry-dsn secret is populated and the updated backend is deployed. Inngest email rules still require dashboard activation.

Validation

• full test suite: 585 passed
• focused alerting/API suite: 124 passed
• Ruff checks passed
• mypy passed
• Bandit passed
• shell syntax and dry-run checks passed
• GCP setup rerun successfully with exactly one uptime check and three enabled policies
• production /health returned {"status":"ok","version":"0.1.0"}

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
cortex	Ready	Preview, Comment	Jul 17, 2026 9:48am