Health endpoint

Liveness + readiness probe.

It is the platform's, not a module's. Every Endora instance serves it because it is an Endora instance: @endora-commerce/platform registers the route from its own application composition, so there is nothing to install, nothing to switch on, and no command that can take it away. It used to be a module called health_checks, which meant a scaffolded instance — whose module set is derived from the modules that declare themselves non-deactivatable — did not have it, and its API container never became healthy.

This page lives in the site's own tree rather than beside a module's sources, on the precedent of Module Lifecycle, whose subject is likewise a part of the platform package.

Public surface

Verb + Path	Purpose
GET /api/v1/_health	Returns 200 with a JSON status of Postgres, Redis, and Meilisearch reachability. Returns 503 if any dependency is unreachable.

Use cases

• Container orchestrator readiness/liveness probe.
• Load-balancer health check for graceful drain.
• Synthetic monitoring.

Extension points

Add new dependencies to the probe by extending the check list in packages/platform/src/http/health.ts. Each probe should be cheap (single ping / SELECT 1) and bounded (≤ 200 ms timeout).