Health

Health/readiness endpoint. - Redis unset → `200 {status: ok, redis: disabled}` (single-worker dev). - Redis configured and reachable → `200 {status: ok, redis: ok}`. - Redis configured but unreachable → `503 {status: degraded, redis: unavailable}`. The 503 is intentional: rate-limit storage is now Redis-backed (see `backend/core/limiter.py`), so when Redis is down the limiter raises on every protected request and the instance can't reliably serve chat / auth / widget. Reporting unhealthy lets Railway pull the instance out of rotation instead of returning 500s to clients.

curl -X GET "https://api.getchat9.live/health"

null