API reference

GET /health

System health check. Returns the status of each backing service so operators and load balancers can decide whether the relay is ready to serve traffic.

No rate limit. No authentication.

Response 200 OK

{
  "status": "ok",
  "services": {
    "db": "ok",
    "redis": "ok",
    "temporal": "ok"
  },
  "timestamp": "2026-04-19T12:34:56.789Z"
}

Field	Type	Description
status	string	"ok" if all dependencies reachable, "degraded" if any is not.
services.db	string	"ok", "error", or "unknown".
services.redis	string	Same.
services.temporal	string	Same.
timestamp	string	ISO-8601 server time.

The endpoint always returns HTTP 200 — readers inspect the status field. This is intentional so that a single failing dependency does not flap the load balancer; consumers who need stricter semantics should map status: "degraded" to their own alerting.

Use cases

• Kubernetes / Docker healthchecks: hit /health and require status === "ok" for readiness; accept "degraded" for liveness.
• Status pages: poll /health and surface per-service status from services.
• Local dev troubleshooting: if any service is "error", check your docker compose stack.

What is not checked

• Chain RPC reachability — tested only lazily, at the moment a workflow activity needs it.
• The CCTP attestation endpoint (Circle Iris).
• Wallet balances and gas budgets.

For those, see the monitoring guide.