Rate limits
Share the clinic’s request budget and respect Retry-After responses.
Two request budgets
The public API applies a default limit of 120 requests per key per minute and 600 requests per clinic per minute. The clinic budget is shared across its integration keys. Splitting work across keys does not increase that clinic budget.
The ingress also applies request-admission controls, including a trusted-IP limit. A request can receive 429 from either admission or the key/clinic limiter.
Respond to 429
A denied request returns 429 and a Retry-After header. Wait for the indicated delay before retrying. Use bounded retries with jitter and coordinate workers so they share backoff instead of producing a retry burst.
Use up to 100 resources per collection page when it suits your workload. Avoid polling every patient separately when a paginated result list can answer the question.
Availability failures
If the service cannot enforce its authorization or rate-limit state, it fails closed with 503. Treat this as temporary unavailability, not permission to bypass a check. Follow the errors and retries guide.