Rate limits

Rate limits protect studios and members from abusive or accidental traffic. Build clients so a temporary limit is handled without data loss.

Current reference status

The current Storefront and Management OpenAPI bundles do not declare a 429 response or rate-limit headers on any operation. Treat limits as a protective boundary that may be added additively, and handle 429 Problem responses if they appear.

Client behavior

If a request is rate-limited, pause that integration and retry later. Use exponential backoff with jitter. For writes, keep the same Idempotency-Key while retrying the same action so the retry cannot apply twice.

Recommended backoff:

Attempt Delay before retry
1 1-2 seconds
2 2-5 seconds
3 5-15 seconds
Later Cap at 60 seconds, then alert

Do not fan out retries across many workers with the same key. That turns a temporary limit into a larger incident.