Every response carries standard rate-limit headers:
Handling 429
When a limit is exceeded the API returns429 with:
RateLimit-Reset seconds. For batch workloads (e.g. issuing points for many orders), spread requests out rather than bursting — 60 writes/minute is one per second sustained.
Idempotent retries
Write endpoints accept an optionalIdempotency-Key header. If a request times out or errors mid-flight, retry with the same key — within 24 hours you’ll get the original response back (marked with Idempotency-Replayed: true) instead of a duplicate operation.