Skip to main content
Every endpoint shares one limit: 3,000 requests per 60 seconds, counted per API key.

​
Headers

Every response carries your current standing, so you can pace yourself without guessing:
RateLimit-Reset is an absolute time, not a countdown. Subtract the current time to get the wait: waitSeconds = RateLimit-Reset - now. Sleeping for the raw value parks your integration for decades.

​
When you go over

You get a 429 with code: "RATE_LIMITED". Wait until RateLimit-Reset and retry.
A 429 is the one response that repeats message at the top level, kept for callers written against the older shape. Read error.code instead.

​
Staying under the limit

Most integrations that hit the limit are making one request per record. Two habits fix that:
  • Ask for more per page. pageSize defaults to 100. Raising it means fewer round-trips for the same data.
  • Resolve ids in one call. Instead of a GET /worker/{id} per row, join the table or send one POST /worker/query with every id you need. See querying.
For writes, the bulk endpoints do the same thing in reverse. One PATCH /deliverable/bulk updates every matching record, and one PUT /deliverable/bulk upserts a whole batch. See standard endpoints.