Admin API
Rate limits
The Admin API uses calculated query costs and a leaky bucket per app.
| Rule | Value |
|---|---|
| Bucket size | 1000 points |
| Restore rate | 50 points / second |
| Maximum single query cost | 1000 points |
| Maximum page size (first) | 100 |
| Maximum query depth | 8 |
| Mutation base cost | 10 points per mutation field |
Cost = 1 per object field; connection fields multiply their children by first (default 20). Scalars are free. Every response includes the cost:
JSON
"extensions": {
"cost": {
"requestedQueryCost": 42,
"actualQueryCost": 42,
"throttleStatus": { "maximumAvailable": 1000, "currentlyAvailable": 958, "restoreRate": 50 }
}
}When the bucket is empty you get HTTP 429 with extensions.code = "THROTTLED" and a Retry-After header (seconds). Back off and retry; request fewer items per page for large syncs.