Rate limits and quotas
FieldTask limits requests in three ways, both for the organization as a whole and for each connection:
| Limit | Applies to | When it's reached |
|---|---|---|
| Requests per minute | Reads and creates, counted separately | 429 rate_limit_exceeded |
| Concurrent requests | Requests in progress at the same time | 429 rate_limit_exceeded |
| Daily allowance | Records created, reads, and recovery reads, counted separately | 429 daily_quota_exceeded |
FieldTask sets the limits for each organization, and a connection can have a lower allowance than its organization. The exact numbers aren't published here; owners and admins see today's usage and remaining allowances in Settings → API connections. Daily allowances reset at 00:00 UTC.
MCP tool calls count against the same limits as REST requests.
Handling 429
Every 429 includes Retry-After in seconds (MCP: retry_after). Wait at least that long before retrying. Retrying sooner doesn't help and still counts as a request.
- For
rate_limit_exceeded, the wait is short: until the next minute, or a couple of seconds for concurrency. - For
daily_quota_exceeded, the wait lasts until the allowance resets.
Things that don't use your allowance
- Replaying a create with its original idempotency key doesn't count as another create.
- Reading an operation, and looking records up by
external_reference, can use a separate recovery allowance once normal reads are used up, so you can still confirm outcomes.
Reduce your usage
- Use the event feed or webhooks instead of repeatedly listing everything.
- Use
page_size=100when reading many records. - Filter on the server, for example
scheduled_dateon jobs, instead of downloading everything and filtering yourself.