Rate limits

Telem does not publish hard rate limits today. There is no per-second or per-day request cap documented for the hosted deployment, and nothing in the SDK, the MCP server, or the agent-tool plugins currently implements client-side throttling, retry, or backoff — a failed or slow request is surfaced as-is (see Errors), never silently retried.

That doesn’t mean unlimited: API keys may carry account-level quotas set for your specific deployment, independent of anything published here. If you’re planning for high volume, contact the Telem team to talk through your needs before you build against an assumption either way.

If a request is throttled

There is no dedicated exception for a 429 response. Like any other status code outside 400/401/403/404, it would raise the SDK’s generic APIStatusError — see the full status code / exception table for how every other failure mode maps.

What the SDK does today

The Python client sets a request timeout (60 seconds by default, configurable via the timeout argument) and raises TelemConnectionError if a request times out or the transport fails before a response arrives — it does not retry automatically. If your use case needs retry-with-backoff around rate limiting or transient failures, implement it at the call site using the status_code on the raised exception.

This page reflects the current state of the project and will be updated if published limits or built-in retry/backoff behavior ship.