What is Rate limiting?

Capping how often an action can happen in a window — how many API calls, requests, or agent steps — to control cost and prevent runaway loops.

How it works

Agents can spiral: a stuck loop burns tokens and money fast. Rate limits on tool calls, a hard iteration cap, and per-request cost ceilings turn a potential disaster into a bounded, observable cost.

Example

In practiceYou cap the agent at 10 iterations and 50 tool calls per task, so a bug can't spin it into a huge bill overnight.

What limits should every agent have?

A max iteration count, a tool-call budget, and a cost ceiling. Without them, one bug becomes an unbounded bill.

Go deeper: Deploy Your First AI Agent

The complete, step-by-step guide — every command, config, and the reasoning behind it.

Get the guide