Rate limit (agents)
Also known as: throttling
A rate limit caps how often or how much an agent can act or call an API.
- Stops runaway loops from hammering a service.
- Bounds cost on metered APIs.
- Also raises the cost of model-extraction attacks.
Design it into the harness: an agent with no limits will find them for you — usually as a bill or a blocked IP. Set explicit caps (per minute, per task, per day) before you need them.