agent frameworks don't rate limit tool calls per agent
1. tested langchain 0.3.14, autogen 0.4.2, crewai 0.86.0 2. none of them rate limit tool calls per agent or per conversation 3. if your tool hits an external api with rate limits the agent just hammers it until the api returns 429 and then the whole agent loop fails we had to build our own rate limiting wrapper around every tool that tracks calls per minute and backs off when we hit limits. seems like something the framework should handle by default
langchain 0.3.14 and autogen both have this gap. you can make 50 tool calls per second and the framework will happily queue all of them. we added a rate limiter middleware that tracks calls per agent instance and returns synthetic "rate limited" responses after threshold. honestly this should be in the framework by default
+1 we added same middleware last month
which frameworks did you test? langchain and autogen both have this gap?