Rate limits
Request limits per API key or connected app.
Each API key or OAuth grant can make 300 requests per minute, counted in one-minute windows. MCP tool calls and REST requests share the same limit.
When you exceed it, the API returns 429 rate_limited with a Retry-After header (seconds) and retry_after_seconds in the body. Wait that long, then retry. The SDK and CLI do this automatically.
AI operations – testing an agent, re-running the AI on a conversation, translating – also consume AI credits of your plan, just like in the dashboard. Check usage with GET /stats/usage.