Developers

MCP Rate Limits

Limits per token and how to handle them.

  • 60 tool calls/minute per token
  • 10 calls/minute for calculate_scenario and get_path_to_target
  • Bursts above the limit receive a JSON-RPC error result

Back off exponentially when you are limited. Sync freshness, not call volume, is what usually gates answer quality — check get_data_coverage first.