you can see how different models interpret tool descriptions.
How's this done? I saw the creator of MCP recommended¹ "investing heavily in tool descriptions" but it wasn't clear exactly how to
OpenTools is hiring: andy@opentools.com
https://opentools.com