Performance of the tools doesn't really matter when you have a full LLM inference loop in between each tool call.
replyI still find it revolting they're writing this stuff in typescript.
replyIn practice it is actually extremely fast because there is no process fork. You're talking nanoseconds for common commands
reply[Disclaimer: I made the thing]