←back to thread

MCP is eating the world

(www.stainless.com)
335 points emschwartz | 1 comments | | HN request time: 0.198s | source
1. fennecbutt ◴[] No.44377134[source]
Even MCP is still relying on tool/function calling in LLMs which is just a finetune/trained behaviour with zero guarantees.

It relies on training the model so that the probability of outputting the python_call token followed by the correct tool is relatively high. But it's still n<1.