yet again, another LLM integration that leaves the door wide open
for you to accidentally end up sending requests that ever end due to
not forcing use of stop tokens ... how many hours have i wasted on this
because developers of these tools don't seem to understand the api's
that they're building these tools around?