Local AI coding agents hit timeouts with LM Studio

This highlights the growing pains of running autonomous agents on local hardware, where prompt prefill can take minutes instead of milliseconds. Default API timeouts in agent frameworks are tuned for fast cloud APIs and fail on slow local inference. Long context windows (30k+ tokens) exacerbate the problem by drastically increasing the time LM Studio spends reading the prompt before generating the first token. The fix requires manually increasing request timeouts in the client extension settings or config files to 10-30 minutes.