Each tool you expose is prompt text the model re-reads every turn —
run_terminal_command alone is ~2.3 KB — so narrow tools
are off by default. Whether they help a small model is an open question.
A local model on WebGPU drives a Linux VM through one
run_terminal_command tool. The folder you pick is mounted
read-only at /mnt/host. Weights load from
dist/models after make model, else from the
Hugging Face CDN. The VM-only page is at /.