lmstudio-ai/lms

lmstudio-ai/lms – Command‑line interface for LM Studio

What it is – lms is a small CLI that talks to the LM Studio desktop application (a local LLM development environment). It lets you start/stop the built‑in API server, list and load models that LM Studio has downloaded, and perform a few management tasks from the terminal.

Key capabilities

• Status & server control – lms status shows whether LM Studio is running; lms server start|stop toggles the local inference server.
• Model inventory – lms ls lists all models stored on disk (with --json for machine‑readable output or --decision to filter decision‑type models). lms ps shows which models are currently loaded for inference.
• Loading & unloading – lms load <path> loads a model into the server, optionally with -y to force maximum GPU acceleration without prompts. lms unload <id> removes a specific model, or --all clears everything.
• Project scaffolding – lms create boots a new project using the LM Studio SDK, useful for developers building extensions or custom pipelines.
• Log streaming – lms log stream pipes LM Studio’s logs to the console for quick debugging.

How to get it – The CLI is bundled with LM Studio 0.2.22+. If the lms command isn’t on your PATH, run npx lmstudio install-cli to install it globally.

Who it’s for – Anyone who prefers a terminal workflow when working with LM Studio: data scientists, ML engineers, or hobbyists who want to script model loading, start the inference server, or integrate LM Studio into CI pipelines.

Development notes – The CLI lives inside the lmstudio.js monorepo and cannot be built in isolation. To contribute, clone the full repo, run npm install && npm run build, then invoke the built CLI via node publish/cli/dist/index.js.

---

All details are taken directly from the repository’s README.