A CLI remote for LM Studio’s desktop LLM engine

What it does

lms is the official command-line companion to LM Studio, the desktop app for running local large language models. It lets you check the app’s status, list downloaded and loaded models, manage the local API server, and scaffold new SDK projects without touching the GUI. It ships with LM Studio 0.2.22 and newer, though a Node-based bootstrap installer is available if the binary does not land in your path.

The interesting bit

The CLI is built for automation, not just convenience. Flags like --json on model listings emit machine-readable output, and you can load or unload models without confirmation prompts. That makes it practical for shell scripts that need to query or orchestrate a local inference engine.

Key highlights

• Ships bundled with LM Studio, with a fallback installer if the binary misses your path
• Start, stop, and monitor the local API server from the terminal
• Machine-readable JSON output for model listings and loaded-model status
• Includes a project scaffolding command for the LM Studio SDK
• Streams application logs via a dedicated subcommand

Caveats

• Requires the LM Studio desktop app; it is not a standalone LLM runner
• Cannot be built standalone—it lives inside the lmstudio.js monorepo

Verdict

Worth a look if you already use LM Studio and want to automate model management or integrate it into terminal-based workflows. Skip it if you need a self-contained CLI inference engine or do not use the LM Studio ecosystem.