Claude-LMStudio Bridge

What is Claude-LMStudio Bridge?

An MCP server that bridges Claude with local LLMs running in LM Studio. It allows Claude to list available models, generate text, perform chat completions, and verify connectivity with your local LM Studio instance.

How to use Claude-LMStudio Bridge?

Install by cloning the repository and running the setup script (setup.sh on macOS/Linux, setup.bat on Windows). The script guides you through configuring Claude Desktop with the bridge. Alternatively, manually create a virtual environment, install dependencies from requirements.txt, and configure a new MCP server in Claude Desktop with the correct command and arguments. Ensure LM Studio is running with its API server enabled (default port 1234). You can customize host, port, and debug logging via a .env file using LMSTUDIO_HOST, LMSTUDIO_PORT, and DEBUG variables.

Key features of Claude-LMStudio Bridge

• List all available models in LM Studio
• Generate text using your local LLMs
• Support for chat completions through local models
• Health check tool to verify LM Studio connectivity
• Bridges Claude Desktop with local LLM infrastructure

Use cases of Claude-LMStudio Bridge

• Check if your LM Studio server is running and reachable
• List available models loaded in your local LM Studio
• Generate text (e.g., a poem) using a local LLM
• Send chat completion queries to your local model from Claude

FAQ from Claude-LMStudio Bridge

What are the prerequisites to use Claude-LMStudio Bridge?

You need Claude Desktop with MCP support, LM Studio installed and running locally with its API server enabled, and Python 3.8 or later.

What should I do if Claude cannot connect to LM Studio?

Ensure LM Studio is running, the API server is enabled (Settings > API Server), and the port in your .env file (default 1234) matches LM Studio’s port. Use the provided debug_lmstudio.py script to diagnose issues.

What if Claude reports that no models are loaded?

Open LM Studio, load a model, and verify it is running successfully. The bridge can only list models that are currently loaded in LM Studio.

How do I fix a "MCP package not found" error?

Reinstall the required packages with pip install "mcp[cli]" httpx python-dotenv. Make sure you are using Python 3.8 or later and that the virtual environment is active.

Can I change the default host and port used by the bridge?

Yes. Create a .env file with LMSTUDIO_HOST and LMSTUDIO_PORT set to your preferred values. The defaults are 127.0.0.1 and 1234.