Lmstudio MCP

🚀 LMStudio-MCP Server

A Model Control Protocol (MCP) server that allows Claude to communicate with locally running Large Language Models (LLMs) through LM Studio.

🚀 Quick Start

LMStudio-MCP creates a bridge between Claude (with MCP capabilities) and your locally running LM Studio instance. This enables Claude to:

• Check the health status of the LM Studio API.
• List all available models.
• Get the currently loaded model.
• Generate completions using your local model.

In this way, you can combine Claude's capabilities with your own private models and fully utilize the locally running models.

✨ Features

The bridge provides the following features:

• health_check(): Verify if the LM Studio API is accessible.
• list_models(): Get a list of all available models in LM Studio.
• get_current_model(): Determine the currently loaded model.
• chat_completion(prompt, system_prompt, temperature, max_tokens): Generate text using your local model.

📦 Installation

Prerequisites

• Python 3.7+
• LM Studio installed and running
• A model loaded in LM Studio
• Claude with MCP access
• Required Python packages (see installation steps)

Steps

1. Clone this repository:

   git clone https://github.com/infinitimeless/LMStudio-MCP.git
   cd LMStudio-MCP
   

2. Install the required packages:

   pip install requests "mcp[cli]" openai
   

📚 Documentation

MCP Configuration

To enable Claude to connect to this bridge, you need to correctly configure the MCP settings. You can:

1. Use directly from GitHub:

   {
     "lmstudio-mcp": {
       "command": "uvx",
       "args": [
         "https://github.com/infinitimeless/LMStudio-MCP"
       ]
     }
   }
   

2. Use local installation:

   {
     "lmstudio-mcp": {
       "command": "/bin/bash",
       "args": [
         "-c",
         "cd /path/to/LMStudio-MCP && source venv/bin/activate && python lmstudio_bridge.py"
       ]
     }
   }
   

For detailed MCP configuration instructions, please refer to MCP_CONFIGURATION.md.

Usage

1. Start your LM Studio application and ensure it is running on port 1234 (default port).
2. Load a model in LM Studio.
3. If running locally (without using uvx), start the LMStudio-MCP server:

   python lmstudio_bridge.py
   

4. In Claude, select "lmstudio-mcp" when prompted to connect to the MCP server.

Known Limitations

• Some models (e.g., phi-3.5-mini-instruct) may have issues.
• Only models with a file size of no more than 20MB are supported.
• Loading times may be long for larger models.

Troubleshooting

If you encounter problems, please check the following:

1. Ensure that LM Studio is correctly installed and running on the default port (1234).
2. Check if there are any firewall or network restrictions preventing access to the port.
3. Verify that the model has been successfully loaded into LM Studio.
4. Check the LM Studio console output for more information.

For more detailed information, please refer to the Troubleshooting Guide.

📄 License

Contributors

• Author: [Your Name]
• Other contributors: ...

Thanks to all who have contributed to this project!