Description

This MCP server, developed by dmayboroda, enables AI assistants like Claude to perform RAG (Retrieval-Augmented Generation) on local files. Built in Python and leveraging the uv package manager, it integrates seamlessly with MCP clients through a simple configuration file. The server allows the client to search and retrieve context from the user's local documents, enhancing its ability to provide relevant and personalized responses. By granting access to private or offline information sources, this implementation extends the client's knowledge base beyond its training data, making it particularly useful for tasks involving confidential documents, personal notes, or frequently updated local resources.

Installation 

git clone https://github.com/dmayboroda/minima/tree/HEAD/mcp-server

Claude Desktop Configuration

Add this configuration to your Claude Desktop config file to enable this MCP server:

Config file location: ~/Library/Application Support/Claude/claude_desktop_config.json

Config file location: %APPDATA%\Claude\claude_desktop_config.json

Config file location: ~/.config/Claude/claude_desktop_config.json

claude_desktop_config.json 
{
  "mcpServers": {
    "minima-local-rag": {
      "command": "node",
      "args": [
        "path/to/server.js"
      ]
    }
  }
}