DocumentBrain

An MCP (Model Context Protocol) server for reading, converting, and saving documents. Built with FastMCP and MarkItDown, it enables LLMs to interact with files on your local machine.

Features

Tools

Tool	Description
Read Any Document	Reads any supported document and returns its text content, including OCR for images.
Save File to PC	Saves content to a file at a specified path on your machine.

Resources

Resource	URI	Description
Local Document Directory	docs://files	Lists all files in the current working directory.
Document Content	docs://file/{filename}	Retrieves the converted text content of a specific file.

Prompts

Prompt	Description
analyze_data	Generates a creative four-line song based on user-provided text.

Installation

Prerequisites

• Python >= 3.12

Install from source

pip install -e .

Install from PyPI

pip install vvin-document-brain

Usage

Run the MCP server

mcp-document-brain

Configure in Cursor

Add the following to your Cursor MCP config file at ~/.cursor/mcp.json:

Using uvx (recommended — no install needed):

{
  "mcpServers": {
    "DocumentBrain": {
      "command": "uvx",
      "args": ["vvin-document-brain"]
    }
  }
}

Cache path issues? If uvx fails with a "Permission denied" error for ~/.cache/uv or ~/.local/share/uv (common on managed/corporate machines), add an env block to redirect both paths:

{
  "mcpServers": {
    "DocumentBrain": {
      "command": "uvx",
      "args": ["vvin-document-brain"],
      "env": {
        "UV_CACHE_DIR": "/tmp/cache/uv",
        "XDG_DATA_HOME": "/tmp/uv-data"
      }
    }
  }
}

Using pip install (if already installed globally):

{
  "mcpServers": {
    "DocumentBrain": {
      "command": "vvin-document-brain"
    }
  }
}

Configure in Claude Desktop

Add the following to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "DocumentBrain": {
      "command": "uvx",
      "args": ["vvin-document-brain"]
    }
  }
}

Example tool usage

Read a document:

Tool: read_any_document
Arguments: { "file_path": "~/Documents/report.pdf" }

Save a file:

Tool: save_file_to_pc
Arguments: { "filepath": "/path/to/output.txt", "content": "Hello, world!" }

Project Structure

DocumentBrain/
├── src/
│   └── document_brain/
│       ├── __init__.py
│       └── server.py        # MCP server with tools, resources, and prompts
├── tests/
│   └── test_server.py
├── pyproject.toml
└── README.md

Supported File Formats

DocumentBrain leverages MarkItDown to convert a wide range of file formats, including:

• PDF
• Word documents (.docx)
• Excel spreadsheets (.xlsx)
• PowerPoint presentations (.pptx)
• Images (with OCR)
• HTML
• Plain text
• And more (see MarkItDown docs for the full list)

Development

# Install dev dependencies
pip install -e ".[dev]"

# Run tests
pytest

License

MIT

Author

Vishwesh Vinchurkar — vishwesh.vinchurkar@gmail.com