A Model Context Protocol (MCP) server that provides access to Linux man pages using FastMCP. This server allows AI assistants to search, retrieve, and explore system documentation directly from your local machine.

• Search man pages: Find documentation by keyword or command name using apropos
• Retrieve specific pages: Get complete man page content by name and optional section
• List sections: Browse available man page sections (1-9) with descriptions
• Clean formatting: Man page content is cleaned and formatted for AI consumption
• Async operations: All operations are asynchronous with timeout protection
• MCP Resources: Expose man pages as resources with man:// URIs

# Clone the repository
git clone https://github.com/guyru/man-mcp-server.git
cd man-mcp-server

# Install dependencies
uv sync

# Install with development tools
uv sync --extra dev

# Clone the repository
git clone https://github.com/guyru/man-mcp-server.git
cd man-mcp-server

# Install the package and dependencies from pyproject.toml
pip install .

# Or install in development mode (editable install)
pip install -e .

# For development with optional dependencies
pip install -e .[dev]

# Using uv
uv run python3 man_server.py

# Using python directly
python3 man_server.py

# With MCP development tools
uv run mcp dev man_server.py

To integrate this MCP server with VS Code, you need to create a configuration file:

1. Create the VS Code MCP configuration directory (if it doesn't exist):

   mkdir -p .vscode

2. Create .vscode/mcp.json with the following content:

   {
     "servers": {
       "man-mcp-server": {
         "type": "stdio",
         "command": "uv",
         "args": ["run", "--directory", "/path/to/man-mcp-server", "python3", "man_server.py"]
       }
     }
   }

3. Update the path in the configuration above to match your actual project directory.

4. Alternative configuration if you're not using uv:

   {
     "servers": {
       "man-mcp-server": {
         "type": "stdio",
         "command": "python3",
         "args": ["/path/to/man-mcp-server/man_server.py"]
       }
     }
   }

This configuration allows MCP-compatible VS Code extensions to communicate with your man pages server.

Search for man pages by keyword or topic:

search_man_pages("permission")  # Find pages about permissions
search_man_pages("network")     # Find networking-related pages

Retrieve the full content of a specific man page:

get_man_page("ls")           # Get ls man page (any section)
get_man_page("chmod", "1")   # Get chmod from section 1 specifically
get_man_page("printf", "3")  # Get printf from section 3 (C library)

List all available man page sections with descriptions:

list_man_sections()  # Shows sections 1-9 with descriptions

The server exposes man pages as resources using man:// URIs:

• man://sections - List of all available sections
• man://search/{keyword} - Search results for a keyword
• man://{section}/{page} - Specific man page content

Examples:

• man://search/network - Search results for "network"
• man://1/ls - The ls command man page from section 1
• man://3/printf - The printf function man page from section 3

• Operating System: Linux with standard man page system
• Python: 3.10 or higher
• Commands: man, apropos (usually pre-installed)
• Dependencies: mcp library

# Test search functionality
uv run python3 -c "
from man_server import man_service
import asyncio
print(asyncio.run(man_service.search_man_pages('ls')))
"

# Test page retrieval
uv run python3 -c "
from man_server import man_service
import asyncio
content = asyncio.run(man_service.get_man_page('ls', '1'))
print(content[:200] + '...')
"

# Run with MCP development tools for debugging
uv run mcp dev man_server.py

The server includes comprehensive error handling:

• Missing pages: Graceful handling with informative error messages
• Timeout protection: Subprocess calls are protected with configurable timeouts
• Fallback methods: If apropos fails, falls back to man -k
• Content validation: Ensures retrieved content is not empty
• Clean formatting: Removes ANSI codes and formatting for AI consumption

This project is licensed under the MIT License - see the LICENSE file for details.

Contributions are welcome! Please feel free to submit a Pull Request.