UV Documentation MCP Server

An MCP server that provides programmatic access to UV documentation through a standardized API. This server now leverages a modern, hierarchical resource structure that exposes both high‐level documentation sections and detailed command documentation (including subsections) as individual MCP resources. Tools have been streamlined, and many documentation queries are handled via the resource layer.

Links

• UV Documentation
• MCP Python SDK

Quick Start

1. Requirements:

Ensure Python 3.13+ is installed (minimum 3.10 for MCP SDK compatibility).

2. Clone the Repository:

git clone https://github.com/StevenBtw/uv-docs-mcp.git
   cd uv-docs-mcp

3. Set Up Virtual Environment and Install Dependencies:

python3.13 -m venv env
   source env/bin/activate   # On Windows use: env\\Scripts\\activate
   pip install -e .

4. Run the Server:

uv run uv-docs

*(If using Claude Desktop or other MCP clients, adjust configuration accordingly.)*

Architecture

The server is built with a modular and hierarchical design that clearly separates concerns:

flowchart TD
    Client[MCP Client] --> Server[Server Layer]
    Server --> Resources[Resource Layer]
    Server --> Tools[Tools Layer]
    Server --> Prompts[Prompts Layer]
    
    Resources --> Cache[Cache System]
    Tools --> Cache

• Resource Layer:

Exposes UV documentation via URIs with the scheme uv-docs://. Resources are structured hierarchically:

• Level 1: Base sections (e.g., uv-docs://cli, uv-docs://settings, uv-docs://resolver)
• Level 2: Individual commands (e.g., uv-docs://cli/uv-run, uv-docs://cli/uv-cache)
• Level 3: Command subsections (e.g., uv-docs://cli/uv-cache/uv-cache-dir)

• Tools Layer:

Provides operational commands to interact with the cache and perform real-time searches.

• update_cache: Updates the documentation cache if the live version differs.
• search_documentation: Searches UV documentation using real-time query execution.

• Prompts Layer:

Supplies prompt templates for AI assistants to generate documentation summaries (e.g., the summarize-docs prompt).

Components

Resources

Resources are accessed using the uv-docs:// URI scheme and provide the full hierarchy of documentation. For example:

• Base Section:

uv-docs://cli returns the list of all CLI commands.

• Command Level:

uv-docs://cli/uv-cache returns metadata about the uv cache command, including available documentation sections.

• Subsection Level:

uv-docs://cli/uv-cache/uv-cache-dir returns the detailed documentation content for the “uv cache dir” section.

Each resource returns JSON data with detailed metadata and, when requested, the full documentation content. This fine-grained breakdown allows MCP clients and AI assistants to query only the information they require.

Tools

The available tools have been streamlined:

1. update_cache

• Purpose: Update the documentation cache if the live version has changed.
• Input:

{
       "force": false
     }

2. search_documentation

• Purpose: Perform real-time searches over the UV documentation.
• Input:

{
       "query": "search query"
     }

Prompts

The server provides pre-built prompt templates for common documentation tasks:

• summarize-docs:

Creates a summary of UV documentation sections.

Optional Arguments:

{
    "section": "cli|settings|resolver"
  }

If no section is specified, summaries for all sections are generated.

Installation & Configuration

MCP Server Configuration

For Claude Desktop or other systems, modify the MCP server settings appropriately:

Development/Unpublished Configuration

{
  "mcpServers": {
    "uv-docs": {
      "command": "uv",
      "args": [
        "--directory",
        "/Users/steven/Git/MCP/uv-docs",
        "run",
        "uv-docs"
      ]
    }
  }
}

Published Configuration

{
  "mcpServers": {
    "uv-docs": {
      "command": "uvx",
      "args": [
        "uv-docs"
      ]
    }
  }
}

Debugging & Inspection

For detailed debugging, use the MCP Inspector:

npx @modelcontextprotocol/inspector uv --directory /Users/steven/Git/MCP/uv-docs run uv-docs

Caching

The server implements a version-aware caching system to optimize documentation retrieval:

• Initialization: The cache is automatically initialized at startup.
• Version Checking: The live documentation version is checked against the cached version.
• Segmented Caching: Each documentation section (e.g., CLI, settings, resolver) is cached separately.
• Manual Updates: Use the update_cache tool when necessary.

Development

Project Structure

uv-docs/
├── src/
│   └── uv_docs/
│       ├── __init__.py
│       ├── server.py         # Server setup and routing
│       ├── resources.py      # Resource handling and MCP resource interface
│       ├── cache.py          # Version-aware cache system
│       ├── prompts.py        # Prompt templates for documentation tasks
│       └── tools.py          # Consolidated tool implementations (update_cache, search_documentation)
├── pyproject.toml            # Project configuration
└── README.md                # Project overview (this file)

Building & Publishing

1. Sync Dependencies & Lockfile:

uv sync

2. Build Package Distributions:

uv build

3. Publish to PyPI:

uv publish

Note: Set PyPI credentials via:

• Token: --token or environment variable UV_PUBLISH_TOKEN
• Or username/password: --username/UV_PUBLISH_USERNAME and --password/UV_PUBLISH_PASSWORD

Debugging

For the best debugging experience, use the MCP Inspector:

npx @modelcontextprotocol/inspector uv --directory /Users/steven/Git/MCP/uv-docs run uv-docs

The Inspector will display a URL for browser-based debugging.

Caching

The server implements a version-aware caching system:

• Cache initializes on server start
• Version checks against live documentation
• Automatic updates when version changes
• Manual updates via update_cache tool
• Each documentation section cached separately