Installation

Installation

This guide will help you set up UV-MCP on your system and configure it for use with your MCP client (like Claude Desktop or Gemini CLI).

Prerequisites

Before installing UV-MCP, ensure you have the following:

1. uv: The uv package manager must be installed.
   • macOS/Linux: curl -LsSf https://astral.sh/uv/install.sh | sh
   • Windows: powershell -c "irm https://astral.sh/uv/install.ps1 | iex"
   • Official Installation Docs
2. Python: Python 3.10 or higher.

Installing the Server

You can install uv-mcp directly from the repository or via pip. Since it’s a tool intended to be run by an agent, we recommend installing it using uv tool.

Option 1: Using uv tool (Recommended)

uv tool install git+https://github.com/saadmanrafat/uv-mcp

This installs uv-mcp in an isolated environment and makes the uv-mcp command available.

Option 2: Using pip

pip install uv-mcp

(Note: Adjust command if installing from a specific source or local path)

Configuration

To use UV-MCP with an AI client, you need to add it to your MCP configuration file.

Finding your mcp-config.json

• Claude Desktop App: usually located at ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows).
• Gemini CLI: typically uses a configuration file passed via flags or located in ~/.gemini/config.json.

Configuration Snippet

Add the following to your mcpServers object:

{
  "mcpServers": {
    "uv-mcp": {
      "command": "uv",
      "args": [
        "tool",
        "run",
        "uv-mcp"
      ]
    }
  }
}

If you are developing locally or running from source:

{
  "mcpServers": {
    "uv-mcp": {
      "command": "uv",
      "args": [
        "--directory",
        "/absolute/path/to/uv-mcp",
        "run",
        "uv-mcp"
      ]
    }
  }
}

Verifying Installation

Once configured, restart your AI client. You should be able to ask the agent:

“Check if uv is installed”

If the agent responds with the version of uv, the connection is successful.