Uiverse Tool

English | 中文

A FastMCP-based code extraction tool for Uiverse.io, supporting component code extraction from Uiverse website for various frontend frameworks.

Features

• 🎨 Support for multiple frontend framework code extraction
• 🚀 Built on FastMCP, supporting both stdio and SSE transport methods
• 🔧 Easy integration with AI workflows
• 📦 Uses Playwright for web content extraction

Supported Frameworks

• HTML/CSS
• React
• Vue
• Svelte
• Lit

System Requirements

• Python >= 3.12
• uv package manager

Installation

1. Clone or download the project

git clone https://github.com/YBMecho/Uiverse_MCP.git
cd Uiverse_MPC

2. Install dependencies

Using uv to install dependencies:

uv sync

3. Install Playwright browsers

uv run playwright install

Configuration

This tool supports two MCP configuration methods: stdio and SSE. Please add one of the following configurations to your MCP configuration file (such as Cursor's mcp.json).

Method 1: stdio configuration (recommended)

{
  "mcpServers": {
    "uiverse-MCP-tool-stdio": {
      "name": "Uiverse MCP Tool",
      "type": "stdio",
      "description": "Uiverse MCP Tool",
      "isActive": true,
      "command": "uv",
      "args": [
        "--directory",
        "D:\\YBMecho\\Desktop\\Uiverse_MPC",
        "run",
        "app.py"
      ]
    }
  }
}

Note: Please replace D:\\YBMecho\\Desktop\\Uiverse_MPC with your actual project path.

Method 2: SSE configuration

{
  "mcpServers": {
    "uiverse-MPC-tool-SSE": {
      "name": "Uiverse MCP Tool",
      "type": "sse",
      "description": "Uiverse MCP Tool",
      "isActive": true,
      "url": "http://127.0.0.1:8000/sse"
    }
  }
}

Note: An HTTP server must be started before using the SSE method.

Usage

Available Tools

1. parse_and_extract

Extract component code from Uiverse links.

Input format: <framework> <link>

Examples:

HTML https://uiverse.io/Na3ar-17/evil-dragon-24
React https://uiverse.io/username/component-name
Vue https://uiverse.io/username/component-name
Svelte https://uiverse.io/username/component-name
Lit https://uiverse.io/username/component-name

Usage rules:

• Links must start with https://uiverse.io/
• Links must contain a specific component path (not just the domain)
• Framework names are case-insensitive

2. list_supported_frameworks

List all currently supported frameworks.

Returns: List of supported frameworks (Markdown format)

Using with AI Assistants

After configuration, you can use it directly in MCP-supporting AI assistants (like Cursor):

Please extract HTML code from https://uiverse.io/Na3ar-17/evil-dragon-24

or

Use React framework to extract code from https://uiverse.io/username/component-name

The AI assistant will automatically call the parse_and_extract tool to retrieve the code.

Project Structure

Uiverse_MPC/
├── app.py                  # FastMCP main application
├── browser.py              # HTML/CSS extractor
├── browser_React.py        # React extractor
├── browser_Vue.py          # Vue extractor
├── browser_Svelte.py       # Svelte extractor
├── browser_Lit.py          # Lit extractor
├── pyproject.toml          # Project configuration
├── build_exe.spec          # PyInstaller configuration
├── dist/                   # Executable output directory
│   └── UiverseExtractor.exe
└── README.md

Building Executable

The project includes PyInstaller configuration to build a standalone Windows executable:

pyinstaller build_exe.spec

The generated UiverseExtractor.exe will be located in the dist/ directory.

Dependencies

• mcp[cli] >= 1.16.0 - MCP protocol support
• fastmcp >= 2.0.0 - FastMCP framework
• playwright >= 1.55.0 - Browser automation

Development Notes

Local Running

Run in stdio mode:

uv run app.py

Adding Support for New Frameworks

1. Create a new browser_<Framework>.py file
2. Implement the extract_<framework>_code(url: str) -> str async function
3. Import and add it to the _dispatch_extract function in app.py
4. Update the SUPPORTED_FRAMEWORKS list

License

Please add the appropriate license information according to your needs.

Contributions

Issues and Pull Requests are welcome!

Contact

For questions or suggestions, please provide feedback through Issues.