RedNote MCP

# RedNote MCP - Xiaohongshu Content Search Tool ## Overview RedNote MCP is a Model Context Protocol (MCP) server for searching and retrieving content from Xiaohongshu (Red Book) platform. It provides intelligent content extraction with automatic login management and parallel processing capabilities. ## Features - **Smart Search**: Keyword-based content search on Xiaohongshu - **Auto Login**: Automatic cookie management and login handling - **Parallel Processing**: Efficient concurrent content retrieval - **Rich Data Extraction**: - Note titles and content - Author information and descriptions - Interaction metrics (likes, favorites, comments) - Images and hashtags - Direct note links ## Technical Stack - **Runtime**: Node.js with TypeScript - **Browser Automation**: Playwright - **Protocol**: Model Context Protocol (MCP) SDK - **Validation**: Zod schema validation - **Package Manager**: pnpm ## Data Structure ```typescript interface RedBookNote { title: string; // Note title content: string; // Note content author: string; // Author name authorDesc?: string; // Author description link: string; // Note URL likes?: number; // Like count collects?: number; // Favorite count comments?: number; // Comment count tags?: string[]; // Hashtag list images?: string[]; // Image URLs (WebP format) } ``` ## Installation ### Prerequisites - Node.js 18+ - pnpm package manager ### Setup 1. Clone the repository: ```bash git clone <repository-url> cd rednote-mcp ``` 2. Install dependencies: ```bash pnpm install ``` 3. Install Playwright browsers: ```bash pnpm exec playwright install ``` 4. Build the project: ```bash pnpm build ``` ## Usage ### Running the MCP Server ```bash pnpm start ``` ### Development Mode ```bash pnpm dev ``` ### Testing ```bash pnpm test ``` ## MCP Client Configuration ### Claude Desktop Add the following configuration to your Claude Desktop config file: **Windows**: `%APPDATA%\Claude\claude_desktop_config.json` **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json` ```json { "mcpServers": { "rednote-mcp": { "command": "node", "args": [ "C:\\ABSOLUTE\\PATH\\TO\\rednote-mcp\\build\\index.js" ] } } } ``` **For macOS/Linux users:** ```json { "mcpServers": { "rednote-mcp": { "command": "node", "args": [ "/absolute/path/to/rednote-mcp/build/index.js" ] } } } ``` Replace the path with your actual project directory. ### Other MCP Clients For other MCP-compatible clients, use the built server file: ```bash node build/index.js ``` ## Tool Usage Once configured, you can use the search tool in your MCP client: ``` Search for "food recommendation" on Xiaohongshu ``` The tool will return structured data including titles, content, author information, and images. ## Important Notes - **First Run**: Manual login to Xiaohongshu is required on first use - **Performance**: Initial searches may take 30-60 seconds due to browser startup and content loading - **Rate Limiting**: Concurrent requests are limited to 3 to avoid platform restrictions - **Image Format**: Images are provided in WebP format - **Cookie Management**: Login state is automatically saved and reused ## Development ### Project Structure ``` rednote-mcp/ ├── src/ │ ├── index.ts # MCP server entry point │ └── xiaohongshu.ts # Core scraping logic ├── cookies/ # Auto-generated cookie storage ├── results/ # Optional: saved search results ├── build/ # Compiled JavaScript output ├── package.json ├── tsconfig.json └── README.md ``` ### Available Scripts - `pnpm build` - Build TypeScript to JavaScript - `pnpm start` - Run the built MCP server - `pnpm dev` - Development mode with auto-reload - `pnpm test` - Run tests (if available) - `pnpm clean` - Clean build directory ## Troubleshooting ### Common Issues 1. **Login Required**: If you see login prompts, delete the `cookies/` directory and restart 2. **Timeout Errors**: Increase the MCP client timeout settings 3. **Browser Not Found**: Run `pnpm exec playwright install` to install browsers 4. **Permission Errors**: Ensure the project directory has proper read/write permissions ## License This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details. ## Disclaimer This tool is for educational and research purposes. Please respect Xiaohongshu's terms of service and rate limits when using this tool.