MyMCP Prompt

Description

MyMCP Prompt is a tool for generating Model Context Protocol (MCP) servers from natural language descriptions. This MVP uses the Google Gemini API to convert user descriptions into functional Python MCP servers with corresponding JSON configurations.

Project Structure

The application consists of:

• Flask Backend (Root Directory):
• app.py: Main Flask application setup (CORS, blueprint registration).
• api.py: Contains the /api/generate-mcp endpoint which interacts with the Google Gemini API to generate server code and configuration.
• requirements.txt: Lists Python dependencies.
• React Frontend (/frontend-vite Directory):
• Provides a web interface (frontend-vite/src/App.js) for users to input server descriptions.
• Displays the generated Python code and JSON configuration.
• Shows the paths where the generated files are saved.

Usage

1. Start the Backend Server:

• Ensure your virtual environment is activated in the root directory.
• Make sure the GEMINI_API_KEY environment variable is set.
• Run the Flask app:

# In the project root directory (mymcp)
flask run

• The backend will be available at http://127.0.0.1:5000.

2. Start the Frontend Development Server:

• Open a new terminal.
• Navigate to the frontend-vite directory:

cd frontend-vite

• Run the React app:

npm run dev

• The frontend will open automatically in your browser, usually at http://localhost:5173.

3. Use the Application:

• Open http://localhost:5173 in your browser.
• Enter a description for the MCP server you want to generate.
• Click "Place Your Order".
• The generated Python code and JSON configuration will be displayed, and the files will be saved to the generated_server directory (or a directory chosen by the LLM).

Change Log

• v0.1.0 (MVP): Initial release with basic MCP server generation using Google Gemini. Backend in root, frontend in /frontend.
• v0.2.0 (Vite Frontend & UI Redesign): Migrated frontend from Create React App to Vite for improved performance, reduced vulnerabilities, and better developer experience. Implemented a new French café-themed UI. Removed the output directory input from the frontend, allowing the backend to choose the output directory. Updated the backend API to handle requests without an output directory.

Future Features

• Integration with additional LLMs (OpenRouter, LiteLLM, OpenAI, Anthropic, SombaNova, Cerebras, LM Studio, Ollama, Groq).
• Support for generating tools/function calls within the MCP server.
• Support for generating agent prompts.
• Improved error handling and user feedback.
• More sophisticated MCP server code generation (e.g., using templates, better structure).
• UI enhancements.
• Unit tests for backend and frontend.