RandomUser MCP Server and Chatbot

This repository contains a RandomUser MCP Server built with FastAPI, a custom chatbot integrating MCP tools with Gemini-2.0-Flash, and a proxy to connect the server to Claude Desktop. The project leverages the Model Context Protocol (MCP) to extend AI capabilities with tools for generating random user profiles, demonstrating a practical application of MCP in both standalone and integrated environments.

This project uses GOOGLE GEMINI LLM, click there to get your free API keys: https://aistudio.google.com/

Overview

The RandomUser MCP Server fetches user data (name, country, age, etc.) from https://randomuser.me/api/ and exposes three tools:

• get_random_user: Fetch a single random user profile.
• get_multiple_users: Fetch multiple user profiles (1-10).
• get_user_by_gender: Fetch a user profile by gender (male/female).

The Chatbot uses these tools via HTTP requests or falls back to Gemini-2.0-Flash for general queries. The Proxy bridges the HTTP server to Claude Desktop's stdio-based MCP interface, enabling tool usage within Claude.

Features

• MCP Server: HTTP-based server with FastAPI, exposing tools via /mcp/tools and /mcp/call_tool.
• Chatbot: Command-line interface blending MCP tools with Gemini-2.0-Flash responses.
• Claude Integration: Proxy script to connect the server to Claude Desktop.
• Extensible: Easy to add new tools or modify existing ones.

Prerequisites

• Python 3.8+
• Claude Desktop (optional, for MCP integration)
• Gemini API Key (for chatbot LLM functionality)

Installation

1. Clone the Repository:

   git clone https://github.com/shelwyn/MCP_with_python.git
   cd MCP_with_python
   

2. Install Dependencies:

   pip install -r requirements.txt
   

3. Set Up Environment: Create a .env file in the root directory:

   GEMINI_API_KEY=your_gemini_api_key_here
   

   Replace your_gemini_api_key_here with your actual Gemini API key from Google.

File Structure

randomuser-mcp/
├── mcp_server.py           # HTTP-based MCP server
├── chatbot.py              # Custom chatbot with MCP and Gemini
├── proxy.py                # Stdio-to-HTTP proxy for Claude Desktop
├── .env                    # Environment variables (not tracked)
├── requirements.txt        # Project dependencies
├── claude_desktop_config.json  # Claude Desktop MCP config (example)
└── README.md               # This file

Usage

1. Running the MCP Server

Start the HTTP server:

python mcp_server.py

• Runs on http://localhost:8000
• Endpoints: /mcp/sse, /mcp/tools, /mcp/call_tool

2. Running the Chatbot

Launch the chatbot:

python chatbot.py

Examples:

You: Get me a random user
Bot: Full Name: Mr John Doe
     Country: United States
     Age: 34
     ...

You: Tell me about the sun
Bot: The Sun is a star at the center of our solar system...

You: exit

3. Integrating with Claude Desktop

1. Start the MCP Server:

   python mcp_server.py
   

2. Configure Claude: Edit claude_desktop_config.json (e.g., C:\Users\<YourUsername>\AppData\Roaming\Claude\claude_desktop_config.json):

   {
     "mcpServers": {
       "randomuser_proxy": {
         "command": "python",
         "args": ["C:/path/to/proxy.py"]
       }
     }
   }
   

   Update the path to proxy.py.

3. Restart Claude Desktop:

   • Close and reopen Claude Desktop.
   • Look for the hammer icon (🔨) in the input box.
   • Test with: "Use get_random_user" or "Call get_multiple_users with count 3".

Requirements

requests==2.31.0
fastapi==0.103.2
uvicorn==0.23.2
sse-starlette==1.6.5
mcp[cli]==0.1.0
google-generativeai==0.4.0
python-dotenv==1.0.0

How It Works

MCP Server (mcp_server.py)

• Tools: Defined in RandomUserTool class, fetching data from randomuser.me.
• FastAPI: Hosts HTTP endpoints, with MCP logic from mcp.server.Server.
• Endpoints:
  • /mcp/tools: Lists tools in MCP format.
  • /mcp/call_tool: Executes tools with POSTed arguments.

Chatbot (chatbot.py)

• Tool Detection: Parses input for keywords (e.g., "random user") to call MCP tools via HTTP.
• LLM Fallback: Uses Gemini-2.0-Flash for non-tool queries.
• HTTP Client: Communicates with http://localhost:8000.

Proxy (proxy.py)

• Bridge: Translates Claude's stdio JSON-RPC requests to HTTP calls.
• Methods: Handles initialize, tools/list, tools/call, and returns empty responses for resources/list and prompts/list.
• Runs as Subprocess: Launched by Claude Desktop via config.

Troubleshooting

No Hammer Icon in Claude:

• Ensure mcp_server.py is running before starting Claude.
• Check logs in C:\Users\<YourUsername>\AppData\Roaming\Claude\Logs.

Chatbot Errors:

• Verify .env has a valid GEMINI_API_KEY.
• Ensure MCP server is running.

Proxy Issues:

• Confirm the path in claude_desktop_config.json is correct.
• Add print("Debug: ...", file=sys.stderr) in proxy.py for logs.

Contributing

1. Fork the repository.
2. Create a feature branch (git checkout -b feature/new-tool).
3. Commit changes (git commit -m "Add new tool").
4. Push to the branch (git push origin feature/new-tool).
5. Open a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Anthropic: For Claude Desktop and MCP support.
• Google: For Gemini-2.0-Flash API.
• RandomUser.me: For the free user data API.

Contact

For questions or suggestions, open an issue or reach out via GitHub.