MCP Server - FastAPI on Azure App Service

This is a Python implementation of the MCP (Model Context Protocol) server using FastAPI and deployed to Azure App Service, providing multiplication, temperature conversion, and weather tools.

Features

• Multiplication Tool: Multiply two numbers
• Temperature Converter Tools: Convert between Celsius and Fahrenheit
• Weather Tools: Get US weather alerts by state and forecasts by coordinates

Local Development

Prerequisites

• Python 3.11+
• uv for faster dependency management and virtual environment creation
• Node.js & npm (for running MCP Inspector)

Setup and Running

1. Create a virtual environment:

cd src
uv venv
# Activate the virtual environment
source .venv/bin/activate  # On macOS/Linux
# OR
.venv\Scripts\activate     # On Windows

1. Install dependencies:

uv pip install -r requirements.txt

1. Run the development server with auto-reload:

uvicorn main:app --reload

The server will start on http://localhost:8000 by default.

API Endpoints

• / - Root endpoint showing server status
• /health - Health check endpoint
• /mcp-server/mcp - MCP server endpoint for tool interactions

Testing with MCP Inspector

1. Start the server (if not already running):

uvicorn main:app --reload

1. Run MCP Inspector:

npx @modelcontextprotocol/inspector

1. Connect to the MCP Server:

   • Transport Type: streamable-http
   • URL: http://localhost:8000/mcp-server/mcp
   • Authentication: None

2. Available Tools:

   • multiply - Multiply two numbers
   • celsius_to_fahrenheit - Convert Celsius to Fahrenheit
   • fahrenheit_to_celsius - Convert Fahrenheit to Celsius
   • get_alerts - Get US weather alerts by state
   • get_forecast - Get weather forecast by coordinates

Azure Deployment

Azure Prerequisites

• Azure CLI installed (az)
• Azure Developer CLI installed (azd)
• Azure subscription

Deployment Process

1. Login to Azure:

az login
azd auth login

1. Deploy using Azure Developer CLI:

# From the project root directory
azd up

This will:

• Create a new resource group
• Deploy an App Service Plan (Linux)
• Deploy the Python web app
• Configure all necessary settings

Post-Deployment

After deployment:

1. Check the application health: https://<your-app>.azurewebsites.net/health
2. View available tools: https://<your-app>.azurewebsites.net/
3. Monitor logs in Azure Portal or using Azure CLI

Connecting to the MCP Server

The MCP endpoint is available at:

https://<your-app>.azurewebsites.net/mcp-server/mcp

To connect using MCP Inspector or other MCP clients:

1. Use the full URL: https://<your-app>.azurewebsites.net/mcp-server/mcp
2. Select "streamable-http" as the transport type
3. No authentication is currently required (the endpoint is publicly accessible)

Security Considerations

• No authentication is currently implemented (public endpoints)
• CORS is wide open for development
• Consider implementing API key or OAuth2 authentication for production
• Validate all user inputs in tools to prevent injection attacks

Troubleshooting

• Check logs: az webapp log tail --name <app-name> --resource-group <rg-name>
• SSH into container: az webapp ssh --name <app-name> --resource-group <rg-name>
• Restart app: az webapp restart --name <app-name> --resource-group <rg-name>

Resources

• Azure Developer CLI Documentation
• FastAPI Documentation
• MCP Protocol Documentation
• Uvicorn Documentation
• MCP Inspector Documentation
• Azure App Service Documentation
• FastMCP Documentation