Weather Info App with MCP Server

A simple weather information application built with FastAPI and integrated with an MCP (Model Context Protocol) server for use with Gemini CLI.

📋 Requirements Checklist

• ✅ FastAPI weather application

• ✅ MCP Server implementation

• ✅ Gemini CLI integration

• ✅ MCP tools demonstration

• ✅ Screen recording (see SCREEN_RECORDING_GUIDE.md)

🎥 Screen Recording

IMPORTANT: This repository includes a screen recording demonstrating:

1. MCP server running

2. gemini mcp list command showing available tools

3. Usage of all MCP tools (get_weather, get_weather_batch, check_api_health)

See SCREEN_RECORDING_GUIDE.md for detailed recording instructions.

Project Structure

Features

• FastAPI Weather API: RESTful API providing weather information

• MCP Server: Exposes weather functionality as MCP tools

• Gemini CLI Integration: Ready to use with Google's Gemini CLI

• Multiple Tools: Get weather for single/multiple cities, health check

Installation

1. Clone this repository:

2. Install dependencies:

Running the Application

Step 1: Start the FastAPI Weather Server

In one terminal:

The API will be available at http://localhost:8000

You can test it:

Step 2: Configure Gemini CLI for MCP

The MCP server uses stdio transport. Create or update your Gemini CLI configuration file:

On Windows: %APPDATA%\Google\Gemini CLI\mcp_config.json

On macOS/Linux: ~/.config/google-gemini-cli/mcp_config.json

Example configuration:

For Windows, use full path like:

Step 3: Use with Gemini CLI

1. Start Gemini CLI

2. List available MCP tools:

3. Use the tools:

Available MCP Tools

1. get_weather

Get current weather information for a single city.

Parameters:

• city (required): Name of the city

• country (optional): Country name

Example:

2. get_weather_batch

Get weather information for multiple cities at once.

Parameters:

• cities (required): Comma-separated list of cities

Example:

3. check_api_health

Check if the weather API is running and healthy.

Example:

Testing

Run the demo script to test the setup:

API Endpoints

The FastAPI server provides:

• GET / - API information

• GET /health - Health check

• GET /weather?city=<name>&country=<name> - Get weather (GET)

• POST /weather - Get weather (POST with JSON body)

Screen Recording Instructions

To create a screen recording demonstrating the MCP server:

1. Start the FastAPI server: python weather_api.py

2. Open Gemini CLI

3. Show gemini mcp list command to see available tools

4. Demonstrate each tool:

   • get_weather for a single city

   • get_weather_batch for multiple cities

   • check_api_health

5. Show the responses and how they work together

Project Files

• weather_api.py - FastAPI weather application

• mcp_server.py - MCP server exposing weather tools

• demo.py - Testing and demonstration script

• get_path.py - Helper to get correct paths for configuration

• test_mcp_structure.py - Verify MCP imports and structure

• requirements.txt - Python dependencies

• mcp_config.json - Example Gemini CLI configuration

Documentation

• README.md - This file (main documentation)

• QUICK_START.md - Quick setup guide

• setup_instructions.md - Detailed setup instructions

• SCREEN_RECORDING_GUIDE.md - Guide for creating demo video

• PROJECT_SUMMARY.md - Complete project overview

Notes

• The weather data is mock/simulated for demonstration purposes

• Make sure the FastAPI server is running before using MCP tools

• The MCP server communicates with the FastAPI server via HTTP

• All paths in the configuration must be absolute paths

Troubleshooting

MCP server not connecting:

• Ensure FastAPI server is running on port 8000

• Check that the path to mcp_server.py in the config is correct and absolute

• Verify Python is in your PATH

Tools not appearing:

• Restart Gemini CLI after updating the configuration

• Check the MCP server logs for errors

• Verify the configuration JSON syntax is correct