🌦 Weather & Resource MCP Agent

Python-based AI agent that uses the Model Context Protocol (MCP) to connect Google Gemini with external tools.

• Fetches live weather data from OpenWeatherMap
• Loads and processes a local delivery log resource
• Built with LangGraph for orchestration and FastMCP for server tooling

---

📌 Features

• Weather Tool: Real-time weather for any city via OpenWeatherMap
• Resource Tool: Load/query a local delivery_log.txt file
• Interactive Chat: Natural conversation; the agent auto-chooses tools
• MCP Protocol: Clean separation between AI logic and external tools

---

🛠 Tech Stack

• Python 3.10+
• MCP — standard for AI tool integration
• FastMCP — quick MCP server creation
• LangGraph — agent state management
• LangChain Google GenAI — Gemini LLM
• OpenWeatherMap API — weather provider
• python-dotenv — env management

---

📂 Project Structure

.
├── weather_client.py       # AI agent (LangGraph + Gemini + MCP client)
├── weather_server.py       # MCP server (weather + resource tools)
├── delivery_log.txt        # Example resource file
├── requirements.txt        # Python dependencies
├── example.env             # Sample environment configuration
├── .gitignore              # Git ignore rules
├── LICENSE                 # Project license (MIT)
└── README.md               # Project documentation

---

🔑 Prerequisites

1. Python 3.10+ installed
2. API keys
   • OpenWeatherMap
   • Google Gemini

---

⚙️ Setup

1. Clone the repository

git clone https://github.com/ArslanJajja1/mcp-weather-resource-agent.git
cd single-server-mcp

2. Create and activate a virtual environment

python -m venv venv

• Windows (PowerShell):

venv\Scripts\Activate.ps1

• Windows (CMD):

venv\Scripts\activate.bat

• macOS/Linux:

source venv/bin/activate

3. Install dependencies

pip install -r requirements.txt

4. Configure environment variables

Copy example.env to .env and fill in your keys:

OPENWEATHERMAP_API_KEY=your_openweathermap_api_key_here
GOOGLE_GEMINI_API_KEY=your_google_gemini_api_key_here

---

🚀 Run

Run the client (it auto-starts the MCP server over stdio):

python weather_client.py

---

💬 Usage

• Ask for weather

  • You: What's the weather in London?
  • AI: The current weather in London is scattered clouds, 18°C, feels like 16°C, humidity 72%, wind speed 3.5 m/s.

• List available resources

  • You: /resources
  • Lists file://delivery_log with description

• Load a resource

  • You: /resource file://delivery_log
  • Loads the file and optionally performs your requested action

---

📜 Commands

• /resources: List all available resources from the MCP server
• /resource : Load a specific resource into the conversation
• exit / quit / q: Exit the program

---

🧩 How It Works

1. weather_client.py launches weather_server.py via MCP (stdio transport)
2. The client uses LangGraph to manage conversation flow and memory
3. Gemini (gemini-2.0-flash) decides whether to answer or call a tool
4. The MCP server executes tools and returns results
5. The client formats and returns a final answer

---

🧪 Tools Exposed by the MCP Server

• get_weather(location: str) — Returns current weather for a location using OpenWeatherMap
• Resource file://delivery_log — Returns the contents of delivery_log.txt

---

🔒 Security & Secrets

• Do not commit your .env file; .gitignore excludes it
• Keys are read via python-dotenv at runtime

---

🧰 Development

• Add more MCP tools in weather_server.py using @mcp.tool() or @mcp.resource()
• Keep dependencies in requirements.txt

---

📄 License

This project is licensed under the MIT License. See LICENSE for details.