# 🌦️ Weather MCP RAG Assistant

An AI-powered weather assistant built using **Model Context Protocol (MCP), Retrieval-Augmented Generation (RAG), LangChain, ChromaDB, Groq LLM, and OpenWeather API**.

The project exposes a custom MCP tool that retrieves real-time weather data and combines it with a RAG pipeline to generate contextual, natural-language weather insights.

## 🚀 Key Features

- Custom **MCP server** built using FastMCP
- Real-time weather retrieval using **OpenWeather API**
- **Retrieval-Augmented Generation (RAG)** pipeline for contextual responses
- **ChromaDB** vector store for storing and retrieving weather context
- Sentence Transformer embeddings using `all-MiniLM-L6-v2`
- **Groq-hosted LLM** integration through LangChain
- MCP tool integration with **VS Code / GitHub Copilot**
- Supports dynamic weather queries for different cities
- Secure API-key management using environment variables

## 🧠 How It Works

```text
User Query
    ↓
GitHub Copilot / MCP Client
    ↓
Weather MCP Server
    ↓
get_weather(location)
    ↓
OpenWeather API
    ↓
Weather Data Processing
    ↓
ChromaDB Vector Store
    ↓
RAG Retrieval
    ↓
Groq LLM
    ↓
Context-Aware Weather Response
```

## 🛠️ Tech Stack

**Language:** Python

**AI / GenAI:** LLMs, RAG, Prompt Engineering, Sentence Transformers

**Frameworks:** LangChain, FastMCP

**Vector Database:** ChromaDB

**LLM Provider:** Groq

**External API:** OpenWeather API

**Protocol:** Model Context Protocol (MCP)

**Development Environment:** VS Code, Git, GitHub

## 📂 Project Structure

```text
Weather-MCP-RAG-Assistant/
│
├── rag/
│   ├── embedding.py
│   ├── llm.py
│   ├── retriever.py
│   └── vector_store.py
│
├── services/
│   └── weather_service.py
│
├── utils/
│   └── parser.py
│
├── config.py
├── server.py
├── requirements.txt
├── env.example
├── .gitignore
└── README.md
```

## ⚙️ Setup

### 1. Clone the repository

```bash
git clone https://github.com/rudraniai/Weather-MCP-RAG-Assistant.git
cd Weather-MCP-RAG-Assistant
```

### 2. Create a virtual environment

```bash
python -m venv .venv
```

Activate it on Windows:

```bash
.venv\Scripts\activate
```

### 3. Install dependencies

```bash
pip install -r requirements.txt
```

### 4. Configure environment variables

Copy the example environment file:

```bash
copy env.example .env
```

Add your credentials to `.env`:

```env
OPENWEATHER_API_KEY=your_openweather_api_key
GROQ_API_KEY=your_groq_api_key
GROQ_MODEL=your_groq_model
EMBEDDING_MODEL=sentence-transformers/all-MiniLM-L6-v2
CHROMA_DB_DIR=./chroma_db
```

> API credentials are kept outside version control using `.gitignore`.

### 5. Run the MCP server

```bash
python server.py
```

## 🔌 MCP Configuration

Configure the local MCP server in VS Code using:

```json
{
  "servers": {
    "weather-mcp": {
      "type": "stdio",
      "command": "python",
      "args": [
        "PATH_TO_PROJECT/server.py"
      ]
    }
  },
  "inputs": []
}
```

Replace `PATH_TO_PROJECT` with the local path to `server.py`.

After configuration:

1. Open the VS Code Command Palette.
2. Select `MCP: List Servers`.
3. Select `weather-mcp`.
4. Start the MCP server.
5. Open GitHub Copilot Chat and allow the MCP tool when requested.

## 🎥 Project Demo

The following demo shows the custom `weather-mcp` MCP tool being invoked through GitHub Copilot to retrieve and generate contextual weather information.

![Weather MCP Demo](assets/weather-mcp-demo.png)

## 💬 Example Usage

Ask Copilot:

```text
Use the weather-mcp MCP tool to get the weather in Pune.
```

Other examples:

```text
Use the weather-mcp MCP tool to get the weather in Mumbai.

Use the weather-mcp MCP tool to get the weather in Nagpur.
```

Example information returned by the assistant includes:

- Current temperature
- Feels-like temperature
- Humidity
- Wind speed and direction
- Cloud conditions
- Weather summary
- Air-quality information when available

## 🔐 Security

Sensitive credentials such as API keys are stored in a local `.env` file.

The following files/directories are excluded from Git:

```text
.env
chroma_db/
__pycache__/
.vscode/
```

Never commit real API keys to the repository.

## 📌 What I Learned

Through this project, I gained hands-on experience with:

- Building and exposing tools through **Model Context Protocol (MCP)**
- Integrating external APIs with LLM-based applications
- Implementing a **RAG pipeline**
- Working with embeddings and vector databases
- Using **ChromaDB** for contextual retrieval
- Connecting LangChain with a hosted LLM
- Integrating a custom MCP server with GitHub Copilot
- Managing environment variables and API credentials securely

## 🔮 Future Improvements

- Add multi-day weather forecasting
- Add weather alerts and recommendations
- Improve retrieval and contextual memory
- Add additional weather and environmental data sources
- Build a standalone web interface using Streamlit or FastAPI
- Containerize the application using Docker

## 👩‍💻 Author

**Rudrani Gulhane**

B.Tech Computer Science Engineering — Artificial Intelligence & Machine Learning

Interested in **AI/ML, Generative AI, RAG, LLMs, AI Agents, and MCP-based applications**.

## 📄 License

This project is intended for educational, portfolio, and experimental use.