somnathsingh31/mcp-quickstart

🚀 MCP Quickstart: Custom Client & Server with LLM Integration

This project provides a quick and practical starting point to understand and experiment with the Model Context Protocol (MCP) by implementing both a custom MCP server and MCP client, along with integration to a Large Language Model (LLaMA 4) via Groq API.

💡 What This Project Does

• MCP Server (servers/)

  • Exposes 3 tools via MCP protocol:

    • get_weather: Get weather for a specified city.
    • get_gold_price: Get the current gold price.
    • get_bitcoin_price: Get the latest Bitcoin price.

• MCP Client (clients/)

  • Connects to the MCP server.
  • Retrieves the list of tools available.
  • Passes tools to a connected LLM (LLaMA 4 via Groq API).
  • Allows the LLM to invoke tools based on user prompts.

• LLM Interface

  • Communicates with the Groq LLM API.
  • Provides tool definitions as context so the LLM can call them when needed.

🔑 Groq API Key: Get your API key from https://console.groq.com/keys to enable LLM integration.

📁 Project Structure

mcp-quickstart/
├── clients/         # MCP client implementation
├── servers/         # MCP server with tool definitions
├── utils/           # Shared helpers
├── main.py          # Start point: runs client-server 
├── requirements.txt # Dependency list for pip users
├── pyproject.toml   # For uv-based dependency management
└── README.md

⚙️ How to Run

Option 1: Using pip

# Clone the repo
git clone https://github.com/somnathsingh31/mcp-quickstart.git
cd mcp-quickstart

# Create and activate virtual environment
python3 -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Run the project
python main.py

Option 2: Using uv

# Clone the repo
git clone https://github.com/somnathsingh31/mcp-quickstart.git
cd mcp-quickstart

# Install uv if not installed
curl -Ls https://astral.sh/uv/install.sh | bash

# Sync environment and install dependencies
uv sync

# Run the project
uv run main.py

This will start the MCP client, connect to the custom MCP server, fetch tool definitions, and pass them to the LLM for dynamic invocation based on user queries.

🧠 Key Concepts Demonstrated

• Custom implementation of the MCP protocol
• Dynamic tool selection and execution by LLM
• Tool call architecture using JSON-based schemas
• End-to-end integration with Groq's hosted LLaMA 4

📚 References

• MCP Protocol
• Groq

📝 License

Licensed under the .