Used for environment variable management, specifically for storing and accessing API keys like GROQ_API_KEY securely within the application.
Provides a unified agent interface for tool invocation, enabling orchestration of multiple MCP servers (Math and Weather agents) through LangChain's framework.
Serves as the implementation language for custom MCP servers, providing both stdio and HTTP transports for agent communication.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Model Context Protocol Multi-Agent ServerWhat's 15 multiplied by 60 and the weather in Delhi?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Model Context Protocol (MCP) Multi-Agent Demo
This project demonstrates how to set up and communicate with custom Model Context Protocol (MCP) servers in Python. It showcases multi-agent orchestration using LangChain, Groq, and MCP adapters, enabling both local and remote tool integration.
Features
Custom MCP Servers: Math and Weather agents, each as independent MCP servers
Multi-Transport Communication: Local (stdio) and remote (HTTP) transports
LangChain Integration: Unified agent interface for tool invocation
Async Orchestration: Efficient, non-blocking agent communication
Related MCP server: MCP Server
Components
1. mathserver.py
A custom MCP server providing math operations (add, multiply) via stdio transport.
2. weather.py
A custom MCP server providing weather information via HTTP transport - (Static content for demo).
3. client.py
A Python client that connects to both servers, discovers their tools, and invokes them using a LangChain agent powered by Groq.
Setup Instructions
Clone the repository
Install dependencies:
Set up environment variables:
Create a
.envfile with yourGROQ_API_KEY:GROQ_API_KEY=your_groq_api_key_here
Run the servers:
Start the weather server (in one terminal):
python weather.pyThe math server is started automatically by the client when needed.
Run the client:
Example Output
Learning Outcomes
How to build and register custom MCP servers
How to enable communication between agents using stdio and HTTP
How to orchestrate multi-agent workflows with LangChain
Requirements
Python 3.8+
See
requirements.txtfor Python dependencies
License
MIT