open-mcp-agents

Exploring agent infrastructure with a focus on MCP-based orchestration with Fast MCP, Langchain and LangGraph.

---

Overview

This project demonstrates a modular agent infrastructure using MCP (Model-Context Protocol). It leverages Fast MCP, Langchain, and LangGraph to connect a client agent to multiple backend MCP microservices (e.g., todo, weather, math servers).

Architecture

Prerequisites

• Python 3.11+
• uv
• Docker & Docker Compose
• A valid GROQ_API_KEY

Setup

1. Clone the repository

git clone https://github.com/adilsaid64/open-mcp-agents
cd open-mcp-agents

2. Set up environment variables

Copy the template and fill in your API key:

cp app/.template.env app/.env
# Edit app/.env and set GROQ_API_KEY

3. Run with Docker Compose

This will build and start the client and todo-server. You can extend the compose file to add more services.

docker-compose up --build

Environment Variables

• GROQ_API_KEY: Your API key for Groq (required by the client)