youtube link : Deep dive and step by step tutorial of A2A and MCP, including code walthrough
This is an LLM-powered application that allows users to execute terminal commands using simple English queries like:
"Delete
readme.txt
on my desktop."
It leverages the A2A (Agent-to-Agent) and MCP (Model Context Protocol) to interpret and execute commands on a Linux system.
Follow these steps to get the system up and running.
git clone https://github.com/ishanExtreme/a2a_mcp-example.git
cd a2a_mcp-example
Export your OpenAI API key to your environment:
export OPENAI_API_KEY="<your-openai-api-key>"
python -m venv venv or conda create ...
source venv/bin/activate # On Windows use: venv\Scripts\activate
pip install -r requirements.txt
Ensure all the services are running in separate terminal tabs/windows:
cd ./client/demo/ui
uv run main.py
cd ./a2a_server
python main.py
cd ./mcp_server
python mcp_server.py
- Open your browser and go to: http://localhost:12000
- Navigate to the Agents section.
- Add a Linux agent using the following endpoint:
http://localhost:10000
You are now ready to start chatting and executing Linux commands via natural language!
- "Delete the
notes.txt
file on Desktop" - "List all files in the Documents folder"
- "Create a new directory called
projects
"
- OpenAI GPT
- A2A Protocol
- MCP Protocol
- FastAPI + Uvicorn
- Python 3.12+
MIT License © 2025