MCP-AWS

🚀 Multi-Agent System: A2A & MCP Integration POC

POC: Integrating A2A, MCP, and OpenAI Agents for AWS Tasks 🖥️✨

🎥 Demo Video

Watch the demo video to see MCP-AWS in action! 🚀

🌟 Features

1. 🚀 Seamless Protocol Integration: Demonstrates the successful integration of the Agent-to-Agent (A2A) protocol with a Model Context Protocol (MCP) server for robust multi-agent communication.
2. 🧠 Leverages OpenAI Agents SDK: Built upon the powerful OpenAI Agents SDK to create intelligent agents capable of understanding and acting on user prompts.
3. ☁️ Automated Cloud Management: Enables direct provisioning and termination of AWS EC2 instances through simple user interactions, showcasing practical tool execution via the MCP.

🛠️ Tools in the MCP Server

The MCP server is a custom server with two tools:

1. initiate_aws_ec2_instance: Creates an AWS EC2 instance.
2. terminate_aws_ec2_instance: Terminates an AWS EC2 instance by its ID.

🚀 Getting Started

Prerequisites

1. Python 3.12+ (for local setup) or Docker (for containerized setup)
2. AWS IAM Role: Create an IAM role with the necessary permissions to manage EC2 instances.
3. Environment Variables: Prepare a .env file with the following variables:
   • AWS_ACCESS_KEY_ID
   • AWS_SECRET_ACCESS_KEY
   • AWS_DEFAULT_REGION
   • OPENAI_API_KEY
   • AMI_ID
   • INSTANCE_TYPE
   • KEY_NAME
   • SECURITY_GROUP_IDS
   • AWS_REGION

🏃‍♂️ Running the App

1. Clone the repository at the root:
   git clone https://github.com/anirban1592/google_openai_mcp.git cd google_openai_mcp
2. Create .env file as shown in prerequisites
3. Run the remote agent example:
   cd openai-agent/ uv run .
4. Clone the A2A client code(by google) at the root dir:
   git clone https://github.com/google/A2A.git cd demo/ui
5. Create an environment file with your API key or enter it directly in the UI when prompted:
   echo "GOOGLE_API_KEY=your_api_key_here" >> .env
6. Run the front end example:
   uv run main.py
7. Refer to the attached video to see it in action

💬 Using the AI Agent

1. To create an EC2 instance:
   Enter your command: Create an EC2 instance
2. To terminate an EC2 instance:
   Enter your command: Terminate EC2 instance with ID <instance-id>

⚠️ Word of Caution

• IAM Role and Credentials: Please create AWS IAM roles and credentials at your own risk. Ensure you follow AWS best practices for security.
• Billing and Security: This app is a proof of concept (POC) and is intended for learning purposes only. We are not responsible for any billing issues or security incidents.

📚 Learnings

This project demonstrates:

1. How to integrate MCP servers with OpenAI Agents SDK
2. How to build a simple AI-driven application for AWS resource management

Enjoy exploring the power of AI and MCP servers! 🌟