Multi-Agent Interaction using OpenAI Agent SDK Example

This project demonstrates how to build and orchestrate multiple intelligent agents using the OpenAI Agent SDK. The system supports:

• Tool-augmented agents – Each agent can invoke domain-specific tools (e.g. SQL queries, metadata inspection).
• Agent memory – Agents maintain conversational context across multiple interactions.
• Agent handoffs – Tasks can be delegated between agents dynamically based on user intent.

The goal is to showcase how to design composable, intelligent agent systems capable of working collaboratively and contextually.

---

Installation

This project uses uv, a modern Python package manager designed for speed and reliability.

Step 1: Install uv

If uv is not already installed, you can install it by running (mac and linux): explore uv docs here

curl -LsSf https://astral.sh/uv/install.sh | sh

Alternatively, if you use Homebrew:

brew install uv

Step 2: Sync the environment

Once uv is installed, synchronise the project environment using:

uv sync

Environment Setup

To run the project, create a .env file in the project root directory and add your OpenAI API key:

OPENAI_API_KEY=your-openai-api-key-here

Running the Project

To start the main application, use the following command:

uv run python src/main.py

Project Overview

• Built using the OpenAI Agent SDK
• Modular and extensible agent definitions
• Enables stateful interaction and inter-agent communication
• Designed for experimentation with agent-based systems