Multi-Agents Project

This repository demonstrates a true multi-agent orchestration system. Multiple agents — inventory, quoting, and ordering — work collaboratively in a pipeline, coordinated by the orchestrator.py module. Each agent handles a specific responsibility, and the orchestrator manages the flow of data and task sequencing.

Features

• True multi-agent orchestration: Agents are coordinated in a pipeline, not combined into a single monolithic agent.
• Modular agents: Inventory, quoting, and ordering agents operate independently.
• Pipeline workflow: Inventory → Quoting → Ordering.
• Extensible: Easily add new agents or tools.
• Example scenarios: Includes CSV-driven request simulations.

Architecture

The architecture follows a pipeline model:

The orchestrator (orchestrator.py) coordinates this flow, passing outputs from one agent to the next and handling errors at each stage.

Getting Started

1. Clone the repository:

   git clone https://github.com/your-username/Multi-agents-project.git
   cd Multi-agents-project

2. Install dependencies:

   poetry install

3. Initialize the database and run test scenarios:

   python main.py

The main.py script loads sample requests from CSV, runs them through the multi-agent pipeline, and outputs responses along with updated financial and inventory states.

Project Structure

.
├── agents.py                 # Inventory, quoting, and ordering agent definitions
├── orchestrator.py           # Multi-agent orchestrator coordinating the pipeline
├── entities.py               # Pydantic models for requests, quotes, and errors
├── main.py                   # Main script to initialize DB and run test scenarios
├── multi-agents-architecture-diagram.png
├── README.md
└── poetry.toml