AI agent

The AI agent feature for the auctioning system

Quick Start

1. Setup Environment Variables

cp .env.example .env

Edit .env with your configuration (LLM credentials, etc.)

2. Running with CLI

Install dependencies:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt

Run the API agent:

python -m app.agents.api_agent.api_agent

Run the SQL agent:

python -m app.agents.sql_agent.sql_agent

Run the Supervisor Agent V2 (Sequential Flow):

python -m app.agents.supervisor_agent.supervisor_agent_v2

The Supervisor Agent V2 implements a sequential flow where:

• API agent is tried first for all requests
• SQL agent is used as fallback when API agent cannot fulfill the request

3. Running with FastAPI

Start the service:

uvicorn app.main:app --host 0.0.0.0 --port 8000 --reload

Service will start on http://localhost:8000 API docs available at http://localhost:8000/docs

See docs/SUPERVISOR_AGENT_V2.md for details.

----- UNDER CONSTRUCTION -----

Run the service:

uvicorn app/main:app --host 0.0.0.0 --port 8080

Service will start on localhost:8080

3. Running with Docker

docker compose up

This starts AI agent service in a Docker.

Features

Multi-Agent System

• API Agent: Handles direct API operations (create items, place bids, get auction info)
• SQL Agent: Executes complex database queries for analytics and reporting
• Supervisor Agent V2: Sequential flow coordinator (API first, SQL as fallback)

Key Capabilities

• Create and manage auction items
• Place and track bids
• Query auction statistics and analytics
• Natural language to SQL conversion
• Automatic fallback for complex queries

Testing

Run unit tests:

PYTHONPATH=/home/runner/work/ai-agent/ai-agent python tests/test_supervisor_v2_unit.py

Tech Stack