Test Agent System

This project is a proof-of-concept designed to test multi-agent delegation using the Google Agent Development Kit (ADK). It features a root agent that delegates tasks to specialized sub-agents based on user input.

⚠️ Disclaimer: Not for Production Use

This project is strictly for testing and demonstration purposes.

• It is not intended for production environments.
• The code is designed to verify that sub-agents are correctly called and that the delegation logic works as expected.
• Security, error handling, and scalability features required for a production system may be minimal or missing.

Architecture

The system consists of a hierarchical agent structure:

1. Root Agent: The main entry point. It understands user intent and routes requests to the appropriate sub-agent.
2. BigQuery Sub-Agent: Handles requests related to BigQuery. It is currently configured to query a public dataset for popular names.
3. API Sub-Agent: Handles requests related to external APIs. It is currently configured to fetch data from a placeholder test API.

Capabilities

• BigQuery: Can query the bigquery-public-data.usa_names.usa_1910_2013 dataset to retrieve the top 10 most popular names.
• API: Can make a test GET request to https://jsonplaceholder.typicode.com/todos/1.

Setup

1. Initialize Virtual Environment:

   python3 -m venv .venv
   source .venv/bin/activate

2. Install Dependencies:

   pip install -r requirements.txt

3. Run the Agent: From the parent folder, run:

   adk web