Workshop ADK

Sample agents built with the Google Agent Development Kit (ADK) for a workshop on building LLM agents.

Prerequisites

• Python 3.13+
• uv for dependency management
• A Google API key with access to Gemini models

Setup

1. Install dependencies:

   uv sync

2. Add your API key to .env in the project root:

   GOOGLE_API_KEY=your-api-key-here

Running an agent

Each agent lives in its own folder with an agent.py exposing a root_agent. From the project root, run:

# Interactive terminal chat with a specific agent
uv run adk run weather_agent

# Browser-based UI (http://localhost:8000) with a dropdown to pick any agent
uv run adk web

Agents in this repo

• weather_agent/ — a single agent with a mock get_weather tool. Good starting example for tools + instructions.
• code_agent/ — a SequentialAgent pipeline (CodeWriterAgent → CodeReviewerAgent → CodeRefactorerAgent) that writes, reviews, and refactors Python code.

Standalone scripts

• weather_agent_starter.py — exercise skeleton for wiring up a Runner + InMemorySessionService around weather_agent by hand (fill in the TODOs).
• weather_agent_completed.py — the completed version of the same script.

Run either directly with:

uv run python weather_agent_completed.py