Bootstrap the Google ADK environment

With credentials ready, install dependencies and understand how the repository is organised.

Create a virtual environment

Copy
cd trip-planner-agent-adk
python3 -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
python -m pip install --upgrade pip

Keeping dependencies inside .venv avoids conflicts with other Python projects.

Install runtime dependencies

Copy
pip install google-adk google-generativeai requests python-dotenv

• google-adk provides the Agent Development Kit CLI and runtime.
• google-generativeai exposes Gemini models for quick smoke tests.
• requests powers the travel assistant’s weather/news fetchers.
• python-dotenv loads .env files while developing locally.

Tip: save the resolved dependencies by running pip freeze > requirements.txt if you plan to share the project with teammates.

Explore the project structure

Copy
ls

You should see:

• personal_assistant/agent.py: the sequential travel orchestrator.
• personal_assistant/tools.py: helper functions for weather and news.
• celebration_planner/agent.py: the parallel celebration planner you will wire up later in the lab.
• README.md: quick-start commands.

Skim personal_assistant/agent.py to recognise how sub-agents are declared. Each agent is an instance of google.adk.agents.llm_agent.Agent with a model, instruction, and optional tools.

Launch the ADK Dev UI

Run the development server from the repository root so both agent packages are auto-discovered:

Copy
adk web .

Open the printed URL (defaults to http://127.0.0.1:8000). You should see a catalog listing Personal Assistant and Celebration Planner. Leave the server running; you will use it in the next step.

If you see authentication errors, confirm that GOOGLE_API_KEY is available in the shell (echo "$GOOGLE_API_KEY"). Restart the terminal after exporting new variables.