Who it is forA developer prototyping an assistant who wants to see tool calls and responses before writing an app around it.
The problemTesting an agent from print statements makes it hard to see what it did and why.
The outcomeA working root_agent you can run from the terminal or debug in the ADK web UI.
What you need first
- Python 3.11 or later
- pip
- A model to call, such as Gemini; ADK is model-agnostic
- For Cloud Run deploys: a Google Cloud project; for local containers: Docker
Steps
- Install with
pip install google-adk(Python 3.11+) - Create an Agent with a name, model and instruction, as in the Quick Start, and assign it to root_agent
- Chat with it in the terminal with
adk run path/to/my_agent - Open the development UI with
adk web path/to/agents_dirto test, debug and showcase it - Add optional integrations with
pip install "google-adk[extensions]"when you need them
Prompt for your agent
Using google/adk-python, create an agent folder with a root_agent that <describe job>, give it <tools>, and run it with `adk web path/to/agents_dir`. Show me a test conversation and the tool calls it made.
Caveat. Releases land roughly every two weeks and this is ADK 2.0, so older tutorials may not match; the README also offers constraints files for safer installs.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for google/adk-python and "Build an agent and test it in ADK's web UI"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer