A first agent project involves more than writing an instruction. Build Your First Agent with Agent Development Kit (ADK) introduces the environment, configuration and execution choices that connect an agent idea with working code.
This free introductory Google course moves through setup, agent identity, practical changes and alternative ways to define and run an agent. It is useful when you want a clearer starting workflow for ADK, including the role of Python, YAML configuration and the interfaces used to examine an agent’s behavior.
Course at a glance
| Provider | |
|---|---|
| Platform | Google Skills |
| Level | Beginner |
| Language | English |
| Estimated study time | 1 hour 15 minutes; official estimate, individual study time varies. |
| Format | Seven web lessons and four quizzes, with setup and configuration practice |
| Access | Free instruction; free Google Skills account required. Product and practical access are separate. |
| Recognition | Course completion badge advertised after required activities; no professional certification or academic credit. |
What you’ll learn
- Explore installation with a virtual environment and the ADK setup workflow.
- Define an agent through its model, name, description and instruction.
- Examine command-line, web and API-server execution, plus programmatic use of Runner and sessions.
- Compare Python-based agents with YAML Agent Config and their appropriate use.
Skills you’ll gain
- ADK development setup
- Agent identity configuration
- Python–YAML comparison
- Execution-mode awareness
- Agent workflow literacy
Understand the pieces of the first development setup
The course begins with step-by-step setup before examining an agent’s identity and a hands-on transformation activity. It then covers execution methods and the two approaches to defining agents. Four quizzes support that progression.
The stated setup objectives include virtual environments, the root_agent naming convention and environment configuration after adk create. These details help you understand the structure of the examples rather than treating installation as an unexplained prerequisite.
The model, name, description and instruction parameters give the identity discussion a concrete form. Track what each setting controls as you study. A well-named agent and a clear instruction are useful development choices, but they do not by themselves establish that every output will meet the intended requirements.
Compare how you define and run the agent
The learning objectives identify adk web, adk run and adk api_server as execution modes, alongside programmatic execution with Runner and sessions. Keep that distinction in your notes: a development interface and a production deployment decision are different parts of the workflow.
Python and YAML definitions are also compared. Rather than assuming one approach is always better, use the examples to understand the choices they support. Follow the course in the context of its tooling version; independently reproducing a setup may involve current package and service requirements.
For optional study preparation, outline a fictional assistant that summarizes supplied public information. Write a name, a short description and the behavior you would want to test. This is independent reflection, not an additional provider assessment, and it does not require a live model call.
Google marks the course Introductory, mapped to Beginner. Programming familiarity is useful for the implementation material without becoming a new formal entry requirement. Allow additional time beyond the seventy-five-minute estimate if you pause to investigate configuration or reproduce the practical examples.
Free learning and practical access
The complete course instruction is advertised as Free on Google Skills, with a free learner account used for access and progress. The public curriculum has seven web lessons and four required quizzes, with no separately listed provisioned cloud lab. Independently running the examples can require local software, model access or computing resources with separate conditions and possible costs; no unlimited free runtime is promised. Google advertises a course completion badge after the required activities. It is not a professional certification or academic credit.
Explore more learning options in the free course catalogue.
Frequently asked questions
Does the course compare Python and YAML?
Yes. The objectives include Python agents and YAML Agent Config, including when each definition approach is useful.
Are the three execution modes the same as production hosting?
No. The course discusses web, command-line and API-server execution, alongside programmatic use. This article does not turn those interfaces into a promise of a hosted production service.
Is model usage included without limits?
No unlimited model or hosting entitlement is promised. The instruction is free; independently running an agent can require separate software, model and computing access.
Questions & discussion
Share a useful question or correction. Comments appear after moderation. Please avoid personal or sensitive information.