This repository contains the ADK (Agent Development Kit) Developer Skill, an AI agent skill that provides up-to-date guidance, context, and best practices for developing applications and agents using the Google Agent Development Kit.
The primary purpose of this repository is to equip AI agents with the knowledge and context needed to write precise, up-to-date, and well-structured code when using the Google Agent Development Kit (ADK). By utilizing this skill, an agent can automatically refer to the latest ADK documentation and base its implementations on provided working examples.
The core skill and its assets are located under the .agent/skills/adk-skill/ directory:
SKILL.md: The main instructional file for the agent. It dictates how the agent should initialize the skill, what to read, and how to approach ADK development tasks.scripts/: Contains utility scripts.update_docs.sh: A shell script designed to pull the latest ADK documentation (llms-full.txt).
resources/: Where the contextual documentation is stored.llms-full.txt: The primary reference manual for ADK syntax, class methods, and usage patterns.
examples/: Contains reference implementations.basic_agent.py: A minimal, working example demonstrating how to set up an ADKAgentand execute it (e.g., usingInMemoryRunner).
As a developer, simply ask your AI agent to create or modify an ADK application within this repository. For example: "Write an ADK agent that returns the weather forecast."
When the agent receives your prompt, it will leverage this skill to automatically:
- Initialization: Execute
update_docs.shto fetch the latest, up-to-date ADK documentation. - Contextual Reading: Parse the newly updated
llms-full.txtto align with the most accurate and recent syntax and usage patterns. - Reference Examples: Inspect
basic_agent.pyto understand the standard boilerplate and structure of ADK applications. - Development: Apply this gathered knowledge to accurately implement or fix the requested ADK agent for you.