adk-python

name	adk-python
description	(💛) Create and manage AI agents using Google's Agent Development Kit (ADK) for Python. Use when the user wants to build single or multi-agent systems, evaluate agents, or deploy them. Includes instructions for fetching fresh ADK documentation and links for TypeScript and Go development.
compatibility	Gemini CLI
metadata	{"version":0.1}

ADK Python Skill

This skill helps you build AI agents using the Agent Development Kit (ADK) for Python.

Core Workflows

1. Initialize ADK Knowledge

Always start by fetching the latest ADK documentation to ensure you have the most up-to-date patterns and API references.

• Option A (Summary): Fetch llms.txt for a high-level overview and index.
• Option B (Deep Dive): Fetch llms-full.txt for comprehensive documentation (large file).

Use the provided script to fetch these:

python3 scripts/fetch_context.py --type summary  # for llms.txt
python3 scripts/fetch_context.py --type full     # for llms-full.txt

2. Create an Agent

Basic pattern for a single agent:

from google.adk.agents import Agent
from google.adk.tools import google_search

agent = Agent(
    name="my_agent",
    model="gemini-2.0-flash",
    instruction="Your instructions here",
    tools=[google_search]
)

3. Multi-Agent Systems

Use LlmAgent and sub_agents to compose systems:

from google.adk.agents import LlmAgent

child_agent = LlmAgent(name="child", model="gemini-2.0-flash", ...)
parent_agent = LlmAgent(
    name="coordinator",
    model="gemini-2.0-flash",
    sub_agents=[child_agent]
)

Cross-Language Development

If the user prefers TypeScript or Go, point them to these resources:

• Go ADK: github.com/google/adk-go
• TypeScript/Web ADK: github.com/google/adk-web
• Java ADK: github.com/google/adk-java

Reference Material

See adk_links.md for a comprehensive list of ADK documentation and repositories.