convention-adk-agent

Standardize Google ADK agent structure with root_agent export and PromptLoader.

Updated Mar 19, 2026
One-click install
npx skills add https://github.com/sunLeee/optimization --skill convention-adk-agent
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: convention-adk-agent
Source: https://github.com/sunLeee/optimization/tree/main/.claude/skills/reference/adk/agent
Command: npx skills add https://github.com/sunLeee/optimization --skill convention-adk-agent

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Google ADK 기반 에이전트 구현에 필요한 패턴을 제시하고, root_agent export의 일관성, PromptLoader 사용의 표준화, 그리고 ToolContext 상태 스키마의 설계를 우선합니다.

Core Features & Use Cases

  • 루트 에이전트 export 패턴: init.py에서 root_agent를 올바르게 export하도록 구조를 확립합니다.
  • PromptLoader를 이용한 프롬프트 외부화: prompts를 외부 템플릿으로 관리하고 시스템 프롬프트 로드를 표준화합니다.
  • 상태 스키마 우선 설계: initial_state 키와 ToolContext.state 구조를 구현 전 문서화하고 검증합니다.
  • 테스트 및 가이드 작성: 에이전트 구조 가이드와 테스트 파일 뼈대를 제공합니다.

Quick Start

새로운 에이전트를 agents/{name} 아래에 생성하고, root_agent export를 구현하며 prompts/를 PromptLoader로 로드하도록 구성합니다.

Frequently Asked Questions about convention-adk-agent

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I structure a Google ADK agent with root_agent export?

To structure a Google ADK agent, create a folder under agents/{agent_name} and implement the root_agent export in __init__.py to establish a consistent modular project layout.

What is the best way to externalize prompts in Google ADK?

The best way to externalize prompts in Google ADK is using PromptLoader to manage prompts as external templates, standardizing system prompt loading across modular agents.

How does ToolContext state schema work in ADK agents?

ToolContext state schema in ADK agents defines initial_state keys and state structure, documenting them before implementation to support reliable development and testing.

Can I standardize multiple modular ADK agents in one project?

Yes, you can standardize multiple modular ADK agents by enforcing a consistent folder structure under agents/{agent_name} with defined root_agent exports and PromptLoader usage.

Do I need to document initial_state before building an ADK agent?

Yes, documenting a well-documented initial_state for ToolContext.state is required before implementation to validate the state schema and ensure reliable agent testing.

Why does my ADK agent structure lack consistency across modules?

ADK agent structures lack consistency when missing a standardized root_agent export in __init__.py and PromptLoader usage, which pattern-driven implementation enforces.