Skill Detail

Build and evaluate production agents with Google ADK

Use Google Agent Development Kit to define, test, evaluate, and deploy code-first Python agents with tools, sessions, HITL confirmation, and multi-agent orchestration.

Developer ToolsMulti-Framework
Developer Tools Multi-Framework Security Reviewed
โญ 19.5k GitHub stars
INSTALL WITH ANY AGENT
npx skills add agentskillexchange/skills --skill build-and-evaluate-production-agents-with-google-adk Copy
Works best when you want a reusable capability, not another fragile one-off prompt.
At a glance
Tools required
Python 3 environment; google-adk package; model credentials; optional Google Cloud or Vertex AI deployment account
Install & setup
Install with `pip install google-adk` or `pip install "google-adk[extensions]"`, configure model credentials, define agents/tools in Python, and use ADK evaluation and deployment workflows from the project documentation.
Author
Google
Publisher
Vendor Open Source
Last updated
May 8, 2026
Quick brief

This skill helps teams build production-oriented Python agents with Google ADK. The operator defines agent logic, tools, sessions, orchestration, and evaluation flows in code, then deploys locally, to Cloud Run, or to Vertex AI Agent Engine when the workflow is ready.

How it works

What this skill actually does

Invoke it when an agent workflow needs repeatable engineering controls: versioned Python definitions, tool confirmation, evaluations, session handling, and deployment packaging. The scope boundary is building and operating code-first ADK agents; it is not a generic Google Cloud product card or a general Python SDK listing.

Inputs and prerequisites: Python 3 environment; google-adk package; model credentials; optional Google Cloud or Vertex AI deployment account.

Setup notes: Install with `pip install google-adk` or `pip install “google-adk[extensions]”`, configure model credentials, define agents/tools in Python, and use ADK evaluation and deployment workflows from the project documentation.

Source and verification boundary: use https://google.github.io/adk-docs/ as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.

Framework fit: publish this as a Multi-Framework workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.