Agent Development Kit (ADK) Documentation
by Google
Google's open-source, code-first framework for building and deploying production AI agents.
Overview
The Agent Development Kit (ADK) is Google's open-source framework for building, debugging, and deploying AI agents at enterprise scale, with SDKs for Python, TypeScript, Go, Java, and Kotlin. The documentation covers the full agent lifecycle: defining agents with instructions and tool calls, composing multi-agent architectures with flexible orchestration (predictable workflow pipelines or dynamic LLM-driven routing), running a built-in evaluation harness, and deploying to Google Cloud (Agent Runtime, Cloud Run, or GKE) or your own infrastructure. ADK is model-agnostic — it works with Gemini and, via adapters, other leading model providers — and integrates with the Model Context Protocol and Google Cloud's built-in auth, tracing, and observability. Guides, quickstarts, and hands-on codelabs get a first agent running in minutes.
At a Glance
- Topic
- Agentic
- Level
- Intermediate
- Format
- Documentation
- Cost
- Free
- Duration
- Docs + quickstarts + codelabs, self-paced
- Provider
- Hands-on
- Yes — code/exercises
- Certificate
- None
What You’ll Learn
- ✓Build a code-first AI agent with instructions, tools, and function calls in Python or another supported language
- ✓Compose multi-agent systems using workflow (sequential/parallel/loop) agents or LLM-driven dynamic routing
- ✓Evaluate agent behavior with ADK's built-in evaluation tooling
- ✓Deploy agents to Google Cloud (Agent Runtime, Cloud Run, GKE) or self-host
- ✓Integrate external tools and MCP servers, and wire in observability and tracing
Highlights
- •Official Google framework, launched at Cloud Next 2025
- •Model-agnostic and open-source (Apache-2.0) with multi-language SDKs
- •First-class multi-agent orchestration and evaluation built in
- •One-command deploy path to managed Google Cloud runtimes
Who It’s For
Best For
- ✓AI engineers building multi-agent or tool-using agent systems
- ✓Teams that want a production deployment path on Google Cloud
- ✓Developers evaluating open-source agent frameworks (vs LangGraph, CrewAI, OpenAI Agents SDK)
Prerequisites
- •Comfortable programming in Python (or TS/Go/Java)
- •Basic understanding of LLMs, prompts, and tool/function calling
FAQ
What is Agent Development Kit (ADK) Documentation?
Official documentation for Google's Agent Development Kit (ADK), an open-source, code-first framework for building, evaluating, and deploying reliable AI agents — for engineers who want model-agnostic multi-agent systems that run from a laptop to Google Cloud. Introduced at Google Cloud Next 2025.
Is Agent Development Kit (ADK) Documentation free?
Agent Development Kit (ADK) Documentation is free to access.
What level is Agent Development Kit (ADK) Documentation for?
Agent Development Kit (ADK) Documentation is aimed at a intermediate audience. Recommended background: Comfortable programming in Python (or TS/Go/Java), Basic understanding of LLMs, prompts, and tool/function calling.
How long does Agent Development Kit (ADK) Documentation take?
Expect roughly Docs + quickstarts + codelabs, self-paced. Most learners work through it at their own pace.
What will I learn from Agent Development Kit (ADK) Documentation?
You'll learn: Build a code-first AI agent with instructions, tools, and function calls in Python or another supported language; Compose multi-agent systems using workflow (sequential/parallel/loop) agents or LLM-driven dynamic routing; Evaluate agent behavior with ADK's built-in evaluation tooling; Deploy agents to Google Cloud (Agent Runtime, Cloud Run, GKE) or self-host; Integrate external tools and MCP servers, and wire in observability and tracing.