Agent Development Kit

Agent Development Kit¶

智能体开发工具包¶

Agent Development Kit (ADK) is a flexible and modular framework for developing and deploying AI agents. While optimized for Gemini and the Google ecosystem, ADK is model-agnostic, deployment-agnostic, and is built for compatibility with other frameworks. ADK was designed to make agent development feel more like software development, to make it easier for developers to create, deploy, and orchestrate agentic architectures that range from simple tasks to complex workflows.

智能体开发工具包 (ADK) 是一个灵活且模块化的框架,用于开发和部署 AI 智能体。虽然针对 Gemini 和 Google 生态系统进行了优化,但 ADK 是模型无关的、部署无关的,并且是为与其他框架兼容而构建的。ADK 的设计使智能体开发感觉更像是软件开发,使开发者更容易创建、部署和编排从简单任务到复杂工作流的智能体架构。

News: ADK TypeScript v0.2.0 released!

新闻: ADK TypeScript v0.2.0 已发布!

ADK TypeScript v0.2.0 is officially released! By popular demand, the ADK team has brought the power of Agent Development Kit to one of the most popular programming languages on the planet. For details, check out the blog post.

ADK TypeScript v0.2.0 正式发布!根据大众需求,ADK 团队已将智能体开发工具包的功能带到了全球最受欢迎的编程语言之一。有关详情,请查看博客文章。

News: ADK Go v0.3.0 released!

新闻: ADK Go v0.3.0 已发布!

ADK Go release v0.3.0 includes numerous bug fixes, introduces new features such as agent-to-agent request callbacks and extendability, and updates dependencies like the GenAI SDK and the ADK Web UI. For release details, check out the release notes.

ADK Go v0.3.0 版本包括众多错误修复,引入了诸如智能体到智能体的请求回调和可扩展性等新功能,并更新了 GenAI SDK 和 ADK Web UI 等依赖项。有关发布详情,请查看发布说明。

News: ADK Java v0.5.0 released!

新闻: ADK Java v0.5.0 已发布!

The ADK Java v0.5.0 release adds new features for tool execution mode configuration and model versioning, along with numerous bug fixes, dependency updates, and significant refactoring to improve the agent and runner architecture. For release details, check out the release notes.

ADK Java v0.5.0 版本增加了工具执行模式配置和模型版本控制的新功能,以及众多错误修复、依赖项更新和重大重构以改进智能体和运行器架构。有关发布详情,请查看发布说明。

Get started:

开始使用:

PythonTypeScriptGoJava

pip install google-adk

npm install @google/adk

go get google.golang.org/adk

pom.xml

<dependency>
    <groupId>com.google.adk</groupId>
    <artifactId>google-adk</artifactId>
    <version>0.5.0</version>
</dependency>

build.gradle

dependencies {
    implementation 'com.google.adk:google-adk:0.5.0'
}

Start with Python Start with TypeScript Start with Go Start with Java

Learn more¶

Flexible Orchestration # 灵活的编排

Define workflows using workflow agents (Sequential, Parallel, Loop) for predictable pipelines, or leverage LLM-driven dynamic routing (LlmAgent transfer) for adaptive behavior.

使用工作流智能体 (Sequential、Parallel、Loop) 定义工作流以实现可预测的流水线,或利用 LLM 驱动的动态路由 (LlmAgent 转移) 以实现自适应行为。

Learn about agents
Multi-Agent Architecture # 多智能体架构

Build modular and scalable applications by composing multiple specialized agents in a hierarchy. Enable complex coordination and delegation.

通过组合层次结构中的多个专业化智能体来构建模块化和可扩展的应用程序。启用复杂的协调和委托。

Explore multi-agent systems
Rich Tool Ecosystem # 丰富的工具生态系统

Equip agents with diverse capabilities: use pre-built tools (Search, Code Exec), create custom functions, integrate 3rd-party libraries, or even use other agents as tools.

为智能体配备多样化功能:使用预构建工具(搜索、代码执行)、创建自定义函数、集成第三方库,甚至将其他智能体作为工具使用。

Browse tools
Deployment Ready # 部署就绪

Containerize and deploy your agents anywhere – run locally, scale with Vertex AI Agent Engine, or integrate into custom infrastructure using Cloud Run or Docker.

在任何地方容器化并部署您的智能体——本地运行,使用 Vertex AI 智能体引擎扩展,或使用 Cloud Run 或 Docker 集成到自定义基础设施中。

Deploy agents
Built-in Evaluation # 内置评估

Systematically assess agent performance by evaluating both the final response quality and the step-by-step execution trajectory against predefined test cases.

通过根据预定义的测试用例系统地评估最终响应质量和逐步执行轨迹来评估智能体性能。

Evaluate agents
Building Safe and Secure Agents # 构建安全可靠的智能体

Learn how to building powerful and trustworthy agents by implementing security and safety patterns and best practices into your agent's design.

了解如何通过在智能体设计中实施安全和安全模式以及最佳实践来构建强大且值得信赖的智能体。

Safety and Security

Agent Development Kit

Agent Development Kit¶

智能体开发工具包¶

Learn more¶

了解更多信息¶