Google AI releases ADK Go: a new open source toolkit designed to help Go developers build powerful AI agents

How to build a reliable AI agent that plugs into existing Go services without needing to connect a separate language stack? Google just released Go Agent Development Kit. Go developers can now build AI agents using the same framework that already supports Python and Java, while keeping everything within the familiar Go toolchain and deployment model.

This bridges the gap for AI developers and backend developers who are already serving services in Go. You no longer need a separate Python-based stack for your agent. You can express agent logic, orchestration, and tool usage directly in Go code, then move the same agents into Vertex AI Agent Builder and Agent Engine when they’re ready for production.

What does the Agent Development Kit offer??

The Agent Development Kit (ADK) is an open source framework for developing and deploying artificial intelligence agents. It is optimized for Gemini and Google Cloud, but is designed to be model and deployment agnostic.

In fact, ADK provides you with:

• A code-first programming model where agent behavior, tooling, and orchestration reside in common source files
• Workflow agents are used to broker sequential, parallel, and cyclic control flows within a system
• Rich tool ecosystem including built-in tools, custom function tools, OpenAPI tools, Google Cloud tools, and ecosystem tools
• Deployment paths covering local run, container, Cloud Run and Vertex AI Agent Engine
• Built-in assessment and safety mode, integrated with Vertex AI Agent Builder

For developers, ADK turns proxies into ordinary services. Instead of treating it as a one-time script that calls LLM, you can run it locally, inspect the trace, and deploy it to a managed runtime.

What ADKs have been added to Go?

The Go version retains the same core feature set as the Python and Java SDKs, but exposes it through the idiomatic Go API. The Google AI team describes ADK for Go as an idiomatic and performant way to build agents that use Go’s concurrency and strong typing.

Here are some key points:

• ADK for Go installed go get google.golang.org/adk
• The project is open source and hosted on github.com/google/adk-go
• It enables building, evaluating, and deploying complex artificial intelligence agents with flexibility and control
• It uses the same agents, tools, and workflow abstractions as other ADK languages

This means that Go services can embed proxy behavior without having to switch languages. You can build a multi-agent architecture where each agent is a Go component composed with other components using the same framework.

A2A protocol support in Go

ADK for Go comes with native support for the Agent2Agent protocol, or A2A.

The A2A protocol defines how agents can call other agents through a standard interface. In the Go version, Google emphasizes that the master agent can orchestrate tasks and delegate them to specialized subagents. These subagents can run locally or as a remote deployment. A2A keeps these interactions secure and opaque, so agents do not need to expose internal memory or proprietary logic to participate.

Google also contributed the A2A Go SDK to the main A2A project. This will provide a protocol-level entry point for Go developers if they want to interoperate with other runtimes and frameworks that also support A2A.

MCP toolbox for databases and tools

A key detail in Google’s official announcement is native integration with MCP Toolbox for Databases. It states that ADK Go provides out-of-the-box support for more than 30 databases through the toolbox.

MCP Toolbox for Databases is an open source MCP server for databases. It handles connection pooling, authentication, and other issues, and exposes database operations as tools using the model context protocol.

In ADK this means:

• You register MCP Toolbox for Databases as an MCP tool provider
• The agent invokes database operations through MCP tools instead of building raw SQL
• The toolbox enforces a safe, predefined set of actions that agents can perform

This fits the ADK model of general tooling, where agents use a combination of built-in tools, Google Cloud tools, ecosystem tools, and MCP tools, all of which are described in the Vertex AI Agent Builder documentation.

Integrate with Vertex AI Agent Builder and Agent Engine

ADK is the main framework supported by Vertex AI Agent Builder for building multi-agent systems.

The latest Agent Builder update describes build paths where you can:

1. Develop agents locally using ADK, now including ADK for Go
2. Get started quickly and develop UI using ADK Test agents with multiple tools
3. Deploy the agent as a managed runtime to Vertex AI Agent Engine

For the Go team, this means that the language used in services and infrastructure can now be used throughout the entire agent lifecycle, from local development to managed production deployments.

This release positions the Go Agent Development Kit as a practical bridge between AI agents and existing Go services, using the same open source, code-first toolkit that supports Python and Java agents. It brings A2A protocol support and MCP Toolbox for Databases into the Go native environment, aligned with Vertex AI Agent Builder and Vertex AI Agent Engine for deployment, evaluation, and observability. Overall, this release makes Go a first-class language for building production-ready, interoperable AI agents in the Google ecosystem.

Check repurchase agreement, sample and technical details. Please feel free to check out our GitHub page for tutorials, code, and notebooks. In addition, welcome to follow us twitter And don’t forget to join our 100k+ ML SubReddit and subscribe our newsletter. wait! Are you using Telegram? Now you can also join us via telegram.