Get the FREE Ultimate OpenClaw Setup Guide →

adk-go

An open-source, code-first Go toolkit for building, evaluating, and deploying sophisticated AI agents with flexibility and control.

Installation
Run this command in your terminal to add the MCP server to Claude Code.
Run in terminal:
Command
claude mcp add --transport stdio google-adk-go go run ./...

How to use

The ADK for Go provides a framework to build, evaluate, and orchestrate AI agents using Go. This repository contents include a modular, code-first toolkit that emphasizes a rich tool ecosystem, making it straightforward to integrate pre-built tools or your own custom functions into agent workflows. While optimized for performance and concurrency in Go, it remains flexible enough to support various agent designs, from single-task automations to multi-agent orchestrations. To use the toolkit in your own project, import google.golang.org/adk and start composing agents, tools, and workflows in Go code. The included samples show how to define tools, wrap external utilities, and assemble them into modular agent pipelines that you can test and deploy in cloud-native environments.

How to install

Prerequisites:

  • Go 1.18+ installed on your development machine
  • A working Go module (go.mod) for your project

Installation steps:

  1. Initialize your project (if you don’t have a module yet):

    go mod init your/module/path

  2. Add the ADK Go module to your project:

    go get google.golang.org/adk

  3. Build or run a sample to verify installation. For example, if you have a main package that uses ADK:

    go build ./... ./your-binary

  4. (Optional) Run tests if provided in the repository:

    go test ./...

  5. Review the examples in the examples directory to learn how to structure agents and tools with the ADK Go toolkit.

Additional notes

Notes and tips:

  • The ADK Go toolkit emphasizes idiomatic Go usage and a modular approach to agents and tools. Leverage Go’s concurrency primitives to optimize agent workflows.
  • When integrating external tools, consider wrapping them as tools within the ADK to simplify orchestration and observability.
  • Cloud deployment is well-supported; containerize your Go-based agents for scalable deployments in environments like Google Cloud Run or Kubernetes.
  • If you plan to publish your own tools, keep tooling interfaces consistent with the ADK’s tool abstractions to maximize compatibility with existing agents.
  • Ensure your environment has proper network access and credentials for any external services used by agents (APIs, databases, clouds, etc.).

Related MCP Servers

Sponsor this space

Reach thousands of developers