go-adk

Build and orchestrate Go AI agents with Google's Agent Development Kit.

1|Updated Feb 19, 2026

Search Tags:#workflow #llm #tooling #mcp #agent-development #a2a #adk-go

Ready to run

npx skills add https://github.com/juburr/mad-skills --skill go-adk

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: go-adk
Source: https://github.com/juburr/mad-skills/tree/main/go-adk
Command: npx skills add https://github.com/juburr/mad-skills --skill go-adk

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

ADK Go provides a code-first toolkit for building AI agents in Go, enabling seamless integration with Gemini or other GenAI backends and offering a rich set of orchestration patterns and tooling to compose multi-agent workflows.

Core Features & Use Cases

Code-first agent building with llmagent, sequentialagent, parallelagent, and loopagent for structured workflows.
MCP and A2A integrations to connect local and remote agents, plus tooling like mcptoolset and agenttool.
Memory and artifact tooling, dynamic instruction providers, and plugin extensibility for production-grade agent runtimes.

Quick Start

Create a Gemini-backed model and instantiate an agent with llmagent.New using a basic Config, then run it with a runner.

Frequently Asked Questions about go-adk

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I build AI agents in Go using Google's Agent Development Kit?▼

To build AI agents in Go, use the ADK to instantiate a Gemini-backed model and configure an agent via llmagent.New, then execute it with a runner. This provides a code-first approach to defining agent names, models, and toolsets.

What orchestration patterns are available for multi-agent workflows in Go?▼

Multi-agent workflows in Go can be orchestrated using sequential, parallel, and loop agent patterns. These structured workflow components allow developers to compose complex multi-agent systems within the ADK runtime.

Can I connect local MCP servers and remote A2A agents in a Go agent runtime?▼

Yes, Go agent runtimes can connect local MCP servers and remote A2A agents. The ADK provides mcptoolset and agenttool integrations to facilitate communication between local and remote agents in your workflow.

Do I need a specific Go version to use Google ADK for agent development?▼

Yes, agent development with Google's ADK requires Go version 1.24.4 or higher. This environment prerequisite ensures compatibility with the ADK interfaces needed to map agent names and toolsets to a runnable runtime.

What is the best way to manage memory and artifacts for production AI agents in Go?▼

For production AI agents in Go, manage memory and artifacts using the ADK's built-in memory and artifact tooling. This is complemented by dynamic instruction providers and plugin extensibility to create robust agent runtimes.

When should I use a code-first approach versus visual builders for AI agent development?▼

A code-first approach is ideal when you need seamless integration with GenAI backends like Gemini and require fine-grained control over tool definitions and orchestration patterns. It suits developers building structured, production-grade multi-agent workflows.