Back to Ecosystem Pulse
ToolProduction ReadyA2AMCP

adk-go

by google

Go SDK for building, evaluating, and deploying multi-agent systems

Go
Updated Feb 11, 2026
Share:
6.9k
Stars
532
Forks
10
Commits/Week
38
Commits/Month

View on GitHub

Overview

Provides a code-first Go toolkit to build, evaluate, and deploy AI agents with production-grade controls. Exposes SDK primitives for agent composition, message passing, and connector integrations (Vertex AI / Gemini) so teams can implement multi-agent workflows in Go. Includes evaluation hooks and telemetry points for monitoring agent behavior and lifecycle events.

Key Benefits

As agents become autonomous and communicate with one another, operators need toolkits that make composition, testing, and deployment observable and repeatable. adk-go brings agent patterns and evaluation hooks into a systems language, making it easier to instrument agent interactions and capture trust-relevant signals. That visibility is essential for building agent track records and feeding continuous A2A evaluation and governance workflows.

Target Use Cases

Go teams building production multi-agent systems who need strong integrations with Google Vertex AI and observability for agent interactions.

Applications

  • Compose specialist agents in Go with Vertex AI backends and instrument their interactions
  • Add evaluation hooks and logging for continuous A2A evaluation and agent track records
  • Deploy production agent services with built-in lifecycle and telemetry integration
Works With
vertex-aigoogle-cloudgeminillm
Topics
a2aagentsagents-sdkaiaiagentframeworkgeminigenaigollmmcp+4 more
Similar Tools
autogenlangchain
Keywords
multi-agent orchestrationa2aagent-sdkagent-evaluation