Labspace - DevDuck Multi-Agent Systems

This Labspace provides an overview of the DevDuck Multi-Agent System, a secure AI coding agent for Node.js programming assistance built with Google Agent Development Kit (ADK). The system features a coordinating agent (DevDuck) that manages two specialized sub-agents (Local Agent and Cerebras) for different programming tasks.

Learning objectives

This Labspace will teach you:

  • Multi-agent coordination and intelligent routing between specialized AI agents
  • Google Agent Development Kit (ADK) architecture and implementation
  • Agent specialization patterns (coordination vs execution vs analysis)
  • Docker Compose orchestration for multi-agent systems
  • Agent-to-agent communication and task delegation
  • Debugging agent decision-making through trace visualization

System Architecture

Three-Agent System:

  • DevDuck Agent: Main coordinator using Qwen3 model
  • Local Agent: General development tasks using Qwen2.5 model
  • Cerebras Agent: Complex problem-solving using Llama-4 Scout via Cerebras API

Launch the Labspace

To launch the Labspace, run the following command:

docker compose -f oci://dockersamples/labspace-devduck up -d

And then open your browser to http://localhost:3030.

Using the Docker Desktop extension

If you have the Labspace extension installed (docker extension install dockersamples/labspace-extension if not), you can click this link to launch the Labspace.