Google ADK (Agent Development Kit) is Google’s framework for building Gemini-powered agents. Idun Agent Platform wraps ADK agents as production services with AG-UI streaming, guardrails, and observability.
Want to start from working code? The agent templates include ADK examples for tool calling and structured I/O. Create an ADK agent
Create the agent in the Manager, then connect your code.Open the agent wizard
From the Agent Dashboard, click Create an agent. In step 1, enter a name and select Google ADK.
Configure the framework
In step 2, fill in:
- Agent definition: Python module path and variable, e.g.
my_agent.agent:root_agent
- App name: application name for ADK session/memory scoping
- Agent Host: Localhost or Remote
- Server Port: the port your agent will listen on
The agent definition uses Python module path notation (module.path:variable), not a file path. This differs from LangGraph. Enroll your agent
Step 3 gives you the enrollment instructions. In your agent’s project directory:pip install idun-agent-engine google-adk
export IDUN_MANAGER_HOST=http://localhost:8000
export IDUN_AGENT_API_KEY=<your-api-key>
idun agent serve --source manager
Set agent.type to ADK and provide the agent definition path and app_name.server:
api:
port: 8008
agent:
type: ADK
config:
name: "my-adk-agent"
agent: "my_agent.agent:root_agent"
app_name: "myagent"
pip install idun-agent-engine google-adk
idun agent serve --source file --path config.yaml
Write your agent code
Create an ADK agent at the module path matching your configuration:
from google.adk.agents import Agent
root_agent = Agent(
model="gemini-2.5-flash",
name="weather_agent",
description="An agent that answers questions about the weather.",
instruction="You are a helpful weather assistant. Answer questions about weather conditions.",
)
If you use Vertex AI models, authenticate with gcloud auth application-default login before running the agent.
The agent field
The value uses Python module path notation with a colon separator:
- Module path: dotted Python module path (e.g.,
my_agent.agent)
- Variable name: the agent instance in that module (e.g.,
root_agent)
This differs from LangGraph’s graph_definition, which uses a file path.
Session and memory services
ADK agents have two persistence layers: session services (conversation state) and memory services (long-term recall). Both default to in-memory when using the Manager. See Memory and sessions for ADK for backend options including Database (PostgreSQL) and Vertex AI.
ADK does not support folder paths that contain spaces. Make sure your project directory path has no spaces.
Next steps
