Registering an agent transforms it from an isolated workload into a discoverable service. By publishing your agent's skills, you enable other developers and orchestrator agents in your organization to find and reuse your work, accelerating overall AI development.
The fastest way to register an agent is to deploy it using Vertex AI Agent Engine because it features automatic registration. When you deploy your agent, Agent Registry automatically detects the resource, extracts its metadata and skills, and publishes it to your registry.
This quickstart shows you how to automatically register an agent deployed to Vertex AI Agent Engine in Agent Registry. To register custom agents hosted on other Google Cloud runtimes, or if you want to manually register remote agents using the Agent Registry API, see Register agents.
Before you begin
To get started, set up Agent Registry.
To follow this quickstart, install the Vertex AI SDK for Python in your environment, along with Agent Development Kit (ADK).
Deploy and register an agent automatically
You can use ADK to define your agent and the Vertex AI SDK to deploy it. You set up your Google Cloud project and then develop and deploy your ADK agent to the Vertex AI Agent Engine runtime.
When you provide your agent through this process, the system deploys it within Vertex AI Agent Engine and automatically adds it to Agent Registry. For a complete example of the deployment process, see the quickstart with Agent Development Kit for Vertex AI Agent Engine.
Deploying an agent to Vertex AI Agent Engine generates a managed resource. As you update or delete your agent in Vertex AI Agent Engine, Agent Registry synchronizes those changes for the agent registration.
Verify registration
To confirm that your agent was successfully registered, you can list, filter, and view the details of this agent in Agent Registry.