Get a complete inventory of your Google Cloud AI assets with one integration

Your team can now discover all AI agents and MCP servers running across your Google Cloud projects from a single integration, replacing the per-project setup that left gaps in your inventory.

The previous Google Agent Engine integration discovered only AI agents, returned limited metadata, and was scoped to a single Google Cloud project at a time. For organizations running AI across multiple projects and regions, this meant an incomplete, thin picture of what was actually deployed.

The Google Agent Registry integration closes that gap. It discovers both AI agents and MCP servers, and captures the context you need to make governance decisions: runtime, agent type, origin, skills, service endpoints, version, and location. Connections between agents, their MCP servers, and their endpoints are mapped automatically, so you can see how capabilities relate to each other. With agents and MCP servers represented as fact sheets alongside your existing application landscape, you can map dependencies, assign ownership, and assess risk the same way you would for any other technology asset.

Discovered items appear in the AI Agent Discovery Inbox, where you can link them to existing fact sheets, create new ones, or reject them. If you have the previous Google Agent Engine integration set up, migration is not mandatory as your existing integration will continue to run. However, we strongly encourage the switch, as the new integration provides significantly more coverage.