We launched Agent Scanners in January with a simple goal: automatically find agents, MCP servers, and APIs your teams are building, wherever they are building them, and bring them into a cross platform registry. Since then, we’ve kept expanding its capabilities with support for more providers and types of assets. The registry now covers more platforms, connects more securely, and always refreshes on its own.
We are releasing several advanced capabilities in our recent release to meet our customers’ needs:
- A new scanner for ServiceNow
- OAuth 2.0 authentication support for the Google Vertex AI and Microsoft Copilot Studio scanners
- Much deeper metadata about each agent, across every scanner
ServiceNow: agents where your workflows already live
Many large enterprises use ServiceNow to run the workflows behind IT, HR and customer service. ServiceNow agents sit close to your most sensitive data. They read and update incidents, change requests, employee records, and customer cases. That’s what makes them useful, and it’s why they need to be visible and governed.
The new ServiceNow scanner discovers agents built with Now Assist AI Agents in AI Agent Studio. For each agent, it captures the name, description, instructions, role and available tools. Where your instance supports it, the scanner also pulls each agent’s A2A agent card, the open standard for describing what an agent does and how to reach it. Your ServiceNow agents are catalogued in the same registry as your agents from AWS, Microsoft, Google, Salesforce, Databricks, Snowflake, and more, allowing you to enforce a consistent governance posture.
Discovered agents can go through manual review before they’re imported. To set up the ServiceNow Scanner , you need.
* AI Asset Governance enabled and synced on your ServiceNow instance
* A ServiceNow Table API integration user with read access to the `alm_ai_*` tables
* An Application Registry entry in ServiceNow, which generates the OAuth client ID and client secret pair that the scanner uses to authenticate
OAuth 2.0 for Google Vertex AI and Microsoft Copilot Studio
Security teams do not prefer using long-lived secrets. To enhance the security posture of scanners, the Google Vertex AI and Microsoft Copilot Studio scanners now support OAuth 2.0 with bring your own app.
You register an app in your own Google Cloud project or Microsoft Entra tenant, grant it only the permissions the scanner needs, and connect Agent Fabric to it. The app stays in your tenant. You control its scopes, rotate its credentials on your schedule, and revoke access whenever you want. Here are the prerequisites for setting up the OAuth2.0 for your scanners:
- Microsoft Copilot Studio: Register an OAuth app in your Microsoft Entra tenant, configure it to request the Dynamics CRM user_impersonation scope, and set the callback URL to your Anypoint Secrets Manager OAuth endpoint.
- Google Vertex AI: Register an OAuth app in your Google Cloud project, grant it access to the Vertex AI API, and set the callback URL to your Anypoint Secrets Manager OAuth endpoint.
Richer metadata now available across all agents discovered by Scanners
Finding an agent should be the easy part. To govern it effectively you need more information such as what model it runs on, what tools it can call, what data it can reach, and who owns it. Our scanners now pull much deeper metadata from each platform, so that context is on the agent card the moment it lands in the registry.
| What we capture | Why it matters |
|---|---|
| Models: Which models each agent runs on | Find every affected agent fast when a model changes or is retired |
| Tools and actions: the functions and APIs an agent can call | See what an agent can actually do, and the actions it can take |
| Data sources: the knowledge bases and data stores an agent connects to | Understand what enterprise data an agent can reach |
| Access and authentication: how an agent is authenticated and who can call it | See who and what is allowed to call each agent |
| Ownership: who created, owns, and last changed the agent | Know who to talk to when something needs to change |
| Multi-agent setup: which agents work together | See how agents hand work to each other |
| Lifecycle : an agent’s status, version, and when it was created or last updated | Spot stale or abandoned agents and track what changed between versions |
| Environment: whether an agent runs in production or sandbox | Focus governance on production agents without noise from experiments |
Coverage depends on what each platform exposes, so the exact fields vary by scanner. The direction is the same everywhere: more of what you need to govern an agent, captured automatically.
What’s next?
Coverage, secure access, and context are what turn a list of discovered agents into a registry you can govern. With this release, your registry catalogues agents from ServiceNow, connects the way your security team expects, and knows far more details about every agent in it.
Next, we’re building on this metadata to show how agents connect to tools, MCP servers, and the data behind them.
See the list of all available Agent Scanners, and learn more about Agent Fabric.
