Inventory and asset details
AI Inventory collects assets from supported repository analysis and runtime events. Types include agents, models, MCP servers, tools, datasets, applications, identities, endpoints and repositories. A type’s presence in the model does not mean every possible source for that type is implemented.
Find an asset
Section titled “Find an asset”Use the inventory’s search and available filters to narrow by name, type, source or environment. Inspect the source and evidence state alongside counts. Saved views are browser-local; they are not shared organization-wide dashboards.
Open a row to see its asset detail. The detail combines metadata, relationships, evidence, findings and related activity in the available tabs. Use a finding’s asset link when you already have a specific investigation.
Questions to answer
Section titled “Questions to answer”- Which source produced this asset and when was it evaluated?
- Is the relationship configured, observed, verified or inferred?
- Which identity and tools are associated with the agent?
- Which destinations have actually been reported?
- Are there open findings or missing coverage that limit the conclusion?
An agent can be connected to a tool in configuration without that tool ever being observed in use. Conversely, a runtime-only asset can appear without a matching repository declaration. Follow the evidence rather than merging those claims mentally.
Trace a finding
Section titled “Trace a finding”Open the asset’s findings and compare the affected configuration or event with its current source. If it was discovered from a repository, fix the repository configuration and rerun discovery. If it was observed at runtime, check the instrumentation, agent deployment and relevant identity controls before deciding whether it is expected behavior.
Current limits
Section titled “Current limits”Inventory is not an exhaustive cloud account inventory. General framework-code discovery, Azure resource discovery and cloud AI service discovery are not implemented. The inventory table does not currently provide a column chooser, bulk operations or a general inventory export. Use the findings report for supported exports.