Connectors
Tools that send findings into AISPM, and systems AISPM sends them to. AISPM collects, prioritizes and remediates what the tools detect.
Connected
| Connector | Direction | Mode | Status | Feeds | Covers | Last report |
|---|---|---|---|---|---|---|
| Inbound | Link | Reporting | UC1UC2 | ASI06LLM01LLM02LLM08 | — | |
Hugging Face HubModel registries / hubs | Inbound | Live | Reporting | UC5 | LLM03 | — |
Model eval monitorModel evaluation | Inbound | Simulated | Reporting | UC4 | LLM03 | — |
Agent permissions scanAgent permissions | Inbound | Simulated | Reporting | UC3 | ASI01ASI02ASI03LLM06 | — |
Status, Feeds and Covers follow the use cases each connector reports for, so toggling a use case in the sidebar changes them. Connectors added in the demo report nothing and do not change the Coverage table.
Hugging Face Hub: live read
Reads a public repo's metadata and the hub's own security status. It does not download model files, and AISPM does not scan anything: it only shows what the hub reports.
Available connectors
Connecting one is a demo only: it moves into the table above and reports nothing.
AI gateway / guardrails
An AI gateway that would report blocked and flagged requests.
A model proxy and gateway that would report routing and policy events.
Model registries / hubs
A model registry that would report model versions and stage changes.
Vector databases
A managed vector database that would report indexes and ingestion events.
A vector database that would report collections and ingestion events.
A vector database that would report collections and ingestion events.
Agent observability
Agent observability that would report agent runs and tool calls.
Skill files
Would report where skill files live (drives, cloud storage, laptops, repos), who changed them and which agents load them. AISPM never reads their content.
Exposure management (CTEM)
AISPM findings are pushed upward into the broader cross-domain exposure graph. AISPM is not a CTEM platform.