> ## Documentation Index
> Fetch the complete documentation index at: https://docs.humanos.tech/llms.txt
> Use this file to discover all available pages before exploring further.

# Risk Metrics

> How to access agent risk intelligence, for your own organization or a partner's.

The intelligence endpoints expose risk metrics for AI agents operating under Humanos mandates.
Instead of sharing raw events or credentials, the API answers business questions: was this agent
authorized, who consented to its authority, how does it behave, and can its record be trusted.

Every figure is computed from signed, tamper-evident data: the agent's mandates, the lifecycle
events chained to them, the identity verification (KYC) of the humans who granted authority, and
the agent's own registration (KYA). Responses carry aggregates only; user identities, mandate
identifiers and contacts never cross organizations.

## Accessing intelligence data

All intelligence endpoints take an `orgDid` parameter naming the organization whose data is
consulted.

* **Your own organization**: pass your organization's DID. No sharing setup is needed; you can
  always consult your own agents.
* **Another organization**: pass that organization's DID. This works only while that organization
  has an active sharing grant to yours, managed by them in the dashboard under Settings. Without
  a grant, the request returns `404 sharing_not_enabled`. The same 404 is returned whether the
  organization does not exist, never shared with you, or revoked its grant, so organizations
  cannot be discovered by probing.

The `did` parameter names the agent to measure. Every figure is scoped to the pair: the agent's
mandates and events within the named organization's data.

## The endpoints

* [Agent trust snapshot](/api-reference/latest/intelligence/agent-trust-snapshot): the risk
  picture of one agent at a single instant. Registration, live mandates, bounds and execution
  headroom, human consent, and chain integrity.
* [Agent trust window](/api-reference/latest/intelligence/agent-trust-window): what one agent
  did over a period. The request funnel, verification activity, counterparties, burn rate, and
  incidents.
* [Resolve an action](/api-reference/latest/intelligence/resolve-an-action): both bodies report
  per-action figures keyed by action URN. This endpoint resolves a URN into its definition, so
  you can see what an execution of that action actually permits.