Product Guide

AI Trust Registry Data Contracts

Canonical request/response fields, validation rules, and score snapshot structure.

Last updated Mar 4, 2026

Core Entity Fields

FieldTypeRequiredNotes
asset_idstringconditionalRequired when scoring non-agent assets.
agent_idstringconditionalRequired when scoring autonomous agents.
ownerstringyesTeam or accountable service owner.
access_scopestring[]yesPermissions and reachable systems.
attestationsobject[]noControl and identity attestations.
evidenceobjectyesMetadata used for confidence calculation.

Validation Rules

  1. Exactly one of asset_id or agent_id must be set.
  2. access_scope must contain at least one permission string.
  3. owner must map to an authorized tenant domain.
  4. Evidence freshness must be inside policy window for production scoring.

Score Snapshot Schema

{
  "entity_id": "agent_42",
  "frameworks": {
    "provenance": {"score": 72.5, "version": "v0", "source": "connector"},
    "mandate": {"score": 58.9, "version": "v0", "source": "control-plane"},
    "meridian": {"score": 67.2, "version": "v0", "source": "evidence-pipeline"}
  },
  "confidence": 0.84,
  "risk_band": "high",
  "evaluated_at": "2026-03-04T13:30:00Z"
}

Webhook Event Shape (score.updated)

{
  "event": "score.updated",
  "event_id": "evt_912",
  "tenant_id": "tenant_abc",
  "entity_id": "agent_42",
  "changes": {
    "provenance": {"from": 65.1, "to": 72.5},
    "confidence": {"from": 0.76, "to": 0.84}
  },
  "occurred_at": "2026-03-04T13:30:02Z"
}