AI Trust Registry Integration Playbook

Integration Sequence

1. Register tenant credentials and webhook endpoint.
2. Wire discovery pipeline to scoring endpoints.
3. Configure policy thresholds by environment (dev, staging, pilot-prod).
4. Validate webhook signatures and idempotent processing.
5. Enable approval automation and remediation workflows.

Reference Flow

1Inventory source

→

2Adapter/Connector

→

3/v1/assets/score and /v1/agents/score

→

4Registry score + confidence snapshot

→

5Policy evaluation and remediation routing

Emits

allow onboardingrequire remediationblock access

Sample Request: Asset Score

POST /v1/assets/score
Authorization: Bearer <token>
X-Tenant-Id: tenant-a
Content-Type: application/json

{
  "tenant_id": "tenant-a",
  "asset_id": "asset-a-001",
  "event_id": "evt-asset-a-1001",
  "score_version": "provenance_v0",
  "discovery_confidence": 0.88,
  "control_readiness": 0.61,
  "evidence_events": [
    {
      "evidence_id": "asset-ev-a-1",
      "event_type": "inventory.scanned",
      "quality": 0.9,
      "completeness": 0.8,
      "weight": 1.0
    },
    {
      "evidence_id": "asset-ev-a-2",
      "event_type": "metadata.extracted",
      "quality": 0.7,
      "completeness": 0.6,
      "weight": 0.5
    }
  ]
}

Sample Response

{
  "accepted": true,
  "tenant_id": "tenant-a",
  "actor_id": "actor-a",
  "request_id": "fixture-tr-asset-positive-a",
  "asset_id": "asset-a-001",
  "entity_id": "asset-a-001",
  "event_id": "evt-asset-a-1001",
  "score_version": "provenance_v0",
  "scores": {
    "provenance_v0": 0.79,
    "meridian_v0": 0.7833,
    "composite_v0": 0.787
  },
  "risk_priority": "medium",
  "trend": {
    "direction": "flat",
    "delta": 0.0,
    "previous_composite_v0": null,
    "current_composite_v0": 0.787
  },
  "lineage": {
    "evidence_event_id": "53f28c42-53cc-4e0f-ac55-3ec0589eb073",
    "score_snapshot_id": "7e889839-f714-4e4e-8f96-4728ef3087fc"
  }
}

Rollout Strategy

Phase A: Shadow Mode

• Score entities but do not block approvals.
• Compare recommendation vs real operator decision.
• Tune thresholds and confidence floor.

Phase B: Guardrail Mode

• Block only critical band entities.
• Auto-create remediation tickets for high band.

Phase C: Enforcement Mode

• Apply policy thresholds to all production onboarding paths.
• Report weekly drift in score distribution.