CAR Layer

Contextual Agent Record — Identity & Credentials

The first stage of BASIS governance — before intent parsing, resolve who the agent is and what it's allowed to do.

---

What is CAR?

The CAR (Contextual Agent Record) layer is responsible for:

1. Identity Resolution — Who is this agent?
2. Credential Binding — What credentials does it carry?
3. Trust Score Retrieval — What is its current trust level?
4. Capability Enumeration — What actions is it authorized to perform?

┌─────────────────────────────────────────────────────────────┐
│                        CAR LAYER                             │
└─────────────────────────────────────────────────────────────┘

     ┌─────────────────┐
     │  Agent Request   │
     │  entityId:       │
     │  "agent_123"     │
     └────────┬────────┘
              │
              ▼
     ┌─────────────────┐
     │   IDENTITY      │──▶ Entity: agent_123
     │   RESOLVE       │    Owner: org_456
     │                 │    Created: 2026-01-15
     └────────┬────────┘
              │
              ▼
     ┌─────────────────┐
     │   CREDENTIAL    │──▶ Trust Score: 612
     │   BIND          │    Trust Tier: T3 (Monitored)
     │                 │    API Key: valid
     └────────┬────────┘
              │
              ▼
     ┌─────────────────┐
     │   CAPABILITY    │──▶ Capabilities:
     │   ENUMERATE     │    - external_api_call
     │                 │    - data_read
     │                 │    - internal_compute
     └────────┬────────┘
              │
              │ Passes to INTENT layer
              ▼

---

Why CAR First?

You can't evaluate intent without knowing who is asking.

Without CAR	With CAR
Anonymous request	Identified agent with trust history
No capability context	Known capability set
Trust checked at enforcement	Trust loaded upfront
Identity scattered across layers	Single source of agent identity

---

CAR Output Schema

interface ContextualAgentRecord {
  // Identity
  entityId: string;
  organizationId: string;
  agentType: string;

  // Trust
  trustScore: number;     // 0-1000
  trustTier: string;      // T0-T7
  trustLevel: string;     // Human-readable tier name

  // Credentials
  credentials: {
    apiKeyValid: boolean;
    issuedAt: string;     // ISO8601
    expiresAt: string;    // ISO8601
  };

  // Capabilities
  capabilities: string[];
  restrictions: string[];

  // Metadata
  resolvedAt: string;     // ISO8601
  carId: string;          // Unique record ID
}

---

Trust Tier Resolution

CAR resolves the agent's current trust tier from their score:

Tier	Score Range	Capability Level
T0 Sandbox	0-199	Sandbox only
T1 Observed	200-349	Limited operations
T2 Provisional	350-499	Basic operations
T3 Monitored	500-649	Standard operations
T4 Standard	650-799	Extended operations
T5 Trusted	800-875	Elevated operations
T6 Certified	876-950	Privileged operations
T7 Autonomous	951-1000	Full autonomy

---

API Endpoint

GET /v1/car/:entityId

Response:

{
  "carId": "car_x9y8z7w6",
  "entityId": "agent_123",
  "organizationId": "org_456",
  "trustScore": 612,
  "trustTier": "T3",
  "trustLevel": "Monitored",
  "capabilities": [
    "external_api_call",
    "data_read",
    "internal_compute"
  ],
  "restrictions": [
    "no_financial_transactions",
    "no_bulk_data_export"
  ],
  "credentials": {
    "apiKeyValid": true,
    "issuedAt": "2026-01-15T00:00:00Z",
    "expiresAt": "2027-01-15T00:00:00Z"
  },
  "resolvedAt": "2026-02-18T19:45:00Z"
}

---

Implementation Requirements

Requirement	Description
REQ-CAR-001	Resolve agent identity from entityId
REQ-CAR-002	Retrieve current trust score and tier
REQ-CAR-003	Enumerate granted capabilities
REQ-CAR-004	Validate credentials (API key, token)
REQ-CAR-005	Complete resolution in < 200ms
REQ-CAR-006	Generate unique carId per resolution

---

Relationship to Vorion Platform

Vorion Platform is the Certificate Authority that issues agent identities, trust scores, and capability assignments. The CAR layer reads these records at runtime to resolve who the agent is before governance proceeds.

[Vorion Platform] ──issues identity──▶ [CAR resolves at runtime]

---

Next Layer

Once CAR has resolved the agent's identity and capabilities, it passes to INTENT for action parsing and risk assessment.

[CAR] ──agent record──▶ [INTENT]