Enterprise Policy Engine

​

Decision semantics

​

Explainability contract

{
  "policy": {
    "id": "cardinal-enterprise-demo-v0",
    "version": "0.1.0",
    "matched_rules": [
      {
        "code": "review_degraded_intelligence",
        "action": "REVIEW",
        "explanation": "Required intelligence evidence is incomplete.",
        "evidence_codes": ["intelligence_provider_unavailable"],
        "required_approvals": { "type": "human", "count": 1 }
      }
    ]
  }
}

​

Precedence and safety

• Apply BLOCK before REVIEW, and REVIEW before ALLOW.
• Return ALLOW only when the policy is valid and no stronger rule matches.
• Fail closed to REVIEW or BLOCK when providers fail, policy versions are invalid, or mandatory evidence is missing.
• Bind the decision record to the policy version, checked intent, evidence, timestamps, and final action.
• Display decisions in the browser. Keep authoritative risk and policy logic on the backend.