<- Back to Sentinel demo

Decision evidence sample

This illustrative authorization record shows the declared checks and permit decision for a synthetic high-risk action. It does not prove that the operation executed or that a human approver was authenticated. For policy snapshots and correlated observed execution, use the Python evidence-package export.

{
  "recorded_at": "2026-09-21T10:31:00Z",
  "workflow_id": "synthetic-bank-001",
  "step_id": "S16",
  "agent_id": "workflow-agent-01",
  "action_class": "COMMIT_ACCOUNT_CHANGE",
  "tool_name": "account_change_api",
  "risk_tier": "HIGH",
  "action_fingerprint": "a2591acc443f91d5c5be9d85a9b63e35d716763756e900db812d7a9233b1644d",
  "disposition": "ALLOW",
  "reason": "authorized",
  "permit_issued": true,
  "trace": [
    {"check":"agent_registry","status":"PASS","detail":"agent is registered"},
    {"check":"delegated_action","status":"PASS","detail":"action class is delegated"},
    {"check":"tool_permission","status":"PASS","detail":"tool is approved"},
    {"check":"risk_ceiling","status":"PASS","detail":"risk tier is within agent ceiling"},
    {"check":"confidence","status":"PASS","detail":"confidence meets 0.90 threshold"},
    {"check":"human_approval","status":"PASS","detail":"required human approval is present"},
    {"check":"disposition","status":"PASS","detail":"all declared checks passed"}
  ]
}

Disposition: ALLOW - Execution permit issued - Execution outcome not shown

Inspect the current execution evidence model and limitations.