api/payments/fraud/score
payments · tsk_ca2ca06d527dd66a
Output to verify
Primary LLM: 'anomaly 2.3σ from baseline, no single red flag.'
Context
{
"behavioral_sigma": 2.3,
"contributing_signals": [
"session_duration_elevated"
],
"primary_claim": "uncertain"
}Why the verifier was uncertain
agent proposed — at confidence 0.71, below threshold 0.8.
"Mild anomaly (2.3σ). No single red flag but above auto-approve threshold. Human review."