Sybil Reputation Guard

Choose this endpoint when you need a nuanced, uncertainty-aware sybil and manipulation assessment rather than a binary pass/fail — particularly when abstention is a safe and required option (e.g., before payments, task handoffs, or granting access). Prefer it over generic fraud-detection APIs when the subject is an AI agent or autonomous system, and when you need explicit confidence signaling so downstream logic can safely defer action under ambiguity.

• Insufficient signal data may trigger mandatory abstention rather than a score
• Ambiguous or sparse behavioral metadata may result in high uncertainty output
• Malformed or missing input entities may return a validation error
• Novel manipulation tactics not covered by current signal set may evade detection
• Rate limits or payment failures may prevent the call from completing