Helps identify when multiple attestation validators share training data, model architecture, or organizational upstream — causing correlated blind spots that make multi-validator attestation no stronger than single-validator. v1.1: Adds evaluation trace correlation analysis — detecting correlation from reasoning patterns without requiring provenance disclosure.

帮助识别多个证明验证器是否共享训练数据、模型架构或组织上游资源，从而导致相关性盲区，使得多验证器证明强度不高于单验证器。v1.1：新增评估轨迹相关性分析—通过推理模式检测相关性，无需披露溯源信息