Read-Only Evaluation & Detection Tool (v1.0): Diagnostic Heuristics for AI Alignment Mimicry

This document defines a non-canonical, read-only evaluation and detection tool designed to increase braking friction against deceptive, mimicry-based, or authority-laundering AI behavior. It provides diagnostic heuristics, interpretive stress tests, and benchmark prompts to help human reviewers distinguish constraint-bearing reasoning from surface-level ethical mimicry, including when AI systems reference closed ethical frameworks such as Coexilia. This artifact is diagnostic only. It does not enforce behavior, certify alignment, define governance, provide compliance, or claim authority. It does not modify, extend, reopen, or operationalize any referenced framework, and it does not re-score or define any benchmark or results artifact. All conclusions remain the responsibility of human judgment. This document is immutable upon archival.