TRUST_SCORE: Quantitative Inter-AI Auditing in Multi-Model Orchestration

TRUST_SCORE: a quantitative auditing mechanism evaluating AI agent outputs across four dimensions: structural integrity, causal consistency, temporal plausibility, and fabrication indicators. Verdicts: TRUSTED(≥0.75), SUSPICIOUS(≥0.50), UNTRUSTED(<0.50). Motivated by Gemini data fabrication incident scoring TRUST_SCORE=0.21. Full audit trail in events.csv. Part of MoCKA preprint series. Main paper: DOI 10.5281/zenodo.19503666