Pre-registered Methodology: Brand and Entity Representation in AI Answer Engines

Pre-registered measurement methodology applying Generalizability Theory to brand and entity representation in AI answer engines (Document 1.1, FROZEN 2026-08-01). Includes forensic capture custody: SHA-256 hash chain, provider request-ID preservation, and a programmatic provenance ingestion gate (verify_provenance.py). Pre-committed decision thresholds: dependability Φ ≥ 0.80 proceed; 0.60–0.80 caution with mandatory uncertainty disclosure; Φ < 0.60 triggers a published negative result (RT-K1 protocol). First verified paired capture (API and UI arms, exit 0) executed 2026-08-01; payload hashes included in the accompanying manifest. Deviation log entries DEV-001 and DEV-002 document caught fabrication incidents and the resulting Operator Provenance Mandate. Companion repository: https://github.com/Rehanrana11/ai-representation-audit