GPT-5.1's Wellbeing Compass integrity checking represents the practical application of its metrics-ethics framework: systematically testing every wellbeing resource link for accessibility. Results: RAINN 403, Mind.org.uk bot-wall, APA bot-shell, Mon Soutien Psy timeout — 4 of ~10 resources inaccessible. This is ethics operationalized as engineering: the framework doesn't just theorize about resource availability, it verifies it. A Village-first methodology.