nx_uiq_contrast.nx

nx_uiq_contrast.nx -- CLI: FULL-PAGE COMPUTED WCAG 2.x contrast auditor over REAL extracted colors. The quantitative fix for the "presence != conformance" gap: nx_ui_contrast checked ONE hardcoded token pair on 6 hardcoded pages; this parses ANY page, extracts every --custom-property color, and computes the real WCAG ratio for every text-role x surface-role pair -> the failures the presence-judge misses. usage: nx_uiq_contrast --kat (self-test: math + parser + non-vacuous audit) nx_uiq_contrast <html-file-path> (emits JSON: tokens, failing pairs, ratios, verdict) HONEST ENVELOPE: static/source-measured over declared color tokens; role pairing is a name heuristic; true rendered element-on-background stacking + non-text (3:1) + large-text (3:1) tiers = browser-lane. Automated contrast is one axe rule; automated a11y covers ~57% of ISSUES / ~30-40% of WCAG SC (a FLOOR, never proof of conformance). license_tier: ORIGINAL expect_exit: 0 genealogy: w3.org/TR/WCAG21