nx_charsim_bench_census.nx

nx_charsim_bench_census.nx -- the STANDING character-sim FEATURE BENCHMARK (operator 2026-07-03: "make sure you have them able to state of the art benchmark against programs on capabilities and breadth and depth of features ... like the virtamate and other systems across asia and americas" + "make sure our charlab is above honey select"). A BREADTH x DEPTH matrix: every row = one capability category, graded DEPTH 0..3 for US (organ+gate evidence ONLY) and for the BEST NAMED INCUMBENT (VaM+TittyMagic / TheKlub [Americas]; Honey Select 2 / AI Shoujo / Custom Maid 3D 2 / BetterRepack [Asia/Illusion lineage]) -- incumbent grades sourced from the banked hsx_*/vamx_* corpus + the operator's axis summary (STATED source, no fabricated specifics). DEPTH legend: 0 absent · 1 v1/basic · 2 solid · 3 SOTA-leading. Re-run after every rung: rows self-flip on evidence; the tallies are the standing scoreboard. LIAR-KILL: the fabricated row must grade us 0. license_tier: ORIGINAL expect_exit: 0