nx_gamebench.nx

nx_gamebench.nx -- THE GAME CAPABILITY BENCHMARK: can the ecosystem EMIT the games we ingested? RULER = the REAL ingested open-source game corpus (knowledge/store/nx_game_catalog.reg, parsed by nx_game_catalog from the banked bobeff catalog) -- NOT a ruler we invented. Each reference TITLE is a real git repo we can read; each declares the CAPABILITY VECTOR it actually needs. We then score OUR organs per capability and COMPUTE (never assert) how much of each title we could emit from the first byte up. Follows the ecosystem_benchmarks.tsv doctrine (name a real external artifact + its bar) and the nx_benchmark_suite_registry doctrine (self-graded = SUSPECT until an external reference is WIRED). HONEST BY CONSTRUCTION: - coverage is COMPUTED from the rows (sum of capability scores / max), not a hand-written number; - a title is WIRED only if we actually read/ran its reference implementation. UNWIRED titles are reported as SELF-GRADED so a high readiness cannot be quoted as parity; - the GAP QUEUE is demand-ranked: a missing capability that N titles need outranks one that 1 needs. That queue IS the build order -- "where are our gaps" answered with arithmetic, not opinion; - every HAVE names an evidence artifact; nx_gamebench_gate liar-kills any that does not exist on disk. Self-contained (imports only nx_syscalls) so it runs on the NAS as an MCP tool. license_tier: ORIGINAL expect_exit: 0