nx_gatemap_gate.nx

nx_gatemap_gate.nx -- WHICH DOMAIN DOES EACH UNWIRED GATE BELONG TO? Proposes, never writes. ★THE FINDING THIS SERVES (id=1785448222): 214 `_gate.elf` binaries exist on this host and only 19 are DECLARED in any knowledge/compare/<domain>.gates. Thirty of forty domains read CLAIM-ONLY not because nobody built the evidence but because nobody WIRED it. 0/40 PROVEN understates what is provable today by roughly an order of magnitude. ★THE RULE IS STRUCTURAL, NOT A GUESS: **a gate belongs to the domain whose MATRIX NAMES ITS STEM.** A `<domain>.matrix` row carries the organ path it claims (`runtime/nx_natstat.nx`, `runtime/_hdl_build/ nx_vcodec_t8noise_gate.nx`). So for gate `nx_X_gate` the question is simply: which matrix mentions `nx_X`? That is a fact about the corpus, not a similarity score, and it CANNOT drift the way keyword matching does. A gate no matrix mentions is REFUSED, not assigned to its nearest neighbour. ★WHY REFUSAL MATTERS AS MUCH AS THE MATCH: a wrong domain assignment is the same slander class as a wrong subject -- it makes a working gate count against a domain it was never about. UNMATCHED is the honest answer and it is cheap to read. ★AMBIGUITY IS REPORTED, NEVER BROKEN BY TASTE. If two matrices name the same stem the gate is listed as AMBIGUOUS with both domains, because picking one silently is how a wiring pass launders a judgement call into a fact. It WRITES NOTHING. Output is a proposal for human review (that is the whole point -- the operator asked to see it before it lands). usage: nx_gatemap_gate (CWD = the store root) exit 0 always -- this is a census, not a pass/fail gate; the verdict line reports the counts. license_tier: ORIGINAL expect_exit: 0