nx_gateverify.nx

nx_gateverify.nx -- BEHAVIOURAL gate-record verification: run a gate, then compare what it PRINTED against what it RECORDED. WHY THIS AND NOT THE TEXTUAL CHECK: an earlier sibling organ tried to judge emit ORDER from source position and was wrong on all seven gates it flagged -- helper definitions, test fixtures and /tmp writes all match an append token, and it even flagged its own gate via test-data literals. That check was DELETED. The one genuine defect in that arc was caught by READING THE LOG AGAINST THE OUTPUT: the gate printed 41/41 while its log said 23/23. This organ mechanises exactly that comparison, so it finds the real thing and cannot manufacture the false ones. ★A gate's stdout is watched by a human for ten seconds; its LOG is what every ruler reads forever. When those two disagree, the log is the one that matters and the one that is wrong. nx_gateverify <gate-name> [logname] e.g. nx_gateverify nx_capgraph_derive_gate verdicts: AGREE · DISAGREE (the record is untrustworthy) · NO-LOG · UNPARSED exit 0 agree/no-log · 1 DISAGREE · 4 run-fail license_tier: ORIGINAL No hw writes (Rule 26). expect_exit: 0