nx_wirecensus.nx

nx_wirecensus.nx -- THE NISHI KNIP, RUNG 1 (S4: registered but never called). nx_wirecensus [allowlist] [cap_consent.log] [actlog.jrnl] defaults: tool_allowlist.conf cap_consent.log knowledge/status/actlog.jrnl WHY ONE ORGAN AND NOT FOUR (July-2026 SOTA, id=1785451391): Knip finds unused files, exports, dependencies and devDependencies in ONE run over a whole monorepo -- ~300k weekly downloads, and Vercel deleted ~300k lines with it. Our equivalent is FRAGMENTED across nx_adopt (functions), nx_dupfunc (duplicate bodies), nx_srcdiverge (trees) and nx_toolreg_reconcile (registry): four organs, four invocations, no single verdict -- and a census nobody runs in one look is a census nobody runs. This organ is the consolidation point. Rung 1 is S4 because it is the class our STATIC censuses structurally cannot see, and because the field's biggest win (16k lines, 35 pct of an Express codebase) came from exactly this cross-reference: live evidence x declared surface. THE SIX CLASSES, so the shape is on the record even where unimplemented: S1 defined, zero EXTERNAL callers ............ nx_adopt answers this today S2 in SOURCE but in no shipped BINARY ........ open S3 built+promoted but never REGISTERED ....... open S4 registered but never AUTHORISED or CALLED . THIS ORGAN S5 conf row present, no consumer re-read it .. nx_toolreg_reconcile now upserts on drift S6 on the NAS but absent from the SSOT tree .. nx_srcdiverge_gate exists, stranded REFUSES RATHER THAN BLESSES: no allowlist, or zero registered tools, exits 2. "0 dark" computed from an empty declared surface reads as a clean bill of health while guaranteeing nothing. license_tier: ORIGINAL Read-only. No hw writes (Rule 26). expect_exit: 0