nx_planecap_gate.nx

nx_planecap_gate.nx -- WHICH PLANES HAVE OUTGROWN THE 1 MiB READER CAP? A verifier, not a fixer. ★WHY THIS EXISTS. Three organs in one session presented a TRUNCATED SCAN AS THEIR VERDICT because a compile-time cap was outgrown by the corpus: nx_debt_hygiene (1 MiB -> hid 21 of 23 sev-9 debts), nx_dbthyg (same bug in the superset organ), nx_tooldiff (512 rows -> INVENTED a capability regression). A grep finds **71** organs carrying `_CAP: i64 = 1048576`. Bulk-editing 71 files blind would be reckless and most of those caps are legitimately bounded (a per-page or per-document scratch buffer is not a hazard). ★THE ACTIONABLE QUESTION IS NOT "who has a cap" BUT "whose CORPUS HAS PASSED IT" -- so this measures the planes and names the ones that are live hazards TODAY. ★IT DOES NOT EDIT ANYTHING. A tool that auto-raised 71 caps would be a mass rewrite justified by a pattern rather than by measurement, which is the same reasoning error in the other direction. This produces the list; a human raises the caps that matter, with the plane size as the evidence. METHOD: every `<prefix>-manifest.txt` under knowledge/store/ names a live plane. Load each with a 32 MiB cap and report its true byte size. Anything at or above 1 MiB is flagged: any reader still sized at 1048576 will silently truncate ITS verdict on that plane. usage: nx_planecap_gate (CWD = the store root) exit 0 = no plane has outgrown the 1 MiB reader cap · 1 = at least one has (list printed) license_tier: ORIGINAL expect_exit: 0