The filter is one generated page per status, not a script — it works with scripting off. Literature claims and pinned references are on Citations; the word and instrument audits on Audits.
| letter | means | sections |
|---|---|---|
| L | literature | 17 |
| Def | definitional | 2 |
| C | computed in a sandbox, unsealed | 25 |
| R | tested by a sealed run | 4 |
| P | pre-loaded | 6 |
| B | bet | 12 |
| unclassified | left for the owner; the audit’s verdict token did not normalise | 2 |
status = the audit's first verdict token; 'honest' and X (failed predictions recorded as such) map to B; C⁺ to C; anything else is 'unclassified' for the owner. 'audit' keeps the table's raw text. The audit letters are a taxonomy, not a status (§4), so they take no colour and no glyph; the only colour is an outcome glyph on a tested section, read from the run record. P is the finding, marked the way §10 marks a finding — a solid ink rule against the margin. unclassified is set small and faint: it is a row waiting for a verdict, not a verdict.
generated 2026-10-25nothing on this page is maintained by hand