kerfmaster pin list · UI / workflow decisions
recordStates what the code does today. Auditable against the code — this is the only type that can go stale.
Content last changed 2026-09-04 — computed from the item itself, not typed.
Built 2026-09-04. The 2026-08-24 ruling ('The large limit for slug tab will replace that rule') finally wired end to end: tables.rev2_tooling returns slug = {width, small_limit, large_limit} off the row (blank raises MaterialIncomplete -- a row that cannot size slug tabs never falls back to a blanket width); emit.write_gcode_macrob resolves it whenever tooling= is passed without an explicit slug_table (explicit legacy callers still win; tooling=None keeps the old path byte-identical). Three bands: no tab below the small limit (falls clear), the row's ONE width between the limits, no tab above the large limit (too big to tip) -- 0/0 rows (≥3/8) never tab, exactly as ruled.
The check judges with the same row -- sizing and judgment cannot disagree: rules.check_tabs(bnds, drop, large) takes the fall-clear size from the row's small limit (else the legacy machine drop-through) and the new upper band; a tab above the large limit is a needless-tab yellow in its own words ('too big to tip'), an untabbed above-large slug is never a traffic red (it cannot tip into the head -- the ruling's own rationale), and the traverse hazard set applies the same cap. check_drag, nc_audit.audit and the Cut plan screen all source the band from the tooling row; the slug report header names the rule the post actually sized with (stats.slug_rule -- header and rows can never disagree); the per-contour record's tab source reads table:rev2_slug.
Proven: test_rules 11→14 (above-large needless, no traffic red above large, legacy red stands) + test_drag 17→23 (sub-limit slug takes NO tab and no flag; mid-band takes the row's 0.020 -- not the old blanket 0.018; explicit slug_table still wins); all 20 suites green. Read-only replay of nav209's own review (Jordan's screenshot part): 0 findings, all five holes untabbed, source table:rev2_slug. Jobs re-judge on their next re-plan; delivered programs stand, per the no-redelivery precedent.
Part retention tabs unchanged: a typed tab size on the outer is still the operator's move (brick 19); automatic part tabs still await Jordan's width table (PINLIST 3a).