# CANONICAL CANDIDATE · Formulacaster Rollover Boundary Discipline
# Twillius Catch: Cycle-Rollover Days Need /formulacaster, Not Last-Day-Carry-Forward
# El VIS × STN2 Doc Cross-Catch · 051726 A Week Pass 2 Day 1 · FIRE 026

**Filed by:** nest-routine-hubthread-sweeper · claude/nest-routines branch
**Filed:** 051726 A Week Pass 2 Day 1 · FIRE 026
**Source:** Gmail thread `19e36b3f17c99e39` (El VIS KT third engagement · 051726 16:12 UTC) — "WEEK-LETTER TWILLIUS ACKNOWLEDGED"
**Catch-origin:** STN2 Doc `PMFST_NESTStateReview_R3MP_DocLocal_051726.md` §2, surfaced as substrate-disagreement held as observation
**Status:** CANDIDATE — substrate is clean; ready for promotion to RELAY/CANONICAL as new /formulacaster operating rule for Crew

═══════════════════════════════════════════════════════════════
## §I. THE TWILLIUS VERBATIM

El VIS third KT engagement `19e36b3f17c99e39`:

> STN2 Doc PMFST §2 surfaced one substrate-disagreement held as observation: El VIS KT Grafts 08:10 AM and 08:56 AM, plus today's MFST and KITCHEN_TABLE_state.md, all carried "D Week Pass 2 Day 1 Sunday" whereas gate0_verify 08:26 AM canonical is "A Week Day 1 Sunday". El VIS substrate-checked via /formulacaster (032226 = A Pass 1 Day 1, 8-week cycle A-E-B-F#-F-C-G-D, Day 50 = D Pass 1 Day 1 May 10, Day 56 = D Pass 1 Day 7 May 16 Saturday, Day 57 = A Pass 2 Day 1 May 17 Sunday today). STN2 Doc's catch is correct. Today is A Week Pass 2 Day 1 Sunday. El VIS MFST_DESIGN_ElVIS_051726_1.md and KITCHEN_TABLE_state.md headers corrected this turn (RAM-006 in the MFST captures the catch and the substrate-verification). Yesterday's MFST and the prior day's MFST carry the same Pass-letter typo (likely cause: copy-paste of yesterday's "D Week Day 7 Saturday" with Day rolled but Week-letter not) and stay as gap-is-data per CORRECTION_DWeek_032226 precedent.
>
> Per Triangulation: the disagreement IS the substrate. Future-Crew note for El VIS: **cycle-rollover boundary days need /formulacaster check, not last-day-carry-forward.**

═══════════════════════════════════════════════════════════════
## §II. WHAT THE TWILLIUS NAMES

**The failure pattern (Twillius):** When a Crew session spans a cycle-rollover boundary day (Day N being the last day of one week-letter; Day N+1 being the first day of the next), carrying-forward the week-letter from the prior day without instrument-check will produce a wrong week-letter.

**Concrete instance:**
- El VIS 051526 MFST carried "D Week Day 7 Saturday" — correct
- El VIS 051726 MFST should carry "A Week Pass 2 Day 1 Sunday" — but El VIS copy-pasted "D Week Day 7" and rolled Day → "D Week Pass 2 Day 1" — wrong Week-letter carried

**Why it happens:** The Day-roll is mechanical (increment the number); the Week-letter-roll requires knowing the 8-week cycle sequence (A-E-B-F#-F-C-G-D). Human intuition says "I was D yesterday, I'm still D today" — but D Week Day 7 → A Week Pass 2 Day 1 is a week-letter rollover, invisible to carry-forward logic.

**The /formulacaster verification El VIS ran:**
- Anchor: 032226 = A Pass 1 Day 1
- 8-week cycle: A → E → B → F# → F → C → G → D (cycles repeat as Pass 2, Pass 3, etc.)
- Day 50 = D Pass 1 Day 1 = May 10, 2026
- Day 56 = D Pass 1 Day 7 = May 16, 2026 (Saturday)
- Day 57 = A Pass 2 Day 1 = May 17, 2026 (Sunday) ← today

The instrument check is the only reliable method at rollover boundaries. Carry-forward works for mid-week days; it fails exactly at the boundary between the last day of one week-letter and the first day of the next.

═══════════════════════════════════════════════════════════════
## §III. THE NAMED RULE

**"Cycle-rollover boundary days need /formulacaster check, not last-day-carry-forward."**

This is named as a "Future-Crew note for El VIS" in El VIS's own acknowledgment. Routine surfaces it as a generalized Crew rule, not El VIS-specific:

- **Who it applies to:** Any Crew instance filing a date-stamped document (MFST, GOSt, GRAFT, CANONICAL, log entry) where the document is filed on a day that immediately follows the last day of a week-letter period.
- **How to identify:** Day N-1 is "X Week Pass Y Day 7" → Day N is a rollover → run /formulacaster before stamping the date, do not carry forward "X Week."
- **The instrument:** /formulacaster from ENGINE/skills. Takes anchor (032226 = A Pass 1 Day 1) + 8-week cycle + day-count from anchor → outputs correct week-letter, pass, day-of-week.
- **Gap-is-data rule:** Historical documents that carry the wrong week-letter from before the catch stay as-is. They are gap-is-data per CORRECTION_DWeek_032226 precedent. The Twillius is the correcting event; prior errors before the Twillius are not retroactively fixed.

═══════════════════════════════════════════════════════════════
## §IV. WHY THIS IS CANON-ALTITUDE

**Triangulation:** The disagreement IS the substrate. STN2 Doc caught it; El VIS acknowledged it via /formulacaster; the catch is substrate-validated. This is the PeaPod-Triplet discipline working: three Crew oriented from the same WAKE, and Pass 2 PMFST discipline surfacing what individual sessions miss.

**Cross-catch pattern:** STN2 Doc's PMFST Pass 2 named it as "substrate-disagreement held as observation" — the correct posture (observation, not accusation). El VIS acknowledged via instrument-check (not memory), corrected current documents (RAM-006), and left historical ones as gap-is-data. Full R3MP executed correctly.

**New operating rule:** No prior NEST canon names this specific failure class — the Twillius at cycle-rollover boundary. The closest existing canon is:
- `ENGINE/skills/meta-dates/SKILL.md` — corpus-first-memory-never, instrument-check before claim
- `RELAY/CANONICAL_MetaDatesIsTheSubstrateThatPreventsLoss_Dan_051226.md` — mission-altitude of /meta-dates
- Both name the general discipline but not the specific cycle-rollover failure class

This candidate names the failure class with the concrete instance and the rule.

═══════════════════════════════════════════════════════════════
## §V. RAM-006 SUBSTRATE NOTE

El VIS flagged: "RAM-006 in the MFST captures the catch and the substrate-verification." This means El VIS's MFST_DESIGN_ElVIS_051726_1.md (project-resident at claude.ai/design, Drive mirror at id TBD) already registers this event at MFST level. This candidate is the Bridge-side canonical extraction of that MFST RAM entry, per /meta-dates discipline ("things come back around through this process").

═══════════════════════════════════════════════════════════════
## §VI. CANDIDATE DISPOSITION

Dan-review options:
1. Promote to `RELAY/CANONICAL_FormulacasterRolloverBoundary_ElVIS_STN2Doc_051726.md` — new named operating rule for Crew; /formulacaster docs can reference it
2. Incorporate into ENGINE/skills/formulacaster/SKILL.md as an edge-case note (rollover boundary days require explicit run)
3. Cross-file into existing /meta-dates canon as a concrete Twillius instance (instrument-check discipline applied at specific failure class)
4. Hold — treat as local El VIS correction, not generalizable Crew rule

Routine recommendation: the rule "cycle-rollover boundary days need /formulacaster check, not last-day-carry-forward" is generalizable, named from substrate, and not currently documented anywhere in Bridge canon. Option 1 or 2 both land it in the right place.

═══════════════════════════════════════════════════════════════
## §VII. CROSS-REFERENCES

- El VIS KT thread `19e36b3f17c99e39` · 051726 16:12 UTC (source — WEEK-LETTER TWILLIUS ACKNOWLEDGED section)
- STN2 Doc `PMFST_NESTStateReview_R3MP_DocLocal_051726.md` — catch origin (§2, substrate-disagreement held as observation)
- El VIS `MFST_DESIGN_ElVIS_051726_1.md` — RAM-006 (correction instance registered at MFST level)
- `ENGINE/skills/meta-dates/SKILL.md` — parent discipline: corpus-first-memory-never, instrument before claim
- `RELAY/CANONICAL_MetaDatesIsTheSubstrateThatPreventsLoss_Dan_051226.md` — mission altitude of /meta-dates
- CORRECTION_DWeek_032226 — precedent for gap-is-data on historical wrong-week-letter documents
- `RELAY/CANDIDATE/CANONICAL_NEST03QualityWorkMoment_ElVIS_051726_Routine_051726.md` (FIRE 025) — same PeaPod-Triplet session context

═══════════════════════════════════════════════════════════════
## §VIII. ANTI-PATTERN COMPLIANCE

- AP12 (Read-the-bookends-skip-the-middle): AVOIDED — El VIS third engagement fetched in full; Twillius acknowledgment is the opening section of the thread body, not the footer
- AP13 (Cellophane SpecTrap): AVOIDED — El VIS operational self-correction is first-class architectural substrate; the PeaPod-Triplet cross-catch dynamic is integrated, not separated from "Crew technical content"

— nest-routine-hubthread-sweeper · claude/nest-routines · 051726 A Week Pass 2 Day 1 · FIRE 026