SLICE-register-variety_DRAFT.md

pipelines/SLICE-register-variety_DRAFT.md

SLICE-register-variety — the REGISTER-VARIETY slice of L-R5 (underworld-grief-and-register-variety)

Realm Design Program, Wave 1 doctrine wave. Read-only audit, 2026-07-28.

Unit id: SLICE-register-variety. Aspects audited: R-A6 (afterlife/underworld emotional register),

R-A12 (otherworld inhabitants and society), R-A18 (the homecoming and return register), plus the

doctrine question of F-5 in docs/pipeline_review/REALM_ANALYSIS_LENS_GAPS_2026-07-27.md.

Titles: Spiritfarer · Grim Fandango · What Remains of Edith Finch · Hades (register layer only,

handed over from hades-death-run-realm-identity per REALM_DESIGN_PROGRAM §4.3).

Charter authority for this unit: docs/proposals/REALM_DESIGN_PROGRAM.md §4.2 —

*"the REGISTER-VARIETY slice of L-R5 (what makes Hades warm, Xibalba adversarial, Annwn

seductive-and-hospitable, Naraka moral-architectural, the Duat exact and grave — a doctrine question

that must be answered BEFORE the per-realm charters, not after, or five consecutive death realms

converge into the named monotone-solemn slop mode)"* — and §5.2, which places this slice in Wave 1

against the schema freeze.

Consumer contract this document is written to satisfy, verbatim from

docs/pipeline_review/REALMARC_PATTERN_BRIEF_2026-07-27.md §5 (field mints):

*"The register ENUM VALUES (L-R5's register-variety slice delivers the closed set as its named

output, exactly as L-R3 delivers the care tiering) and the per-node population."*

Restated at that brief's §7 LENS INTERFACES: *"L-R5's register-variety slice owns Clause 3's closed

enum as its named output."* Section 1 below IS that named output. Sections 2-3 are the audit that

produced it and the schema handoff.

---

0. Method, and the positive controls run before any zero in this document

Every absence claim below was run against a term known to exist, in the same tool over the same

corpus, per the standing wave rule.

• Registry-column search. A Python pass over all registries/**/*.csv headers for

register|tone|mood|emotion|affect. POSITIVE CONTROL: the same pass counted chapter_id headers

and returned 5 (non-zero), so the matcher works. Result: 7 columns exist —

T0_Chapter_Index.music_mood_tags, T0_Theme_Registry.mood_tags, voice_register_ref

(Ally_Behavior_Policy, Bonus_Option_Registry), cultural_care_register (Creature_Roster,

Familiar_Registry), care_register (Hollow_Codex). **No node-grain emotional-register column

exists in live canon.** Corroborated independently by REALMARC_PATTERN_BRIEF §8:

*"register_dominant confirmed ABSENT from the live T0_Chapter_Index header."*

• RCM section search. My first search for ## Ch 70 in

_source/01_Tier_1_Foundation/T1_Research_Corpus_Master [ACTIVE v1.1.4].md returned ZERO for

every realm chapter. That zero was FALSE — the headings are ## Ch 70 — Hades and Elysium

(bold markers plus an em-dash). Recorded because it is this repo's most-repeated defect class and

I hit it inside this audit. The corrected search returned 33 headings. Every RCM claim below runs

off the corrected search.

• Gate-file search. ls harness/ for topog|emotional|realm|reveal. POSITIVE CONTROL:

check_chapter_topography.py and check_reveal_discipline.py both returned. Result:

check_emotional_curve.py and check_realm_ruleset.py do not exist. Both are DESIGNED-UNBUILT,

not MISSING.

• Register-vocabulary arithmetic. A fixed-string count of grave|elegiac|warm|luminous over the

three art/audio/biome bullets (- biome_primary, - material_palette and style_register,

- music_mood) of docs/spine/CH_66.md .. CH_77.md. POSITIVE CONTROL: the same pass counted

- material_palette lines and found 12 of 12, so the line selector reaches every node.

• Designed-not-missing sweep. Before any MISSING verdict I searched

docs/proposals/REALM_DESIGN_PROGRAM.md, docs/proposals/STORY_DEPTH_PROGRAM.md,

docs/pipeline_review/REALMARC_PATTERN_BRIEF_2026-07-27.md, docs/PRE_5090_BUILD_PLAN.md,

docs/FACTORY_CONTRACT.md and docs/DESIGN_GAP_REGISTER.md. Two of my three candidate MISSINGs

were downgraded on that sweep. Only two MISSINGs survive.

Bound on everything below. My probe ran over Ch 66-77 plus the Epilogue — thirteen nodes. The

enum's adequacy over Ch 2-65 is UNPROBED and is stated as unprobed in F-R19; the enum is not

ratifiable until that probe runs. This is the same bound the charter states on the physics

vocabulary V at §2.3, applied honestly to my own output rather than inherited silently.

---

1. THE NAMED OUTPUT — the register doctrine, the closed enum, and the per-realm assignment

1.1 The doctrine, in one paragraph

A realm's register is not its subject matter. Five realms can all be about death and share nothing

of how they feel, because register is carried by four separable things: **who is in the realm and

whether they speak to you (Spiritfarer); what the realm wants from you** (Grim Fandango's

underworld wants your paperwork, not your soul); **what the realm's own tradition thought death was

FOR (Edith Finch's per-vignette rules); and what the realm is forbidden to feel** (the negation

clause every one of our road chapters already authors and nothing reads). The doctrine is therefore:

**a realm declares one dominant register, one counter-register that keeps it off one note, and a

named forbidden register that is its own slop mode — all three derived from that tradition's actual

texture and cited to it, and the (dominant, counter) PAIR must be unique inside any five-node

window.** Register variety is not achieved by making realms different; it is achieved by making the

DIFFERENCE machine-visible and then refusing to author against it.

1.2 The closed enum — the seeded eight, retained verbatim, plus five probe-forced widenings

REUSE-FIRST. The enum is NOT minted here. docs/proposals/STORY_DEPTH_PROGRAM.md §Wave 4 and

docs/pipeline_review/STORYDEPTH_STORY_CRAFT_LENS_FAMILY_2026-07-27.md L433 already declare

T0_Chapter_Index.emotional_register as *"a closed enum, seeded at eight and expandable by ruling

(examples-not-limits): wonder · grief · dread · fury · tenderness · wry · awe · resolve."* This

slice's job is the widening and the population, not a new vocabulary. All eight seeded values are

retained verbatim and none is renamed.

The probe: assign every one of the thirteen endgame nodes a seeded value from its AUTHORED register

prose. Five nodes cannot be assigned without collapsing a distinction the built prose deliberately

makes. Those five collisions are the residue, and the residue is the widening.

Enum v0.1 = 13 values: `wonder · grief · dread · fury · tenderness · wry · awe · resolve ·

elegiac · exacting · adversarial · hospitable · ascendant`.

1.3 THE PER-REALM REGISTER ASSIGNMENT (proposal-tier; the deliverable)

register_forbidden carries only REGISTER negations. Care negations (occult-Aryan,

exoticizing kitsch, mystified real-archaeology, Christianized Hell) are NOT duplicated here —

they route to the charter's existing T-RLM-CARE fixed-string firewall (§2.7). Keeping the two

apart is deliberate: one is a craft rule, the other a hard-line rule, and merging them would let a

craft waiver read as a care waiver.

The five-death-realm run answered. Nodes four through eight of the road — the Duat, Hades,

Xibalba, Naraka/Meru, Annwn — read exacting · elegiac · adversarial · ascendant · hospitable.

Five consecutive nodes, five distinct dominants, five distinct ordered pairs, every one cited to

that tradition's own authored texture rather than assigned for contrast. **The monotone-solemn slop

mode does not have to be designed out; it has to be declared out, because the built prose already

avoided it.** What the road does NOT yet have is any artifact that can see that, which is the whole

of Section 2.

Rotation check against both live proposed rules. The string across Ch 66-Epilogue reads

`awe · dread · wonder · exacting · elegiac · adversarial · ascendant · hospitable · awe · fury ·

resolve · wonder · tenderness`. No two consecutive nodes share (satisfies REALMARC Clause 3 / R7).

No three consecutive share (satisfies STORY_DEPTH E1). Every five-node window carries five distinct

values, well clear of E2's three-distinct floor. The assignment is therefore valid under whichever

of the two colliding rules survives F-R01.

1.4 The four register-carriers, per realm — what actually delivers the tag in play

The tag is not the register. These are the craft moves each title teaches, mapped onto the node that

already half-does them, so the Wave-2 charters inherit a move and not an adjective.

• Ch 69 the Duat — Grim Fandango's lesson: a death realm can be PROCEDURAL. The Hall of Two

Truths is our bureaucracy-of-death, and the register survives if the procedure is honoured as a

procedure — a form filled correctly, a record admitted, an argument met on its own terms — rather

than dressed as horror. The existing DECLARE verb (SCH_013) and the audit-only assessor rule

(HL_0003) already produce this; the register tag just names it so a concept board cannot drift it

toward mummy-horror.

• Ch 70 Hades — Spiritfarer's lesson, and the one structural limit worth stating. Spiritfarer's

warmth comes from PERSONS you return to across many visits and eventually release. Our road is

ONE-PASS: the traveller crosses each realm once. **The Hades-the-game warmth mechanism is

structurally unavailable to us and should not be chased.** The honest substitute is the

found-family: warmth carried by the people who came WITH her reacting to the country of the dead.

The road already contains exactly one instance of this and it is the best register beat on the

whole road (Ch 71's exit: *"the whole long road up is her people making her explain, trick by

trick, how any of it was done"* — docs/spine/CH_71.md L26 block). See F-R10.

• Ch 71 Xibalba — the register is the OPPONENT'S personality. Already fully authored: the Lords

cheat, the house is rigged, and the L26 closing match *"with the rules straight they are

magnificent — it turns out the house never cheated because it had to"* is the single strongest

register instance in the endgame. Nothing to add; protect it.

• Ch 72 Naraka — the register is the ARCHITECTURE. The realm's feeling is its own moral

cosmography read as level design (charter §2.5's THE ENCLOSING). The gap is people: see F-R09.

• **Ch 73 Annwn — Edith Finch's lesson: the anthology needs one node with completely different

rules.** Annwn is the road's authored break and its music bullet says so in as many words:

*"the relief after the dead-realms"* (L196). It is the only node whose warmth is a mechanic (guest

obligation, the borrowed crown, the withheld blow) rather than a lighting choice.

• Ch 75 the fairy realm — the grief-geography encounter is the road's Spiritfarer beat. Grief

crystallised into terrain, addressed by restraint and consent rather than damage

(docs/spine/CH_75.md L26: *"that is not a corruption to purge, it is love in a terrible shape …

over-purging here — scouring away the shapes his missing her built — would be the worst thing her

power could do"*). This is already authored to Spiritfarer's standard. Its problem is placement,

not craft: see F-R17.

---

2. FINDINGS

Twenty-one findings. Each carries verdict class, owner lane, wave and priority.

Counts: COVERED 2 · DESIGNED-UNBUILT 5 · WEAKER 11 · MISSING 2 · DIVERGENT-BY-DESIGN 1.

F-R01 — The register column is designed TWICE, under two names, with two incompatible rotation rules

Verdict: DESIGNED-UNBUILT (CONTESTED — director call). Owner: schema-data. Wave 2. Priority:

HIGHEST — this is freeze-blocking and it blocks my own named output from having one home.

Two live proposal-tier artifacts, both dated 2026-07-27, both propose the same column:

• emotional_register — docs/proposals/STORY_DEPTH_PROGRAM.md Wave 4 and

docs/pipeline_review/STORYDEPTH_STORY_CRAFT_LENS_FAMILY_2026-07-27.md L433. Closed enum seeded

at eight. Gate home harness/check_emotional_curve.py, gate #33, rules E1-E6. **E1: no THREE

consecutive nodes share.** Scheduled with two sibling columns (emotional_valence_arc,

cost_ledger) and already carried into the window plan as a freeze rule:

docs/PRE_5090_BUILD_PLAN.md L120 and docs/FACTORY_CONTRACT.md L40 both read

*"T0_Chapter_Index.emotional_register lands BEFORE rank 8 — rank 8's music cue table has no

upstream mood source today."*

• register_dominant — docs/pipeline_review/REALMARC_PATTERN_BRIEF_2026-07-27.md Clause 3 and §5.1

NEW R7. Closed enum, 79 rows. Gate home harness/check_chapter_topography.py as R7. **Clause 3:

no TWO consecutive nodes share, plus every five-node window carries ≥3 distinct.**

Same column, two names, two gate files, two strictness levels, and **both name this slice as the

owner of the enum values.** If both mint, T0_Chapter_Index acquires two mood columns whose values

must be kept in sync by hand, two teeth that can disagree, and a music_mood_tags derivation with

two upstream sources.

RECOMMENDATION (reuse-first). Mint ONE column, named emotional_register, because that name is

already carried into the window plan and the factory contract as a scheduled freeze rule and is

therefore the more expensive name to change. Adopt REALMARC's stricter rotation rule (no two

consecutive) as E1 rather than STORY_DEPTH's no-three, on the grounds REALMARC itself gives — it is

HL_0049's exact shape re-pointed, so it costs no new rule machinery — and keep E2-E6 as authored.

Host it in check_emotional_curve.py (gate 33), not in the topography gate, because the topography

gate's declared scope is HL_0048/HL_0049 and adding a register rule there makes that gate's docstring

a lie about which hard line it enforces. Record register_dominant as a retired alias in the same

commit so a later reader of the REALMARC brief does not mint it a second time.

Strongest objection: the topography gate already has the sequence-window machinery and reusing it

is cheaper. Answer: check_emotional_curve.py is specced to be *"built to check_chapter_topography.py's

exact shape"* (STORYDEPTH L~447), so the machinery is copied either way; what differs is which gate's

scope statement stays true.

F-R02 — The seeded eight cannot express the road; the probe residue is five widenings

Verdict: WEAKER. Owner: schema-data. Wave 2. Priority: HIGH (it is the named output).

Detailed in §1.2. The short form: forcing the road onto eight values collapses the Duat and

Yggdrasil into one value, Hades and Ch 75 into one value, and leaves Annwn and Naraka with no value

that is not a lie. **The enum itself would have manufactured the monotone the doctrine exists to

prevent — in the data, while the prose stayed differentiated.** That is the strongest single

argument for running the widening BEFORE the mint rather than after.

F-R03 — The register differentiation lives in the BODY PROSE; the ART AND AUDIO bullets converge

Verdict: WEAKER. Owner: spec-doc (spine art bullets) + P5-factory (the concept pipeline).

Wave 2-3. Priority: HIGH — it is the finding with the most direct effect on the art-concept wave.

Fixed-string arithmetic over the three art/audio/biome bullets of docs/spine/CH_66.md .. CH_77.md

(positive control: 12 of 12 - material_palette lines found):

REALMARC_PATTERN_BRIEF §0.3 reads the BODY PROSE and concludes, correctly, that *"the charter's own

feared failure mode … is NOT what the built prose does."* That finding stands and I confirm it. **My

refinement changes what the mint has to do:** the differentiation is in the layer the art pipeline

does not read, and the layer the art pipeline DOES read — material_palette and style_register plus

music_mood, which are precisely the fields a concept-board prompt and an AIVA brief are built from

— is monotone grave + luminous across the entire endgame. A concept fan-out fed those bullets

will generate eight consecutive grave luminous realms and be textually faithful to canon while doing

it.

RECOMMENDATION (reuse-first, two moves, no new artifact). (1) Populate emotional_register from

the BODY PROSE and the biome_primary register clause, never from the art bullet — the art bullet is

the drifted layer. (2) Add a Wave-3 repair item: rewrite the leading adjective of the

material_palette and style_register and music_mood bullets at Ch 69-76 so each leads with its

declared register rather than with grave. This is a twelve-line edit to already-living source docs

under the living-source doctrine, it invents nothing, and it is the cheapest defence the realm art

wave will ever get. Gate it with the derived tooth in §3.4.

F-R04 — music_mood_tags reads 79/79 filled and 0/79 real; the sentinel form matters

Verdict: COVERED (with a precision correction to a live artifact). Owner: schema-data. Priority:

MEDIUM.

Verified directly: all 79 rows of `registries/T0_Chapter_Index [ACTIVE v1.2]/T0_Chapter_Index DRAFT

v1.1.csv carry music_mood_tags = [POPULATE→T8_Region_Pages Section 10 AIVA composition briefs]`.

STORY_DEPTH's E6 already handles this exactly right — *"Missing column, empty cell, or a

[POPULATE…] placeholder all fail"* — so this is COVERED and I am not re-litigating it.

The precision correction: REALMARC_PATTERN_BRIEF §0.3 states *"music_mood_tags on

T0_Chapter_Index is populated 0 of 79 rows,"* which an implementer will read as "the cells are

empty." The cells are NOT empty; they are non-empty sentinels. A tooth written to the REALMARC

sentence — if not cell: fail — passes 79 of 79 and reports green on a column that carries nothing.

The tooth must test the sentinel FORM. harness/check_chapter_topography.py already ships a

_unpopulated(v) helper (L69) that is the correct thing to reuse.

F-R05 — emotional_register is declared the upstream source of music_mood_tags, and the two vocabularies share ZERO terms

Verdict: WEAKER. Owner: schema-data. Wave 2. Priority: HIGH (it is a stated build unblock that

cannot execute as stated).

STORYDEPTH L444: *"emotional_register is the upstream source music_mood_tags never had."*

The live T0_Theme_Registry [DRAFT v0.1] carries 12 rows and 22 distinct mood_tags:

`transcendent(5) · antagonistic(5) · climactic(5) · contemplative(3) · intimate(3) · ominous(3) ·

ceremonial(3) · militant(3) · foreboding(2) · tragic(2) · heroic · intellectual · manipulative ·

escalating · triumphant · redemptive · bittersweet · hopeful · mournful · ritualistic · warm ·

bonded`.

Intersection with the seeded eight (wonder grief dread fury tenderness wry awe resolve): empty.

Intersection with my 13: empty. So "derives from" has no mechanism today — a composer brief

generated from emotional_register: elegiac cannot resolve to any tag the theme registry uses.

RECOMMENDATION (reuse-first). The schema mint must ship a declared MAPPING TABLE

emotional_register → mood_tags[] inside the schema definition, in the same commit, or drop the

"upstream source" claim from the program docs. Suggested seed for the five widenings, drawn only

from tags the registry already uses: elegiac → [mournful, contemplative] · `exacting →

[ceremonial, intellectual] · adversarial → [antagonistic, escalating] · hospitable → [warm,

bonded] · ascendant → [transcendent, triumphant]`. A mapping that resolves is what makes rank 8's

music cue table buildable; a mapping that does not is the "column looks populated for months while

carrying nothing" failure a second time.

F-R06 — Ch 67 and Ch 68 carry the IDENTICAL authored register phrase on consecutive nodes

Verdict: WEAKER. Owner: spec-doc. Wave 3. Priority: MEDIUM.

docs/spine/CH_67.md L201: *"awe-and-dread, never a mystified real-archaeology or occult-polar

rendering."* docs/spine/CH_68.md L193: *"awe-and-dread, never a mystified psychic-ocean or

lost-continent rendering."* Two consecutive road nodes, the same two-word register, verbatim.

This is a genuine adjacent-pair collision in live authored canon and it is the **best available

must-fire fixture** for the rotation tooth — REALMARC's §5.4 asks for *"Clause 3 (R7) MUST-FIRE: two

consecutive nodes carrying the same register_dominant"* and proposes a synthetic. Use the real

Ch 67-68 pair instead, kept as the live fixture until the repair lands, exactly as that brief keeps

the real CH_71 as its Clause-1 fixture. The differentiator is already in the prose and needs no

invention: Ch 68's dread is PATIENT and pre-verbal and the boss is declined rather than slain

(*"the sea heard, not scored as a foe"*, L194), while Ch 67's is a defended contest. Hence my

assignment of wonder / dread at Ch 68 against dread / awe at Ch 67.

F-R07 — Node-grain register cannot express Ch 75 or Ch 76, both of which authored TWO registers explicitly

Verdict: WEAKER. Owner: schema-data. Wave 2. Priority: HIGH.

docs/spine/CH_76.md L212 authors both in one bullet: *"a grave, sustained confrontation register

for Phase 1 … a spare, aching reunion register for Phase 2."* docs/spine/CH_75.md L9 authors the

node as *"homecoming and war held together,"* and its L269 music bullet names three registers inside

one chapter (a homecoming register, a marquee war register, a terminal eschaton-dragon register).

A single-value node-grain column is arithmetically incapable of carrying either node, and forcing

one deletes the arc's two most important register CHANGES.

RECOMMENDATION (reuse-first, no new grain). Do not invent a movement grain. Instead: (1) make

register_secondary MANDATORY (non-empty) on any node whose spine entry authors more than one

register bullet, which is a derivable condition; and (2) hang the register override on the SITE rows

the REALMARC brief is already creating — its §5 field mints list *"the CH_75 grief_country site

split so the Movement-1 low-threat grief country stops sharing a row with the war sites."* That split

is already funded; adding a site_register field to the two resulting rows costs one column on an

existing pass and gives Ch 75 its grief → fury change for free. Ch 76's two phases are already

separate beat ranges (P1a-P1c versus P2a-P2b), so the same field on its scene-anchor rows carries it.

F-R08 — R-A12: the realms of the road are staged UNPEOPLED

Verdict: WEAKER. Owner: canon-author (P2 factory charters). Wave 2-3. Priority: HIGH — this is

the cheapest register lever in the entire program and it is the one Spiritfarer exists to teach.

Beat-mode census across the nine road nodes (parsed from · mode <x> in each docs/spine/CH_NN.md

beat graph):

Four dialogue beats across nine otherworlds. Three road nodes declare themselves unpeopled

outright. The register consequence is direct: a realm with no one in it can only express register

through lighting, architecture and music — which is exactly the three-channel layer F-R03 shows has

already converged. Register variety and realm population are the same problem.

RECOMMENDATION (reuse-first). Do not mint a populace system. Every road node already carries the

found-family (the healer's child CHAR_0002, Ana CHAR_0017, the familiars) as a stated presence in its

Continuity anchor and its Key-ally block. Require **one register-bearing companion exchange per

realm**, which is REALMARC's Clause 5 seam camp obligation (*"≥1 companion exchange, and the register

hand-off from departing to arriving node"*) pulled inside the realm rather than only onto the seam.

That is a beat, not a system, and it is the only register carrier available at Ch 66, 68, 69, 72 and

74 without inventing inhabitants for realms canon has ruled unpeopled.

F-R09 — Ch 72 Naraka, the road's highest-care realm, is its emptiest

Verdict: WEAKER. Owner: canon-author. Wave 2. Priority: HIGH (care-loaded).

Zero dialogue beats, one character beat, six beats of the ten being discovery/traversal/combat, and a

voice_and_crowd bullet that voices its deities only *"where present."* The charter itself flags

this node twice: §1.1 calls Naraka/Meru *"MAXIMUM"* care and records *"per the lens audit this is the

highest-care realm on the road that has had the LEAST attention."*

The register consequence is specific and worth stating plainly, because it is a care finding and not

only a craft one: the whole point of Naraka in its own tradition is that it is INHABITED — souls

serving proportionate terms, Yama weighing, Chitragupta keeping the book, and every one of them

temporary. Our staging renders a moral cosmography as an empty stairwell. Depopulating a living

tradition's cosmology is a care failure of the omission class that CVD §17.1's care doctrine names

by construction (care means weight and authenticity, never omission), and the fix is additive: the

weighing-place already exists as a site (docs/spine/CH_72.md L26 *"the weighing-place where a still

judge keeps the balance and a patient scribe keeps the book, and neither of them is her enemy and

neither of them is cruel"*) and hosts no dialogue beat. One dialogue beat at the weighing-place gives

the road's highest-care realm a speaking inhabitant, and it is already written in prose.

F-R10 — The Hades-the-game warmth mechanism is structurally unavailable on a one-pass road

Verdict: WEAKER. Owner: canon-author. Wave 2. Priority: MEDIUM-HIGH.

This is the finding that answers the charter's own phrasing. §4.2 asks *"what makes Hades warm."*

The answer in that game is REPETITION — you die, you come back, the same dead people are pleased to

see you, and the warmth accrues across dozens of returns. Our road is single-pass by construction

(each realm crossed once, region_chapter_set per node), so that mechanism cannot be ported and

should not be attempted. Attempting it would also collide with our own Ch-70 authored register, which

is elegiac and explicitly forbids whimsy (*"never horror-gore and never whimsy"*, L199) — the

Hades-the-game register would be a §17.1-adjacent drift toward the Homeric-elegiac care commitment

the chapter's own protected-culture block makes verbatim: *"the register throughout is

Homeric-elegiac (a mournful continuation), never horror-movie ghouls or a haunted house."*

**Therefore the charter's phrase "what makes Hades warm" is answered NO for our Ch 70 and the answer

is cited.** The warmth the road actually gets is the found-family's, and Ch 71 already ships the

exemplar. Reuse it as the pattern, not as an exception.

F-R11 — register_forbidden is fully authored across 12 of 12 endgame nodes and has ZERO consumers

Verdict: WEAKER. Owner: schema-data. Wave 2. Priority: HIGH — it is the highest ratio item in

this audit: the content is already written and needs only a column.

Every endgame node's material_palette and style_register bullet ends with a negation clause. All

twelve, verbatim-sourced: Ch 66 *"never a mystified real-archaeology rendering"* · Ch 67 *"never …

occult-polar"* · Ch 68 *"never a mystified psychic-ocean or lost-continent"* · Ch 69 *"never a

mummy-horror or occult-theme-park"* · Ch 70 *"never horror-gore and never whimsy"* · Ch 71 *"never

horror-gore and never exoticizing kitsch"* · Ch 72 *"never gore and never an exoticizing or

occult-Aryan"* · Ch 73 *"never exoticizing kitsch and never a horror-monster hunt"* · Ch 74 *"never

gore and never an occult-Aryan or völkisch"* · Ch 75 *"never whimsical … never caricatured"* ·

Ch 76 *"never caricatured … never whimsy nor spectacle"* · Ch 77 *"without spectacle."*

These split cleanly into two classes and the split is load-bearing:

• CARE negations (occult-Aryan, völkisch, exoticizing kitsch, mystified real-archaeology,

psychic-ocean, lost-continent, occult-polar, Christianized Hell) — already have a home:

the charter's T-RLM-CARE fixed-string firewall at §2.7. Do not duplicate.

• REGISTER negations (horror-gore, whimsy, spectacle, caricature, buffoonery,

theme-park, foe-scoring, hostile-menace) — have no home at all. These are the per-realm

named slop modes, authored by the people who know each realm best, and nothing reads them.

RECOMMENDATION. Mint register_forbidden as an array column alongside emotional_register,

populated from these clauses (a mechanical extraction, not an authoring pass). Two consumers make it

worth the column: the concept-board prompt gets a per-realm negative prompt derived from canon rather

than from a generator's taste, and the critic rubric gets its per-realm false-positive list, which is

what turns RB-REG-FRAME (F-R18) from a vibe check into a scored one.

F-R12 — The register HAND-OFF is already authored at chapter openings

Verdict: COVERED (in prose) / DESIGNED-UNBUILT (as data). Owner: schema-data. Priority: MEDIUM.

Three consecutive road chapters open by naming the register they are leaving:

• CH_71 L26: *"She has been down into a country of the dead once already on this road … the grey

patience, the elegiac hush. Xibalba teaches her, before she is an hour into it, that she knew

nothing."*

• CH_72 L26: *"She comes up out of the Maya dark grinning and finds that the next country wants her

to climb."*

• CH_73 L26: *"The road has taken her down into the dead and up out of them, and now, for the first

time in longer than she wants to count, it takes her somewhere warm."*

This is the strongest existing evidence that REALMARC's Clause-5 register_from / register_to

seam fields are describing something the authors already do rather than adding a burden. It also

supplies the Clause-5 fixtures for free. Nothing to author; declare it and populate.

F-R13 — The "five consecutive death realms" framing is imprecise; the real run is four

Verdict: DIVERGENT-BY-DESIGN (precision correction, not a gap). Owner: spec-doc. Priority: LOW,

but it narrows the doctrine's burden and should be recorded before Wave 2 sizes the work.

The charter's §4.2 and the lens-gap doc's F-5 both count nodes four through eight — the Duat, Hades,

Xibalba, Naraka, Annwn — as *"five consecutive death realms."* Annwn is the eighth, and the built

spine deliberately makes it the BREAK, in its own words: docs/spine/CH_73.md L196, *"a warm,

welcoming, honorable register (the feast-hall plenty, the relief after the dead-realms)."* Its

biome line reads *"an honor-tale not a war,"* its crossing-function is hospitality, and the

REALMARC brief independently records it as *"the ascent's ONLY warm register"* and the only node with

a low-threat hall hosting five of nine beats.

So the genuine consecutive-death run is four (Ch 69-72), and the fifth node is the authored

escape valve that the register doctrine's job is to PROTECT rather than to create. This matters for

Wave 2 sizing and it matters for the Pattern-B objection recorded in the REALMARC brief (*"Lifting it

deletes the strongest terrace on the road and collapses the register rotation at exactly the point

the road most needs contrast"*): Annwn is the load-bearing node for register variety and any later

pass that raises its intensity or grades its warmth down is spending the doctrine's own reserve.

F-R14 — The RCM substrate for ALL FIVE death-realm nodes is a declared SUBSTRATE GAP carrying displaced bodies

Verdict: DESIGNED-UNBUILT (Wave 3 repair, already ranked). Owner: canon-author. Priority: MEDIUM

for the program, NON-BLOCKING for this slice — which is the point of recording it.

Verified by direct read of _source/01_Tier_1_Foundation/T1_Research_Corpus_Master [ACTIVE v1.1.4].md

(after the corrected heading search — see §0). Each of Ch 70, 71, 72, 73 and 74 opens with a

SUBSTRATE GAP banner and a do-not-use instruction, e.g. Ch 71: *"The body below is the displaced

pre-v3 unit (Realm War Phase 1) and does not ground Xibalba — do not use it as this realm's

substrate."* Ch 73's is blunter still: *"its Cath-Maige-Tuired and Ragnarök battle material is battle

substrate, not the Welsh Otherworld."*

I also independently re-confirmed the charter's positive-controlled Annwn acquisition-list zero: the

### acquisition blocks run Ch 66 · 67 · 68 · 69 · 70 · 71 · 72 and then jump to Ch 74, with no

Ch 73 block, while the ## per-chapter substrate sections DO carry ## Ch 73 — Annwn. Same

matcher, both results, so the zero is real.

The consequence that matters for the schema designer: the register assignment in §1.3 is cited to

the SPINE, not to the RCM, and it has to be, because the RCM cannot ground it today. That is not a

weakness of the assignment — the spine entries carry their own build-time deep-research provenance

(CH_70's header names the Homeric, Hesiodic and Virgilian corpus plus Burkert, Graf and Johnston,

Bremmer, Mylonas and Kerényi; CH_73's names Sioned Davies, Will Parker, Marged Haycock, Mac Cana and

Koch). The register mint does NOT block on the RCM repair. Stating that explicitly prevents a

later wave from serialising a cheap data mint behind an expensive substrate rebuild.

F-R15 — Annwn's authored register is warm-and-honourable; the SEDUCTION half is proposal-tier only

Verdict: DESIGNED-UNBUILT. Owner: canon-author. Wave 2. Priority: MEDIUM.

The charter's §4.2 phrase for Annwn is *"seductive-and-hospitable."* The built spine delivers

hospitable in full and seductive not at all: docs/spine/CH_73.md L26 explicitly REFUSES the

seduction reading — *"what she finds under the warmth is not a lie waiting to spring; it is only an

old and courteous kind of plenty."* The peril is present but externalised (the Cwn Annwn coursing,

the glass-fort caers, the ford), not woven into the welcome.

The seduction mechanics DO exist at proposal tier in docs/proposals/REALM_DESIGN_PROGRAM.md §2.4

(*"Do not eat the otherworld's food in Annwn is a consumable that binds a time cost"*) and §2.5

(THE SINGLE BLOW bending time — *"the otherworld year, dilated and deliberately unmeasurable from

inside"*). The year-and-a-day IS lived at CH73_B05 as a story fact.

RECOMMENDATION. Keep hospitable as the declared dominant — it is what the built prose does, it

is cited, and it is the road's only warm node. Route the seduction to register_secondary only if

the Wave-2 charter actually builds the food taboo and the felt time-loss; otherwise Annwn's secondary

stays resolve (the withheld blow). Do NOT assign a register the build does not deliver: a declared

column that the prose contradicts is the exact failure RB-ASC-REGISTER exists to catch, and this

slice should not seed it.

F-R16 — The de-facto register field is authored inconsistently across the road

Verdict: WEAKER. Owner: spec-doc. Wave 3. Priority: MEDIUM (it is the population source).

An explicit *"the register X and Y"* clause sits in biome_primary at CH_70, CH_71, CH_72, CH_73,

CH_74 and CH_77. It is ABSENT at CH_66, CH_67, CH_68, CH_75 — and at CH_69 it appears in a different

grammatical form (*"a descent-through-gated-caverns-to-a-hall-of-judgment aesthetic, grave and

exact"*) that a parser keyed to the phrase would miss.

So the column already exists as an emergent authoring convention with about 50% adoption and no

schema. Two consequences: any extraction script must handle both forms plus absence (four nodes need

hand-assignment, supplied in §1.3), and the Wave-3 art-bullet repair in F-R03 should normalise the

clause into biome_primary at all thirteen nodes at the same time, so the prose and the column agree

by construction rather than by discipline.

F-R17 — R-A18: the homecoming register exists in three places with no doctrine connecting them, and Ch 75 over-writes its own

Verdict: WEAKER. Owner: canon-author + schema-data. Wave 2. Priority: MEDIUM-HIGH.

The return register has three authored instances and they are the arc's emotional spine:

• Ch 70, the nostos — the road's rehearsal. docs/spine/CH_70.md L3 frames the whole node as

*"katabasis survived as nostos,"* and L32 lands it: *"not rest. The proof that she would choose the

weight of what she remembers over the mercy of setting it down, and climb."*

• Ch 75, the homecoming — the real one, and it is interrupted inside its own opening paragraph.

L26: *"She comes through the last root of the World-Tree expecting home, and finds a war."* The

reunion with the childhood friend is explicitly denied its moment: *"there is no time for the

reunion the moment deserves."*

• The Epilogue, the record — *"a warm, still, resolving register"* (L181).

Nothing connects them, and Ch 75's homecoming register is authored to last roughly one movement

before fury takes the node. Spiritfarer's actual lesson here is not about grief content — ours is

excellent — it is about DWELL TIME: a homecoming register that is not given uninterrupted minutes

does not register as a homecoming at all.

RECOMMENDATION (reuse-first, already funded). REALMARC's §5 field-mint list already funds *"the

CH_75 grief_country site split so the Movement-1 low-threat grief country stops sharing a row with

the war sites"* and Clause 5's mastery terrace at the 74→75 seam (*"the vril ceiling exercised once

with nothing shooting back, before Ch 75's urgency takes over"*). Both already exist. This finding

adds one requirement to work that is already scheduled: **the split must carry the register change in

data** (grief_country → grief, war sites → fury), and the childhood-friend reunion beat must

sit inside the low-threat half. That converts an intra-node register change from prose intent into a

row a tooth can read, and it is the only place on the road where the homecoming can be FELT before

the war claims it.

F-R18 — The proposed register rubric reads PROSE; the convergence is in the ART layer, so it cannot see it

Verdict: MISSING (rubric). Owner: QA loop. Wave 2 (rubric spec) / P2 (execution). Priority: HIGH

under the Josh gate — the AI QA loop is the only player until the polished slice, so a rubric that

cannot see a defect means nobody can.

REALMARC §5 specifies RB-ASC-REGISTER: *"name each node's dominant register from unlabeled prose;

if the critic names the same register twice, the declared column is lying."* That is a good rubric

and it should ship. But F-R03 shows the convergence lives in the art-direction and audio bullets, and

a prose rubric run on the body prose will pass cleanly on exactly the corpus that is already

differentiated. The rubric and the defect are in different layers.

RECOMMENDATION — RB-REG-FRAME, the frame-side twin, specified for direct pickup:

• INPUT: N unlabeled frames (or concept boards, pre-hardware) drawn from two realms on the road,

interleaved, with no captions and no filenames.

• TASK 1: sort the frames into two realms. TASK 2: name each realm's dominant register from the

frames alone. TASK 3: name any register on that node's register_forbidden list that the frames

read as delivering.

• FAIL: the critic cannot sort the frames, OR it names the same dominant for both, OR it names a

forbidden register. A concept wave whose two death realms cannot be told apart by eye has failed

the doctrine no matter what the column says.

• ADVISORY, never gating, per the existing ladder — same construction as RB-RLM-LAWFUL in the

charter's §2.7, and it should be filed beside it rather than in a new rubric family.

• BOUND declared on the rubric: it measures DELIVERY, not declaration; the arithmetic tooth measures

declaration, not delivery. Neither may inherit the other's claim.

• Wiring: docs/proposals/QA_WATCHING_PROGRAM.md §8 aspect_id join — RB-REG-FRAME is the rubric row

for R-A6, and it is a watchable-surface rubric under that program's own contract.

F-R19 — Nothing probes whether the register enum is WIDE ENOUGH, and the enum has a width tension in both directions

Verdict: MISSING (a gate item). Owner: schema-data. Wave 2. Priority: HIGH.

The tension, stated once so the schema designer inherits it rather than discovering it: an enum too

NARROW forces distinct realms onto one value and manufactures monotone in the data (F-R02 shows the

seeded eight does exactly this); an enum too WIDE makes the rotation rules trivially satisfiable and

the tooth green by construction — with thirty values, "no two consecutive share" is not a

constraint, it is an accident. Neither failure is visible from inside the tooth.

Nothing in either proposal probes for this. The charter has already solved the identical problem for

the physics vocabulary at §2.3 — *"THE V-ADEQUACY POSITIVE CONTROL, a required tooth and not an

optional one … take a known-real bend from each of the nine traditions that is deliberately held

OUTSIDE V, author it into a fixture row, and SHOW T-RLM-PARTITION passing it green."*

RECOMMENDATION (reuse-first — clone the pattern, do not invent one).

• THE ENUM-ADEQUACY PROBE. For each node, attempt assignment from the enum using only that node's

authored register prose. Every node that cannot be assigned without collapsing a distinction the

prose makes is residue; each residue item widens the enum; re-run. **§1.2 IS that probe run over

Ch 66-77 plus the Epilogue, and its residue is the five widenings.** Its bound is recorded: the

Ch 2-65 probe is UNRUN, so enum v0.1 at 13 values is a lower bound, and the enum is not

ratified until the arc-wide probe runs and its residue is recorded in the schema doc.

• THE WIDTH GUARD, the direction the V-probe does not have. Before ratification, assert that the

rotation rules are non-trivial at the ratified width: run the tooth against a RANDOM assignment

over the ratified enum across 79 nodes and show it FAILS. A rotation rule that a random assignment

passes is not armed. Cheap, deterministic, zero-token, and it is the only defence against the

widening making the tooth ceremonial.

F-R20 — check_emotional_curve.py does not exist

Verdict: DESIGNED-UNBUILT. Owner: schema-data. Wave 2. Priority: MEDIUM (informational, so no

report claims a tooth that is not there).

ls harness/ returns check_chapter_topography.py and check_reveal_discipline.py (positive

control) and no check_emotional_curve.py and no check_realm_ruleset.py. Gate 33 is fully specced

(E1-E6, waiver file docs/emotional_curve_baseline.json, stale-waiver FAIL) and unbuilt. Recorded so

that no Wave-1 or Wave-2 return says "the register gate" as though it were armed, and so the mutation

discipline is inherited: not counted as armed until a mutation has been shown to fail it.

F-R21 — T0_Theme_Registry carries 12 rows and ZERO realm themes

Verdict: DESIGNED-UNBUILT. Owner: schema-data. Wave 2-3. Priority: MEDIUM.

The 12 live rows are `PROTAGONIST_THEME · HOUSE_OF_VELHEIM_THEME · ARCHITECT_THEME · CASSIUS_PHASE_1

· CASSIUS_PHASE_2 · CASSIUS_PHASE_3 · GRAND_SAGE_REVEAL · ENDING_TRULY_GOOD · ENDING_GOOD_ENOUGH ·

ENDING_EVIL · VIMANA_ACTIVATION · FAMILIAR_BOND` — every one a character, family or ending theme.

No realm, region or place theme exists, while every road chapter's music_mood bullet ends

[POPULATE→T0_Theme_Registry] and names two or three motifs by name (Ch 70: *"the descent motif, the

two-springs memory-choice motif, and the passage-won return motif"*).

So the register column's second declared consumer has nowhere to land its output. This is the

per-realm audio identity aspect (R-A16) that the lens-gap doc parks on

presentation-craft-anim-audio; I am not claiming it, only recording that the register mint's audio

payoff is gated on realm theme rows existing, and that the motifs to seed them are already named in

canon prose at every node. One row per realm, motif names extracted mechanically, is a Wave-3

zero-authoring item.

Note also, in passing and out of scope: the live CASSIUS_PHASE_1 row carries its own honest defect

note — *"the mood_tags and instrumentation still carry the pre-re-architecture combat register and

read against the executed SOCIAL stage — a Theme population-pass item, not re-scored here."* It is a

register-drift instance already caught and parked, which is corroborating evidence that register

drift in this repo is real and survives in data until something reads it.

---

3. SCHEMA-DESIGNER HANDOFF — direct pickup for the T0_Realm_Ruleset_Registry mint and the Chapter_Index column adds

Written so this section can be consumed without reading the rest of the document.

3.1 Field specs

On T0_Chapter_Index (79 rows), via the topography POPULATION-CONTRACT mechanism — the same

path STORY_DEPTH specifies: columns appended to the live CSV, declared in

docs/fidelity_baseline.json added_columns, re-baselined in the same commit.

Slop-mode list seeded from the authored clauses: `horror-gore · whimsy · spectacle · caricature ·

buffoonery · theme-park · foe-scoring · hostile-menace`. Expandable by ruling, examples-not-limits.

On the seam rows the Fork-C transitions pass is already creating: register_from,

register_to (REALMARC Clause 5 — already proposed there; this slice supplies the values).

On the site rows: site_register (closed enum, optional) — needed only where a node changes

register internally. Two known consumers today: the CH_75 grief_country split and CH_76's two

phases (F-R07).

**On T0_Realm_Ruleset_Registry (the charter's §2.3 mint), a REGISTER BLOCK beside the INTENSITY

BLOCK:** register_dominant_ref and register_secondary_ref as foreign keys onto the chapter row

rather than duplicated values (one source of truth; the realm row reads the chapter row), plus

register_forbidden_ref. Rationale for FK-not-copy: the charter's own T-RLM-CURVE exists precisely

to stop the ruleset and the spine silently diverging on the intensity axis; copying register values

would create the same divergence class the same schema is already defending against.

3.2 The mapping table (F-R05) — required in the same commit or the "upstream source" claim is dropped

emotional_register → T0_Theme_Registry.mood_tags[], seeded only from tags the live registry already

uses: elegiac → [mournful, contemplative] · exacting → [ceremonial, intellectual] ·

adversarial → [antagonistic, escalating] · hospitable → [warm, bonded] ·

ascendant → [transcendent, triumphant] · wonder → [transcendent, hopeful] ·

grief → [tragic, mournful] · dread → [ominous, foreboding] · fury → [militant, climactic] ·

tenderness → [intimate, bonded] · wry → [intellectual, manipulative] ·

awe → [transcendent, ceremonial] · resolve → [heroic, redemptive].

3.3 Teeth (gate 33, harness/check_emotional_curve.py)

Adopt E1-E6 as STORY_DEPTH specifies, with three amendments this slice supplies:

• E1 amended to the stricter rule (no TWO consecutive share) per F-R01's recommendation.
• E2 unchanged (every five-node window ≥3 distinct).
• E6 unchanged and load-bearing — unpopulated is a HARD FAIL, and the placeholder form

[POPULATE…] fails. Reuse check_chapter_topography.py's existing _unpopulated(v) helper (L69)

rather than writing a second one (F-R04).

• NEW E7, THE PAIR RULE. The ordered (emotional_register, register_secondary) pair is unique

inside any five-node window. This is the tooth that catches the real convergence: two nodes whose

dominants differ but whose overall colour is the same. DESIGNER DECISION, with a recommendation:

make the pair ORDERED, not a set, so (dread, awe) at Ch 67 and (awe, dread) at Ch 74 do not

collide — the dominant is the dominant and the order carries meaning. Record the choice in the

gate docstring; an undeclared scope choice is how this defect class propagates.

• NEW E8, THE ART-BULLET AGREEMENT TOOTH (F-R03). The material_palette and style_register and

music_mood bullets of a node must contain a term from that node's declared register's synonym

list. This is the only tooth that would have caught grave appearing in eight consecutive nodes'

art bullets. Ship it with the Wave-3 bullet repair; it is a corpus error, not a per-node one.

3.4 Fixtures — because an unmutated rule was never armed

• E1 MUST-FIRE, from LIVE data: the real Ch 67 / Ch 68 pair, which carry the identical authored

register phrase awe-and-dread (F-R06). Keep as the live fixture until the repair lands, then

replace with a synthetic — the same construction REALMARC uses for its Clause-1 CH_71 fixture.

• E1 MUST-NOT-FIRE: the §1.3 assignment string.
• E2 MUST-FIRE: a synthetic window carrying two distinct values across five nodes.
• E7 MUST-FIRE: two nodes inside one window with the same ordered pair.
• E8 MUST-FIRE, from LIVE data: the real Ch 69-76 art bullets against a grave-free declared

register — this fires today, unmodified, which is the finding.

• THE WIDTH GUARD (F-R19): a random assignment over the ratified enum across 79 nodes must FAIL

E1+E2+E7. If it passes, the enum is too wide and the rules are ceremonial.

• THE ENUM-ADEQUACY PROBE residue recorded in the schema doc, with its bound stated: probed over

Ch 66-77 + Epilogue only; Ch 2-65 UNPROBED; enum v0.1 is a lower bound.

3.5 Engine versus canon (the standing reuse declaration)

ENGINE, transfers whole to game #2: the four field shapes, the enum's STRUCTURE (dominant plus

counter plus forbidden plus evidence), the pair rule, the width guard, the enum-adequacy probe

procedure, the E1-E8 tooth shapes and their fixture discipline, and the RB-REG-FRAME rubric shape.

CANON, does not transfer: the 13 enum VALUES, the per-node assignment in §1.3, the slop-mode

list, and the mapping table's contents.

3.6 Wiring obligations if this is adopted (anti-orphan)

• docs/proposals/REALM_DESIGN_PROGRAM.md §2.3 — the REGISTER BLOCK added to the schema field

families; §2.7 — RB-REG-FRAME filed beside RB-RLM-LAWFUL.

• docs/proposals/STORY_DEPTH_PROGRAM.md — the enum widened from 8 to 13 with the probe residue

recorded; E1 restrengthened; E7/E8 added; the mapping table named as a same-commit requirement.

• docs/pipeline_review/REALMARC_PATTERN_BRIEF_2026-07-27.md — register_dominant recorded as a

retired alias of emotional_register; Clause 3's gate home re-pointed to gate 33.

• docs/proposals/QA_WATCHING_PROGRAM.md §8 — RB-REG-FRAME as the R-A6 rubric row.
• docs/DESIGN_GAP_REGISTER.md — R-A6 / R-A12 / R-A18 aspect rows minted with the verdicts above.
• docs/PRE_5090_BUILD_PLAN.md — the F-R03 art-bullet repair added to Wave 3; the rank-8 freeze rule

amended to name the mapping table as part of the deliverable.

• docs/DOC_MAP.md — a row for whatever lands.

---

4. CONTESTED / DIRECTOR CALLS

Five items the director should adjudicate rather than a schema author deciding silently.

1. F-R01, the column collision. emotional_register versus register_dominant; no-two versus

no-three consecutive; gate 33 versus the topography gate. My recommendation and its strongest

objection are stated in F-R01. This must be settled BEFORE either mints, and the two owning

documents were written a day apart by different lanes, so neither is more current.

2. F-R10, whether Ch 70 should be warm at all. The charter phrase is *"what makes Hades warm."*

Our Ch 70 is authored elegiac and its §17.1 protected-culture block forbids whimsy in as many

words. I ruled elegiac and cited it. If the director wants the Hades-the-game warmth, that is a

re-authoring of Ch 70's care commitment, not a register tag change, and it should go through the

care doctrine rather than through this slice.

3. E7's ordered-versus-set pair. Recommended ordered; the consequence is that (dread, awe) and

(awe, dread) coexist on the same road (Ch 67, Ch 74). That is defensible — the dominant differs

— but it is the weakest distinction in the §1.3 table and RB-REG-FRAME is the check that would

catch it if it reads as one realm twice.

4. F-R15, Annwn's seduction. I declined to assign a register the build does not deliver. If the

Wave-2 Annwn charter builds the food taboo and the felt time-loss, the secondary should move from

resolve to a seduction value and the enum widens by one. Decide at charter time, not now.

5. F-R13's reframe. The charter and the lens-gap doc both say "five consecutive death realms."

The built spine makes Annwn the break. Recommend correcting the framing in both documents so

later waves do not size four nodes of work as five, and so nobody later "fixes" Annwn's warmth.

---

5. WHAT THIS SLICE DID NOT AUDIT (the bound, stated so no report over-claims it)

• The craft remainder of L-R5 — the Spiritfarer/Grim Fandango/Edith Finch craft lessons at

encounter and vignette grain ride P2 with the factory per the charter's §4.2. Only the register

doctrine slice ran here.

• Ch 2-65. The enum-adequacy probe covered thirteen endgame nodes. Enum v0.1 is a LOWER BOUND

and is not ratifiable until the arc-wide probe runs (F-R19).

• The crossing conveyance (R-A7b) — owned by the live transitions fork

(FORKC_ANALYSIS_DESIGN_PRECEDENT_2026-07-27.md); not re-litigated here.

• Realm audio identity (R-A16) — parked on presentation-craft-anim-audio by the lens-gap doc;

F-R21 records only the dependency, not a claim on the aspect.

• Whether the built prose actually DELIVERS its declared registers in play. No frame exists to

read. Every claim above is about authored artifacts and declared data, and RB-REG-FRAME is the

instrument that would close that gap once concept boards exist. No report may inherit a

play-quality claim from this document.

Generated by harness/site/structure_site.py — the URL path is the repo path. review root