L-R3-sacred-space-living-tradition_DRAFT.md

pipelines/L-R3-sacred-space-living-tradition_DRAFT.md

L-R3 · sacred-space-and-living-tradition-adaptation — lens audit (DRAFT)

Wave 1 doctrine lens, read-only, 2026-07-28. Repo C:\dev\humanity-forgotten (canon). No file in

either repo was edited. Written to be consumed directly by (a) the Wave-2 HARMONY DOCTRINE authoring

and (b) the T0_Realm_Ruleset_Registry schema mint.

Title-set audited: Never Alone (Kisima Innitchuna) · Okami (dual-listed with L-R1) ·

AC Origins — Curse of the Pharaohs (Duat / Aaru) + Discovery Tour · Raji: An Ancient Epic ·

Mulaka · Pentiment.

Aspects audited: R-A2 (adapting a real cosmology at scale) · R-A9 (judgment/moral-weighing as

a mechanic) · R-A11 (deity presence without instantiation) · R-A15 AS A BOUND ON ART CONCEPTS

· R-A20 (the care tiering of the realm set).

Calibration in force: CVD §17.1 as ruled 2026-07-22 — care means **weight and authenticity, never

omission; only the hard-line FLOOR is absolute; care-tier inflation is itself a defect**; the

devil's-advocate pass against over-conservatism is run on every ruling below and its results are

recorded in §1.5 and §4 rather than suppressed.

---

0. Method and positive controls

Every zero below was positive-controlled before it was reported. The controls, with the term known to

be present that proved the matcher could fire:

Chapter refs were read tri-dialect where searched (Ch 72 / Ch-72 / _ch72_); all registry work

was done in Python over the live CSVs, never shell grep.

---

PART 1 — THE NAMED GATING OUTPUT: the per-realm care tiering table

1.1 Why one column cannot carry this (the ruling the table rests on)

The charter already found half of this and got it right: care_tier is a ratified two-value enum

(standard · elevated) whose derivation authority is fixed — *"care_tier — the zone's

cultural-representation care register (standard · elevated), derived from the protected flag and the

page's explicit care-register overlay where one is ruled"*

(_source/01_Tier_1_Foundation/T1_Region_Page_Template [ACTIVE v2.0].md L120-121). It cannot say

"living," and this lens does not widen it. The charter's generation_class split (§3.3) is CORRECT

and this lens RATIFIES it as the second axis.

But two axes are still one short, and the missing one is what actually blocks the art fan-out. Three

independent facts are being collapsed:

• CARE BAND — *how the living community around a tradition is portrayed.* Ratified two values.

Governs critic routing and review depth. Its real trigger is **a living community on stage in the

chapter**, which is what hard_line_status = CULTURAL_PROTECTED already encodes

(T0_Chapter_Index: exactly 4 rows — CH_29, CH_42, CH_50, CH_57).

• GENERATION CLASS — *whether a model may raw-generate the accurate depiction of this place.*

Proposed by the charter, ratified here. Its trigger is **whether a real tradition is being

depicted**, independent of whether its people are on stage.

• SOURCE-CORPUS STATE — *whether a citable reference corpus exists to build Panel A from at all.*

MINTED BY THIS LENS. Without it, generation_class = living is an unfundable instruction: it

forbids Lane G on Panel A while nothing tells the fan-out that Annwn has no acquisition-listed

corpus to assemble a Lane-R plate sheet from (§2 F-10). A rule that bars the only available lane

and provides no other is a rule the fan-out will quietly break.

The proof the axes are genuinely independent is in the live data, and it runs BOTH ways:

• Atlantis — generation_class = invented (nothing real to misrepresent) AND an **elevated

anti-fringe duty** attached to its *reception* history, not its content. Two different columns,

both true. (The charter's C-4, confirmed.)

• Ch 7 South India vs Ch 72 Naraka and Meru — the same living religion. T0_Chapter_Index CH_07

reads standard care; Tamil/Malayali/Siddha/Kalaripayattu/Meenakshi/Murugan per primary-source;

CH_72 reads MAXIMUM care — Hinduism is a LIVING major world religion (~1.2B adherents). The

chapter where the player walks among living Hindu worshippers at a living temple is STANDARD; the

dimensional realm with no living community present at all is MAXIMUM. That gradient is

inverted, and it is inverted because the realm *felt* bigger, not because the duty is bigger

(§2 F-09).

So the table below rules three columns, and the third is the one that unblocks the wave.

1.2 THE PER-REALM CARE TIERING TABLE

Enum values used, all closed:

• care_band ∈ {standard, elevated} — the RATIFIED enum, not widened. No maximum, no open,

no high, no prose.

• generation_class ∈ {invented, historical, living, unsettled} — the charter's §3.3 enum.

This lens's output eliminates every unsettled.

• source_corpus_state ∈ {listed, section_only, absent} — NEW. listed = has an RCM

Acquisition-Priority block; section_only = has an RCM per-chapter substrate section but no

acquisition block; absent = neither.

• deity_depiction ∈ {iconographic_ok, iconographic_bounded, n/a} — NEW, and it is the

R-A11/R-A15 bound the fan-out actually needs (§2 F-05). It governs art, and is orthogonal to

the separate instantiation_class field proposed at §3.

The deity_depiction bound (the R-A11 / R-A15 column, per row)

1.3 What the art fan-out MAY and MAY NOT generate — the four rules that bind every row

• R1 — LANE BY generation_class, on Panel A only. living and unsettled → Panel A is **Lane R

(rights-cleared reference) only**; Lane P on Panel B must carry a Lane-R underlay. historical →

Lane G barred from Panel A, legal on Panel B over a Lane-R underlay. invented → Lane G legal on

both. After this table there are no unsettled rows on the road, so the strict default fires

nowhere and the gate stops being a bottleneck. Rows 1 and 10 (Atlantis, fairy realm) are the ONLY

Lane-G-on-Panel-A rows in the whole program.

• R2 — CITATION BINDS ON ALL ELEVEN ROWS REGARDLESS OF LANE. generation_class = invented licenses

the *pixels*, never the *sourcing*. Atlantis's accurate depiction is Critias's own text —

concentric rings of land and water, the temple of Poseidon, orichalcum "known only by name," the

hot and cold springs — and a raw-generated Atlantis Panel A is a model's generic-fantasy Atlantis,

which fails Panel A's own definition even though the lane is legal. This closes a real hole in the

charter's §3.3 wording (§2 F-16).

• R3 — source_corpus_state = absent BLOCKS THE BOARD, NOT THE LANE. Annwn's first board cannot

ship until the Mabinogi corpus is acquisition-listed. This is the one hard block this lens

imposes, and it is a three-line registry edit plus an acquisition pass, not a wave.

• R4 — deity_depiction GOVERNS ART; instantiation_class GOVERNS PLAY. They are separate fields

and a board may never be bounded by the play field. Concretely: the Duat board renders Anubis at

the scales AND the Duat's boss is still Apep. Both are true at once and today nothing in the corpus

says so.

1.4 Scope note on the prompt's "Mesoamerican pair"

Canon carries one Mesoamerican realm (Xibalba, Ch 71) and its earthly seed pair (Ch 47 / Ch

48, the Maya world). Verified: all three read elevated consistently — heading, T0_Chapter_Index

cell, and every one of the 11 BE_0047*/BE_0048* boss rows plus BE_0071. **This is the only

culture-cluster on the road whose care register is coherent across the spine, the chapter registry

and the boss registry simultaneously, and it is the template the other eight rows should be brought

to.** Recorded as a COVERED positive, not a gap.

1.5 The devil's-advocate pass, and the four inflations this lens REFUSES

Run per CVD §17.1 ("*a devil's-advocate pass calibrated specifically against over-conservatism, so

both failure directions are checked — defanging as rigorously as under-protection*"). Four

recommendations that a timid reading would have produced, and why each is refused:

• REFUSED: promoting Ch 71 and Ch 72 to hard_line_status = CULTURAL_PROTECTED. Tempting — they

are the two most-living traditions on the road and they sit at NONE while Ch 29/42/50/57 sit at

CULTURAL_PROTECTED. But that flag's real meaning is **a living community on stage in the

chapter**, and it drags Ch 57's content floors with it (no antagonist-family presence, no combat

framed against community representatives). Xibalba and Naraka are **dimensional realms with no

living community present**, and their marquee bosses — the Lords of Xibalba, Vritra — are ratified

fought encounters. Promoting them would import floors that would defang two of the road's best

fights to solve a problem that belongs in a different column. NONE is CORRECT. The real gap is

generation_class, which this table now fills.

• REFUSED: keeping MAXIMUM on Ch 72. CVD §17.1 names exactly three maximum-depth chapters —

Ch 29 Sámi, Ch 42 Haiti, Ch 57 Australia — and HL_0009 enumerates the same three. Ch 72 minted a

fourth at the spine without amending either. Nothing is subtracted by de-escalating it to

elevated: every listed commitment stands verbatim, the Vritra death-blow fork stays offered

(the node itself already rules "care_tier stays MAXIMUM on BOTH outcomes" — the same logic holds at

elevated), and review depth is unchanged. What is gained is a valid enum value and the end of an

inverted gradient against Ch 7.

• REFUSED: recommending a cultural-consultation lane. Never Alone's headline is community

co-development, and the naive lens finding is "we have none." But the CVD explicitly deprecated

the cultural-consultation flag at lock (*"The cascade-era flag system using bracketed tags such as

the verification flag, the cultural consultation flag, and the bridge flag is deprecated as of CVD

lock"*), and this is a solo build with a three-subscription AI factory. Recommending a consultation

apparatus is recommending something that cannot exist. DIVERGENT-BY-DESIGN. The translatable

half of Never Alone's lesson is in F-12/F-14 and costs one enum value.

• REFUSED: reading audit-only as "do not depict." This is the single most expensive available

timidity on the whole program, and §2 F-05 is its full treatment. Rendering the Hall of Two Truths

without Anubis at the scales is not care — it is an inaccurate depiction of an ancestral tradition's

own picture of its own cosmos, delivered under a panel header that says "accurate," and it produces

the depopulated establishing shot the charter's own de-slop critic is built to catch.

---

PART 2 — FINDINGS

Each: verdict class · owner lane · wave · priority. Priority is player-impact × solo+AI feasibility.

F-01 · The only machine-readable care column in the corpus has a 15-value vocabulary against a 2-value ratified enum

WEAKER · schema-data · Wave 2 · HIGH

registries/T0_Boss_Encounter_Registry [DRAFT v0.1]/Sheet1.csv carries a live care_tier column.

Exhaustive distribution over all 281 rows: standard 198, elevated 51, high 12,

standard_elevated 5, maximum 2, open 2, plus **nine one-off prose sentences slugified into the

cell** — e.g. standard-with-the-war-register-antisemitism-guardrail-and-the-honored-realm-duties,

elevated-the-realms-own-dead-given-dignity-win-is-rest,

hard-line-with-the-single-sanctioned-reveal-and-the-al-hallaj-reverence. The ratified enum is two

values. Five of the nine road bosses carry off-enum values (BE_0071 elevated OK; BE_0072 maximum;

BE_0073/BE_0074 open; all BE_0075*/BE_0076/BE_0077 prose). Separately BE_0069 (the Duat)

reads standard while its spine heading reads "elevated Egyptian duties" — a live contradiction

inside the corpus.

Reuse-first fix: normalise this column to the ratified two values and move everything else into

the new fields at §3; the deterministic tooth is T-RLM-CARE per the charter §2.7, which is the

declared home. No new registry.

F-02 · No per-realm care field exists anywhere the art fan-out can read (the gating gap this lens closes)

MISSING · schema-data + spec-doc · Wave 2 · HIGH

The realm care register lives in exactly two places, both prose: the spine node's `## Protected-chapter

/ living-culture framing (...) heading text, and T0_Chapter_Index.cultural_depiction_compliance`,

a free-prose cell. Neither T0_Chapter_Index nor T0_Region_Index has a care_tier column (columns

enumerated; neither carries one). No region page exists for any realm (the 12 pages under

_source/02_Tier_2_Region_Pages/ are all Earth regions), so the Zone Catalog's ratified care_tier

extension column has no realm rows to sit in. harness/check_care_scope.py is a presence gate for

one firewall in three homes, not a per-realm care gate.

Net: the charter's own requirement — *"the per-realm care tiering, delivered as a table the

art-concept fan-out reads before generating anything"* — has no field to read. §1.2 is the table;

§3 is the field set.

F-03 · Hard-line high-care scoping is closed at three earthly chapters and reaches no realm

WEAKER · schema-data · Wave 2 · HIGH

T0_Hard_Lines [ACTIVE v0.2] applies_to_high_care_chapter, exhaustively enumerated over all 123

rows: {none: 107, Ch29|Ch42|Ch57: 6, Ch29|Ch57: 4, Ch57: 5, Ch29|Ch42: 1}. **Not one row scopes to

any realm chapter.** Consequence: docs/spine/CH_71.md cites HL_0110 and HL_0111 as binding at

Xibalba, but HL_0111's own applies_to_high_care_chapter reads Ch29|Ch42|Ch57 — the spine invokes

a rule beyond its declared scope, so the citation is decorative and no tool will enforce it at Ch 71.

HL_0110 survives (its rule_text is class-scoped: "living-religion synthetic-voice DEFAULT"), HL_0111

does not (its rule_text enumerates Ch 29/42/57 by name).

Reuse-first fix: widen HL_0111's scope column and rule_text to the living rows of §1.2 rather

than minting a realm-specific hard line. One registry edit + baseline refresh.

F-04 · "audit-only" is a spine coinage with no tier-doc definition, cited to a hard line that does not contain it

WEAKER (mis-citation) · spec-doc · Wave 2 · HIGH

audit-only appears 22× in CH_68.md, 16× in CH_69.md, 4× in CH_67.md, 1× each in CH_66/CH_71

— and zero times in any T1 foundation doc (positive-controlled, §0). It is attributed in-line to

HL_0003: *"Ma'at and the assessor-deities audit-only per HL_0003"* (CH_69.md L102, L179). HL_0003's

full text reads: *"The faithful community and its living tradition are never the story antagonist and

are never villain-framed, defamed, or set up for extermination... This is collective, framing-level

protection of the community as a whole; it does not render individual members mechanically

un-attackable."* It is a community-framing rule. It says nothing about deity instantiation, and

its closing clause points the opposite way. This repo has caught this exact class before (the three

T1_Integrity_Paths §17.5 mis-citations, CLAUDE.md).

Reuse-first fix: define the rule where it belongs — as an instantiation_class field on the

realm ruleset row (§3) — and re-point the spine citations, or mint the rule properly in

T0_Hard_Lines if it is to remain a hard line.

F-05 · audit-only conflates INSTANTIATION with DEPICTION, and the art fan-out will read it as "do not render"

MISSING · spec-doc → schema · Wave 2 · HIGH — this is the named R-A15 bound

Nothing anywhere states which of these audit-only means: *never an NPC/boss* (instantiation) or

*never on screen* (depiction). The Ch 69 text supports the first — "*NONE of them is ever reproduced

as a character, an NPC, or the boss*" — but an art fan-out reading only the badge will bound Panel A.

The consequence is exact and severe: **the Hall of Two Truths is Book of the Dead Spell 125, and the

Papyrus of Ani's own picture of it IS Anubis at the scales, Thoth recording, Ammit waiting, the

forty-two assessors ranged.** A Panel A that omits them is not accurate; it is the tradition's hall

with the tradition's own iconography removed, shipped under a header saying "this is what the place

actually looks like." It also lands squarely on two named slop modes the charter already lists — the

depopulated establishing shot and the empty realm — and on the recorded

"no-enemy mythological realms" defect.

The AC Origins precedent is decisive here and is the reason this lens exists: Ubisoft shipped a

playable Egyptian afterlife (Curse of the Pharaohs — the Duat, Aaru/the Field of Reeds) in which the

Egyptian divine beings are rendered at full monumental scale, and shipped Discovery Tour beside

it as a separately-framed, curator-narrated, cited scholarship layer. **They separated game from

scholarship by SEPARATING THE LAYER, not by emptying the game.** Our audit-only rule performs the same

separation by thinning L1 — which CVD §17.1 explicitly forbids: *"Care tiers add depth, never

subtract content... High-care governs cultural authenticity only; it never lowers the action, combat,

or exploration ceiling."* We already own the Discovery Tour analog: the L2 investigatable-depth and L3

connective layers. Put the scholarly weight there, where it already lives, and let L1 render the

tradition's own iconography at full spectacle.

Reuse-first fix: split into two declared fields — deity_depiction (art, §1.2 table) and

instantiation_class (play, §3) — on the realm ruleset row. Zero new systems.

F-06 · The deity-presence policy is inconsistent across the road, undeclared, and the gradient is inverted

WEAKER · spec-doc · Wave 2 · HIGH

Measured across the nine road nodes: audit-only mentions run CH_66:1, CH_67:4, CH_68:22, CH_69:16,

CH_70:0, CH_71:1, CH_72:0, CH_73:0, CH_74:0. What the nodes actually do:

• Mesopotamian deities — never bodied (Tiamat, Apsu, Marduk, Nammu).
• Egyptian deities — never a character (Ma'at, Anubis, Thoth, Ammit, Osiris, 42 assessors).
• Greek deities — on stage as named characters: Hades, Persephone, Charon, and the judges Minos,

Rhadamanthus, Aeacus, all "honored and never fought" (CH_70 uses "honored" 61×).

• Hindu — Yama is "the JUST judge (Dharmaraja)," Chitragupta keeps the ledger; deities present and

honored; zero audit-only.

• Welsh — Arawn is a full NPC: host, and "by the end, a firm friend and ally."
• Norse — gods of Ásgard, the Norns, the light-elves honored and present.
• Finnish — Louhi is the fought ultimate boss (constitutional, CVD §17.1).

The gradient is inverted: the traditions held most tightly (Mesopotamian — no practitioners;

Egyptian — ancestral with revival) get the *strictest* treatment, while traditions with comparable or

larger living practice (Hellenic revival, living Welsh culture, living Heathenry, living Hinduism) get

full on-stage characters. And the CVD rules the permissive direction outright: *"Deities are equally

real, portrayable, and usable as characters — a good deity may appear as a disguised human quest-giver

or a rescuer... the only firewall is that no deity-name sits on a mortal NPC and no revered or real

figure is cast as the villain."*

Reuse-first fix: declare instantiation_class per named being on the ruleset row and make the

gradient a *design* decision instead of an accident of authoring order. This lens does not

recommend tightening Hades/Annwn/Yggdrasil; it recommends asking, at Wave 2, whether Ch 68/69 are

tighter than the doctrine requires (see §4 C-1).

F-07 · No deity anywhere on the nine-realm road grants the player anything

WEAKER · spec-doc + P5-factory · Wave 2 (rule) / P2 (build) · MEDIUM-HIGH

Boon/blessing/divine-aid regex across CH_66-77: CH_67 7 (all "blessing," and all about the Sampo's

blessing dispersing on sea and land — not a boon to the player), CH_77 1, all others zero

(positive-controlled against CH_15 = 2). So on a road of nine realms populated with honored deities,

sovereigns and functionaries, not one of them gives the player a thing — no boon, no

transformation, no invocation, no rescue-and-re-encounter.

This is the R-A11 aspect's shipped craft, and both regionally-authored titles in my set do it:

Mulaka (Lienzo, Rarámuri tradition) makes the tradition's own demigods summonable powers while

the *sukurúame* sorcerers are the enemy class — the tradition's own categories carry both roles.

Okami goes further and makes the player *be* Amaterasu, with the Celestial Brush — the tradition's

own visual grammar — as the core verb. Holding an honored deity inert on screen is the defanging

failure direction; the CVD's own text licenses the fix.

Reuse-first fix: the deity-granted boon is a signature-law EXPLOIT slot on the ruleset row

(the charter's §2.3 already has EXPLOIT, "how a clever player turns the law into a tool"). One

authored value per road realm. It is bounded by the D-RR.5 constraint the charter already arms in

T-RLM-COMPOUND — a boon must never grant a mastery gate or confer school membership.

F-08 · defeat_protected / has_nonkill_path are blank on every realm boss, including the five canon rules must never be slain

DESIGNED-UNBUILT → schema-data · Wave 2 · HIGH

Both fields exist and are populated on 16 of 281 rows — all of them Ch 2-5 (the slice), including

BE_0005_S3 "Naga Padoha, the bound earth-shaker (Batak…)" at defeat_protected = TRUE. So the

semantics exist and have already been used for a tradition's honored/bound being; the realm road just

never got the pass. Every BE_0066–BE_0074 row reads defeat_protected = '' and

has_nonkill_path = ''. The sharp instances:

• BE_0067 Louhi — win_condition = REDUCE, primary_verb = strike, has_nonkill_path blank —

against CVD §17.1 constitutional text: *"she is NEVER slain, captured, or crushed — she is

repelled and withdraws to Pohjola still sovereign."* The one boss whose non-slaying is written into

the constitution carries no data flag saying so.

• BE_0070 Cerberus — win_condition = PERCEIVE;PACIFY, primary_verb = pass (correct!) but

has_nonkill_path blank. The boss whose *only* canonical outcome is non-kill has the non-kill flag

empty.

• BE_0074 Níðhöggr — verb re-root, canon "driven back and never slain," flag blank.
• BE_0068_S1 the Mušḫuššu-class host — REDUCE/strike, canon "recalled by the mother when she

is heard, never slain," flag blank.

• BE_0072 Vritra — a Josh-ruled offered spare/slay fork, and neither branch is in data.

Reuse-first fix: populate the two existing columns on the 9+ realm boss rows; no schema change.

This is the cheapest high-value item in the audit.

F-09 · Care-tier inflation at Ch 72, and an inverted gradient against Ch 7

WEAKER · spec-doc + schema-data · Wave 2 · HIGH

T0_Chapter_Index CH_72: MAXIMUM care — Hinduism is a LIVING major world religion (~1.2B adherents);

BE_0072.care_tier = maximum. CH_07: `standard care; Tamil/Malayali/Siddha/Kalaripayattu/Meenakshi/

Murugan per primary-source`. Same living religion; the chapter with the **living community and a

living temple on stage is standard, the dimensional realm with no community present** is maximum.

CVD §17.1 and HL_0009 both name exactly three maximum-depth chapters (Ch 29, 42, 57) and Ch 72 is not

among them. Under the ruled calibration, care-tier inflation is itself a defect.

Reuse-first fix: §1.2 rules Ch 72 to elevated with every commitment intact; and the real

question the inversion exposes — whether Ch 7 is *under*-scoped — routes to the earthly-care pass, not

here.

F-10 · Annwn is the one realm with no acquisition-listed corpus, and it is a living row

MISSING · schema-data + spec-doc · Wave 3 (but blocks Annwn's first board) · HIGH

Independently positive-controlled (§0): the RCM's Acquisition-Priority H3 blocks run Ch 66, 67, 68,

69, 70, 71, 72 and jump to Ch 74; there is no Ch 73 block, while the same matcher returns H2

substrate sections for all nine including "## Ch 73 — Annwn." So a generation_class = living

realm — Panel A restricted to Lane R by rule — has **no acquisition-listed reference corpus to

assemble a Lane-R plate sheet from**. That is precisely the condition under which a fan-out reaches

for Lane G and produces a model's guess at a living culture's Otherworld under an "accurate" header.

This is why §1.2 mints source_corpus_state and why R3 blocks the board rather than the lane.

Reuse-first fix: add the Ch 73 acquisition block naming the Four Branches of the Mabinogi and

*Preiddeu Annwn* (the realm's own corpus is already NAMED in the substrate section's gap note per CVD

§19.8 — it just was never acquisition-listed). One RCM edit.

F-11 · Three realm region_id primary keys still carry the CVD-retired coinages — with a care consequence not yet recorded

DESIGNED-UNBUILT (known) · schema-data · Wave 2 · MEDIUM

T0_Region_Index mythic_realm rows: pohljar (Ch 67), tiamarath (Ch 68), maatherion (Ch 69).

CVD §17.7 deprecates all three by name and rules *"Where any preceding doc reference contradicts a

name on this list, the legacy reference is deprecated… do not maintain dual references."* **This is

already recorded** in docs/pipeline_review/REALM_ANALYSIS_CANON_INVENTORY_2026-07-27.md ("Three

primary keys still carry the retired coinages"), so it is not minted fresh here.

The care consequence IS new: region_id is the ratified snake_case key and the region page's home

(T1_Region_Page_Template L25: *"_source/02_Tier_2_Region_Pages/<region_id>.md"*). So the Duat's

art board directory, its art bible, its Zone Catalog rows and its rights ledger would all key on

maatherion — a retired occult-adjacent coinage — on the one realm carrying an elevated Egyptian

duty and an Agartha firewall. Re-key before the first Duat board, not after.

F-12 · No axis anywhere records whether a source is authored from inside the tradition (the Never Alone lesson, translated)

MISSING · schema-data · Wave 2 · HIGH

The live citation discipline is use_class, a seven-value enum rendered inline on region pages

(_source/02_Tier_2_Region_Pages/flores_island.md §16): `SUBSTRATE · ESOTERIC-PRIMARY ·

AUTHORIAL-REBUTTAL · REFERENCE · TONE · DUPLICATE · FRINGE-SKIP`. **Not one value records authorship

standpoint.** Never Alone's entire structural achievement is that the authority in the game is the

community's own — co-developed with Iñupiat elders and storytellers, with **elders' documentary

material as the unlock reward**. We cannot co-develop on a solo build (§1.5), but we can make the

insider voice a citation property, which is the half that survives translation.

Reuse-first fix: one new value or one adjacent boolean on the existing use_class rail —

INSIDER-PRIMARY (authored by a member of the tradition) — and one rule: **every living row in

§1.2 must carry at least one INSIDER-PRIMARY citation on its Panel A citation line.** For Xibalba

that is trivially satisfiable (the Popol Vuh reaches us through translations made with living K'iche'

keepers, a fact the node already states). No new registry, no new lane.

F-13 · The cultural-consultation apparatus was deliberately retired; the loop is one-way by design

DIVERGENT-BY-DESIGN · director note · — · —

CVD (both the §15 authoring-discipline section and its bullet restatement): *"The cascade-era flag

system using bracketed tags such as the verification flag, the cultural consultation flag, and the

bridge flag is deprecated as of CVD lock… Where canon is unsettled, surface to author for resolution

before drafting."* So the living-tradition apparatus is substrate reads + adversarial critics + the

Josh backstop, with no community voice at any point — by ruling, not by omission. Recorded here so

no later wave re-litigates it as a gap. The compensating mechanisms are F-12 (insider citation) and

F-14 (insider voice as the L2 reward).

F-14 · L2 rewards our words about the tradition; Never Alone's L2 reward is the tradition's own voice

WEAKER · P5-factory · P2 · MEDIUM-HIGH

CVD §17.9 defines the investigatable-depth track as "journal, Grimoire link, inscription, language

fragment, or cultural-tradition content" — all authored by us. Never Alone's unlock reward is

documentary footage of Iñupiat elders speaking for themselves, and that is *why* its L2 layer

reads as respect rather than as a museum placard. The realm road is where this matters most, because

five of nine rows are living.

Reuse-first fix: the rail already exists — region pages already render inline [SRC: SRC_00483]

attributions and the substrate carries author/title/year per T0_Research_Corpus_Registry. Rule that

on every living row, the L2 completionist reward surfaces a **named, attributed source in the

tradition's own framing** rather than a paraphrase. Costs an authoring convention, not a system.

Pentiment is the craft teacher for making that apparatus diegetic rather than a pop-up — its

whole design carries scholarly apparatus (marginalia, the manuscript frame, typeface-by-speaker)

without breaking immersion, which is our natural-voice doctrine (Josh 2026-07-26) already ruled

and therefore COVERED; the extension worth taking is making the *source itself* a diegetic surface

(the Grimoire as the object you are making), not adding a citation UI.

F-15 · Raji is a shipped, regionally-authored precedent for the exact Ch 72 design, and nothing cites it

COVERED (confirming) · spec-doc · Wave 2 · MEDIUM

Raji: An Ancient Epic (Nodding Heads Games, Indian-authored) adapts Hindu and Balinese tradition with

the deities honored and narrating while asuras and rakshasas are the fought enemy class. That

is our Ch 72 shape line for line — CVD §17.7 on Ch 72: *"Deities honored; asuras and rakshasas are the

fought adversaries."* An Indian studio adapting its own tradition made the same call. This is worth

recording in the Ch 72 art bible's citation appendix, because the strongest available answer to "is

this respectful" is that the tradition's own authors shipped it. **No change needed — this is a

finding that our design is right and should stop being re-litigated timidly.**

F-16 · The costume rule licenses the lane but must not be read as licensing unsourced Panel A

WEAKER (charter amendment) · spec-doc · Wave 2 · MEDIUM-HIGH

REALM_DESIGN_PROGRAM.md §3.3: *"Lane G is permitted on Panel A only where the realm is wholly

invented. Atlantis and the fairy realm qualify."* Correct as a lane rule; dangerous as a sourcing

rule. Panel A's own definition is *"What the source tradition itself says the place looks like"* with

a mandatory citation line where *"an empty one means the option is not an accurate depiction at all

but a guess with good lighting."* Atlantis HAS a detailed source — Critias's concentric rings, the

temple of Poseidon, orichalcum, the hot and cold springs — so a raw-generated Atlantis Panel A fails

Panel A even though its lane is legal.

Reuse-first fix: one clause in §3.3 — "generation_class governs the LANE, never the SOURCING;

the Panel A citation line binds on all eleven rows." read_boards.py's citation-resolution tooth (a

charter deliverable, currently ABSENT — verified) already implements the check; it just must not

be scoped to exclude invented rows.

F-17 · The lens-gap doc's own wiring obligation is undone; the QA join will keep reporting ROSTER-INCOMPLETE

DESIGNED-UNBUILT · schema-data · zero-token, now · MEDIUM

REALM_ANALYSIS_LENS_GAPS_2026-07-27.md §8 requires minting the R-A1..R-A20 aspect rows into

docs/DESIGN_GAP_REGISTER.md's ASPECT×LENS fold. Verified: that file is 28,142 chars and contains

realm ×0, Xibalba/Naraka/Annwn/Duat ×0, R-A ×1, aspect ×0. So no realm aspect row exists,

and under the QA program's aspect_id join contract the realm rubrics report ROSTER-INCOMPLETE

indefinitely — including R-A20, which this document is the producer for. Nothing this lens returns can

land in the join until the rows exist.

Reuse-first fix: mint R-A2, R-A9, R-A11, R-A15, R-A20 with this lens named as their teacher, plus

the F-1 cell-14 split. Zero model cost.

F-18 · Care-register heading vocabulary is unbounded, and three nodes carry no care section at all

WEAKER · spec-doc · Wave 2 · MEDIUM

Across all 79 spine nodes the ## Protected-chapter… heading uses at least four different section

titles ("Protected-chapter / living-culture framing", "Protected-chapter combat framing",

"Protected-chapter framing", "Elevated-care and protected-chapter framing") and at least ten distinct

tier lexemes (standard care, standard tier, elevated care, elevated care in practice,

living-religion care, maximum tier, MAXIMUM care, OPEN care, `standard tier with elevated-care

sub-nodes, standard tier with elevated reveal-care). CH_70` carries no tier word at all — the

only road node without one. CH_39 and CH_40 have no protected-chapter section at all

(positive-controlled: the same regex returns headings for 76 of 79 nodes; both files do contain "17.1"

in body text, so the section is absent, not the subject). And **CH_29 and CH_42 — two of CVD's

three named maximum-depth chapters — carry no care-tier word in their heading either**, reading only

"Protected-chapter combat framing (ruling 5)."

Reuse-first fix: the heading is prose and should stay prose; the *value* moves to the field set at

§3, where a tooth can read it. Do not standardise 79 headings by hand.

F-19 · Okami's myth-literacy ladder is already our ruled L2/L3 contract — with one extension worth taking

COVERED (confirming) + one WEAKER extension · spec-doc · Wave 2 · MEDIUM

Okami rewards Shinto literacy without ever requiring it, which is exactly CVD §17.9's ruled split:

*"The surface track is self-contained. The action-adventure RPG arc reads complete on the surface track

without engaging the investigatable-depth or canonical-truth tracks."* COVERED; no change.

The extension nothing on the roster carries: Okami makes **the tradition's own visual grammar the core

verb** (the Celestial Brush). Our nearest instance is the charter's Ch 66 exemplar signature law

INSCRIPTION BINDS MATTER, which makes the game's own inscription system the realm's physics. That is

the right instance and it should be the road's showcase for the harmony doctrine, because it is

the one place where the tradition's grammar, the physics delta and the player's verb are the same

object.

F-20 · No region page, no zone rows, and 9-of-25 field fill on every realm region row

DESIGNED-UNBUILT (known) · schema-data · post-slice (P2) · MEDIUM

Verified independently: fairy_realm 11/25, atlantis/hades/xibalba/naraka_meru/annwn/

yggdrasil all 9/25; every environment/terrain/density/weather/movement/permanence column empty;

page_status = dimensional; protected_terrain_ref empty on all. Already recorded in

REALM_ANALYSIS_CANON_INVENTORY_2026-07-27.md, and the realm build is post-slice by the standing law

in docs/PRE_5090_BUILD_PLAN.md. Reported as DESIGNED-UNBUILT so no later wave mints it as MISSING.

Care-specific note: because care_tier on the Zone Catalog is *"derived from the protected flag,"*

and protected_terrain_ref is empty on every realm row, the ratified derivation would emit

standard for Xibalba and Naraka and Meru. §1.2's explicit values are what stop that.

F-21 · The charter's art pipeline is wholly unbuilt (baseline for wave planning)

DESIGNED-UNBUILT · P5-factory · Wave 2+ · LOW (informational)

Verified absent: harness/read_boards.py, docs/art_boards/, docs/art_bibles/,

harness/check_realm_ruleset.py, registries/T0_Realm_Ruleset_Registry. Present and available as the

declared reuse rail: harness/read_captures.py. Recorded so this lens's outputs are understood as

inputs to a mint, not as corrections to a built thing.

---

PART 3 — SCHEMA-DESIGNER CONSUMABLE: the care block for T0_Realm_Ruleset_Registry

Drop-in field set for the Wave-2 schema mint. Nine fields, four of them new, all closed-vocabulary.

Placement: a CARE AND PROVENANCE block adjacent to the charter's §2.3 IDENTITY AND PROVENANCE family;

the ART-DIRECTION HOOKS family reads it, per the charter's rule that the ruleset row is upstream of

the art wave.

Two teeth this block arms, both built on rails that already exist:

• T-RLM-CARE — extend harness/check_care_scope.py's HOMES shape, do not clone it. That gate's

own docstring states the principle exactly: *"A firewall recorded in only one place rots when the

others drift."* The road carries five distinct fringe firewalls — Atlantis/hyperdiffusionism

(Ch 66), Hyperborea-Thule/Ariosophy (Ch 67), Lemuria-Mu-Kumari-Kandam (Ch 68), Agartha (Ch 69),

von-List-Armanen/völkisch (Ch 74) — plus the Vril-Society firewall recurring at Ch 67/68/69/74, and

each currently lives in exactly one place: its spine node's Anti-fringe section. Register each

with 2-3 homes and the presence gate does the rest. Fixed-string scan, never an escaped pattern

(the repo's own recorded lesson).

• The board-readiness tooth — generation_class + source_corpus_state are both required

non-empty before a realm's first board is generated. unsettled is treated as living;

absent + living FAILs with the RCM acquisition block named as the unblocker.

Positive-control fixture required before the block is ratified (mirroring the charter's own

V-adequacy probe reasoning): author one fixture realm row with generation_class = living and

source_corpus_state = absent and show the board-readiness tooth going red, and one with a Panel A

Lane-G image on a living row and show the costume tooth going red. A rule never mutated is not armed.

ENGINE vs CANON, declared per the standing order: the nine fields, their enums, the two teeth and

their fixtures are ENGINE and transfer whole to game #2. The eleven rows' *values* in §1.2 are

CANON and do not.

---

PART 4 — CONTESTED / DIRECTOR CALLS

• C-1 · Are Ch 68 and Ch 69 tighter than the doctrine requires? F-06 shows the audit-only gradient

is inverted against practitioner-community size. The CVD licenses deities as usable characters. But

Ch 69's audit-only rule has a second, non-care reason — the Architect must preside at the Hall

without a deity as the boss, and Apep carries the fight so the Architect never lifts a hand. That is

a reveal-and-structure reason, not timidity, and this lens does not recommend loosening it. Ch 68

has no comparable structural reason on the record. **Recommendation: hold Ch 69's instantiation rule;

ask at Wave 2 whether Ch 68's twenty-two audit-only clauses are doing structural work or inherited

caution.** Either way F-05 stands independently — the *depiction* bound is loosened at both nodes

regardless of what instantiation does.

• C-2 · care_band for the Duat. §1.2 rules elevated. The live corpus disagrees with itself:

spine heading "elevated Egyptian duties" vs BE_0069.care_tier = standard. The charter reads

elevated. This lens rules elevated and flags that one of the two live cells must change — a

director-level apply, not a subagent one (standing lesson on user-authorized canon edits).

• C-3 · Ch 72 de-escalation from MAXIMUM to elevated. Reasoned in §1.5 and F-09. It reads like a

reduction of care and is not one — every commitment survives verbatim. But it touches a

self-declared tier on the road's most sensitive tradition, so it is surfaced rather than applied

silently. Recommendation: apply, with the §1.5 reasoning quoted into the change record.

• C-4 · Whether hard_line_status should gain a fifth value (e.g. TRADITION_DEPICTED) instead of

a new generation_class field. Rejected here for the same reason the charter rejected overloading

care_tier: a column asked to mean two things. Recorded so it is not re-opened.

• C-5 · Ch 7 South India at standard. F-09's inversion could be read as "Ch 7 is under-scoped"

rather than "Ch 72 is inflated." Both may be true. Ch 7 is outside this lens's realm scope and

routes to the earthly-care pass; flagged so it is not lost.

---

Verdict tally

COVERED 2 (F-15, F-19 confirming halves) · DESIGNED-UNBUILT 4 (F-08, F-11, F-20, F-21) ·

WEAKER 9 (F-01, F-03, F-04, F-06, F-07, F-09, F-14, F-16, F-18) · MISSING 5 (F-02, F-05, F-10, F-12,

F-17) · DIVERGENT-BY-DESIGN 1 (F-13). Total 21 findings.

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