supernatural/REALM_DESIGN_PROGRAM.md
CANON SUBORDINATION — this document is PROPOSAL-TIER: it serves canon and never outranks it.
The canon: the CVD · the T1 foundation docs · the T0 registries (registries/) · the spine
(docs/spine/CH_*.md) · the region pages (_source/02_Tier_2_Region_Pages/). Authority order:docs/DOC_MAP.md§ 0.
Canon served (HAND-WIDENED, then scanned from this document's own citations — widen it by hand where it is thin):
docs/spine/CH_66.md..docs/spine/CH_74.md(the Realm Road it rules) ·docs/spine/CH_75.md..docs/spine/CH_77.md(the endgame fairy realm) · CVD §19.8 (the realm roster) · CVD §17.7 ·CVD §6·CVD §17·CVD §19·T1_Audio_Spec·T1_Personal_Dimension_Spec·T1_Region_Page_Template· (+4 more tier) ·T0_Chapter_Index·T0_Environment_Grammar_Registry·T0_Hard_Lines·T0_Realm_Ruleset_Registry· (+2 more t0)
READ THAT CANON FIRST — open it and derive from it before you build anything from this document.
If this document disagrees with canon, CANON WINS and this document is the defect — fix the
document, never the canon. Nothing here is applied until it is ratified into canon.
SUPERSEDED IN PART — the build order only (Josh RULED 2026-07-27, DECISIONS fifth sitting). This
charter was synthesized against Josh's morning realm directive of the same date, which ordered the fairy
realm "last, after all nine mythological realms," with the Prologue and Ch 1 last of all. The fifth
sitting supersedes that ORDERING: the vertical slice is PROLOGUE→CH 13, built Ch 2-13 first and then
Prologue + Ch 1, and the fairy realm designs and builds ONCE, at slice-end — after eleven earthly
chapters have matured the concept-art pipeline, which is Josh's stated reason. The nine mythological
realms (Ch 66-74) stay in the post-slice factory era, and the endgame fairy-realm return (Ch 75-77 at
the sovereign ceiling) EXTENDS the slice's build rather than preceding it — the art bible and the
sovereign kit get built once and consumed twice. Everything else in this charter stands: the doctrine,
the roster axis, the pipeline, the lens set, the freeze-ordered lens priorities, and the Josh picks are
untouched. Where a section below states the old order as a live directive, this banner wins. Authority:
docs/spine/DECISIONS_PENDING_JOSH.md fifth sitting; collision resolution recorded indocs/pipeline_review/SLICE_SCOPE_COST_2026-07-27.md.Synthesis of four read-only analyses, 2026-07-27, against Josh's realm directive of the same date.
Register per CVD §17.14. This charter authors no world canon. It is proposal-tier throughout: it
proposes a roster axis, a doctrine, a schema concept, a pipeline, a lens set, an execution plan, a
short list of genuine Josh calls, and the wiring that keeps none of it orphaned. Every schema, gate,
registry and tier-doc edit named here mints later under its own fresh-context critic and the full
gate ritual. Nothing below is a licence to write into canon without that pass.
Josh's terms: the mythological realms and the fairy realm are the grandest displays of the artistic
capability and the AI pipeline engines — complex, well structured and architected, with physics
blending perfectly and harmoniously with metaphysical lore, magic, and superstition and cultural
lore; epic realms that strongly build intensity toward the end-arc fairy realm; everything beautiful
and rich; art-concept generation led by the director with choices presented to Josh (the most
accurate depiction · the recommended enhanced version for playability and fun · surrounding
recommendations for story, side quests, easter eggs, exploration, adventure, action, nearby important
zones); the fairy realm built last after all mythological realms, with the Prologue and Ch 1
game-build last of all; everything automated by deep-dive fan-outs, alternatives analyses, and
adversarial critics across all lenses; enhancement licence granted on spine, tier docs and vision
where the program needs it.
Restated as a machine, that is five deliverables and one gate:
blank. Section 1.
physics that a deterministic tooth can read. Section 2.
with adversarial critics before Josh ever sees a board. Section 3.
ruleset schema authored with no teacher of realm rulesets is the failure the lens roster exists to
prevent. Section 4.
Prologue and Ch 1 last of all. Section 5. **Re-pointed by the 2026-07-27 fifth sitting (see the banner):
"last" now means at SLICE-END, after Ch 2-13 — not after the nine mythological realms.**
and never as bare questions. Section 6.
nine boss rows with care and reveal disciplines attached, a ratified slate and order (D-RR, CVD
§19.8), and constitutional rulings like the Louhi rule at CVD §17.1. What they lack is a WORLD
(geography, region rows, environment data) and an ART LAYER (theme, creature, voice, sfx rows).
The program is therefore ADDITIVE, not corrective. That single fact sets its risk profile.
that vril alters real physical law rather than suspending it, that vril effects and real physical
law compose emergently, and that the fairy realm's vril and the human world's vril are the same
substrate with density producing the dimensional separation. The realms are one physics at
different densities; the program makes that claim testable.
scenes that show the fairy realm's physics as ORDINARY, felt from inside by a being who has never
known anything else. That scene cannot be authored until the nine partial views exist to be the
thing it is ordinary relative to. Josh's build-last ruling is the only order in which the origin
pair can be written correctly, and the doctrine's refraction law says why.
---
Maturity grades carry the inventory's meanings: A build-ready; B designed and reveal-clean at the
spine but missing geography, substrate or registry population; C named and ruled with a thin or
displaced design body; D stub or blank.
Three intensity readings are given per realm because they are three different instruments and the
program's central intensity finding is that they must not be tuned as one:
HL_0049. It ALTERNATES by hard line; it does not rise.
realms of the dead, climbs out, and reaches the ceiling at Ch 75.
build-toward-the-fairy-realm requirement without touching a hard line.
One reading convention for the CARE lines below, so no row is mistaken for a column value. The
ratified care_tier enum has exactly two values, standard and elevated, per T1_Region_Page_Template.
Words like LIVING and MAXIMUM in the rows below are the analyses' descriptive readings of care
WEIGHT and of whether a tradition has practitioners today — they are prose, not enum values, and
this charter asserts none of them into that column. THE GOVERNING TABLE (integrated 2026-07-28;
ruled at the fourteenth sitting, ruling 9 — CVD §6.2 UNTOUCHED): per-realm care tiers and duties
are settled by docs/pipeline_review/REALM_CARE_TIERING_2026-07-28.md, the L-R3 deliverable
rebuilt under the 14 director adjudications (2-value care_tier + a separate DUTY column; the
6-value care_band untouched; CH_66-69 standard-tier/elevated-duty; Duat standard; Hades elevated;
Ch-72 RULED ELEVATED and the Dream Tradition row RULED INVENT at the seventeenth sitting,
2026-07-28) — where this charter's §1 prose and that table disagree, the table wins. The legality axis the prose readings are really
pointing at is the separate proposed generation_class field at §3.3, and the authority that
settles each realm's
value is L-R3's named output.
Road doctrine binding every node: the realms are the decompression chamber between the human world's
vril-depleted atmosphere and the fairy realm's ceiling; no House of Velheim operative appears
anywhere on the road; each node carries exactly one crossing-function; nodes were repurposed, never
added, so the 77-plus-2 count holds.
Crossing-function origin-recognition. Boss BE_0066 the Orichalcum Warden, a real embodied fought
colossus under ratified D-EMB-66; defeat and recognition are the same beat. Intensity valley
(corrected 2026-07-28 against the live registry — the eighth-sitting Pattern A landing set
CH_66 valley; this charter's earlier "standard" predated it), mode astral. Vril high to anchor. Proposed demand 1. Care: literary invention, and simultaneously
an ELEVATED anti-fringe duty (Donnelly, Blavatsky root-races, Lemuria and Mu, Hancock, documented
hyperdiffusionist racism, corrective authorial-only). Named blank: three sites, the thinnest count
on the road; region row atlantis 9 of 25 fields; asset registries unpopulated.
physics via the Sampo contest. Louhi IS the ultimate boss, BE_0067, in her iron-taloned eagle form
as an honored sovereign matriarch who is RIGHT to fight and withdraws still sovereign — CVD §17.1
constitutional text. Iku-Turso the mid-fight escalation. The Sampo breaks by design, so the prize
is the passage. Intensity PEAK, mode direct. Vril high. Proposed demand 2. Care: national-literary
with a real living-tradition edge (Finnish national epic; no Sámi conflation, no Lapland-witch
trope; Thule-Gesellschaft and Ariosophy firewalled). Named blank: region row and asset registries;
the runic-song tradition is referenced-never-reproduced, which constrains the audio program in a
way no audio doc records.
Crossing-function primordial listening. BE_0068_S1 a real struck-and-purged leviathan-scale
defensive host under D-EMB-68 plus the J10 loosening; Tiamat, Apsu, Marduk and Nammu audit-only,
never bodied. The Enuma Elish's slay-Tiamat combat is deliberately inverted to listening and held
as the game's own respectful re-reading. Intensity standard, mode environmental. Vril high.
Proposed demand 3. Care: Mesopotamian, no living practitioners; Lemuria/Mu/Kumari-Kandam firewall
is the realm-name's whole reason for existing; the Rainbow Serpent referenced-never-reproduced.
Named blank plus a defect: the Body prose is a KEEP-VERBATIM block still reading to the pre-J10
non-lethal framing while the typed layer reads to the fought-and-broken host — a narrative-pass
debt sitting inside a hard-line node.
C data. Crossing-function judgment. Dual structure: Movement 1 the twelve gated hours defending the
barque against BE_0069_APEP, a genuinely-fought separate ultimate boss (care-clean, the enemy the
tradition unmade itself, HL_0121); Movement 2 the Hall of Two Truths, where the Architect
(CHAR_0003) reaches his terminus in a materialized-argument Final Negotiation, the fifty-two-chapter
payoff of the Ch-17 first contact. Apep carries the fight so the Architect never lifts a hand.
Ma'at, Anubis, Thoth, Ammit, Osiris and the forty-two assessors audit-only. Intensity PEAK, mode
puzzle_routed. Vril anchor. Proposed demand 4. Care: ELEVATED (ancestral with living descendants
and active revival). Named blank and the heaviest stale load in the corpus: four live artifacts
still assert the retired World-Tree structure at Ch 69 (WGM §8 heading, T0_Chapter_Index
region_primary, VS_CH69_001's name, the CVD's own chapter-anchor gloss), plus fourteen deferred
Thread-5 refs in T1_Threads_22.
descent-and-return. BE_0070 Cerberus at the Return-Gate, the honored doorkeeper passed and never
killed, three routes (force, song, token); the descent inward is unopposed and the RETURN is the
trial; combat routes through the restless dead of the egress, never the honored shades. Deities
real and sober. Intensity standard and the spine states the trough is DELIBERATE. Vril LOW at every
site — the deliberate inverse of the coming fairy abundance. Proposed demand 5. Care: classical
Greek, standard. Named blank: no WGM geography; RCM substrate gap (its acquisition list is still
the displaced pre-v3 fairy corpus, and Homer, Hesiod and Virgil are not yet listed); enum drift
(spine says choice-routed, registry says vril_management). Six sites, the richest set on the road.
the ballgame, and canon names it the flagship of the minigame-as-boss category. BE_0071 the two
supreme Lords; the winning verb is CLEVERNESS OVER FORCE. Explicit reveal firewall: every mirror or
judgment image here is myth-geography only and is never wired to the Ch 76-77 thesis
mirror-lattice. Intensity PEAK, mode puzzle_routed. Vril LOW — the floor. Proposed demand 6. Care:
LIVING, the sharpest tier (living Maya communities; the Popol Vuh a living cultural possession,
Guatemalan intangible heritage since 2012; the Lords grave and order-keeping, diminished never
exterminated). Named blank: no WGM geography; RCM substrate gap carrying a blunt do-not-use
instruction on the displaced body; enum drift (spine contest-routed, registry puzzle_routed).
Crossing-function the moral-cosmography ladder, ratified over Chinese Diyu (D-RR.1). BE_0072 Vritra
the Enclosing Dragon; the winning verb is RELEASE, each shattered coil bursting a pent river she
rides upward. A Josh-ruled offered fork: the Rigveda-faithful death-blow, or break the last hold
and spare him un-slain, seeding the Ch-75 Last Coil; neither branch coded evil, and either outcome
looses the waters so the fight never dead-ends. Intensity PEAK, mode vril_management. Vril low at
entry, moderate at the weighing place, ANCHOR at the world axis, exiting rising — canon's own
climb-out. Proposed demand 7. Care: MAXIMUM (living Hindu, Buddhist and Jain cosmology; deities
honored and never villainized; the protagonist echoes the Indra-role, honoring and never
supplanting). Named blank: no WGM geography; RCM substrate gap; and per the lens audit this is the
highest-care realm on the road that has had the LEAST attention.
plus, the thinnest substrate on the road. Crossing-function the Otherworld's hospitality-and-peril.
BE_0073 Hafgan the Rival King at the ford, a king defending his own claim; the winning verb is
RESTRAINT, won by one clean committed blow and the refusal of the resurrecting second. Arawn is the
honored host and by the end a friend. The load-bearing distinction: Annwn is honored as its OWN
Welsh tradition and is never conflated with the game's fairy realm, which merely echoes it.
Intensity standard. Vril moderate to high. Proposed demand 8. Care: CONTESTED across the analyses
(see contradiction C-3) — the inventory reads the spine's tier as open, the art pipeline places
Annwn in the historical column while noting it sits closer to elevated, and the lens audit places
it in the LIVING tier with practitioners today; generation_class therefore reads unsettled, which
the §3.3 gate treats as living until L-R3 settles it. Named blank: no WGM geography; RCM substrate
gap whose body is displaced Realm-War material; and Annwn is MISSING ENTIRELY from the RCM's
Acquisition Priority Master List — POSITIVE-CONTROLLED AND CONFIRMED, and the control is recorded
because an absence-claim without a search that can match is this repo's most-repeated defect. The
list carries a per-chapter block for Ch 66, 67, 68, 69, 70, 71, 72 and then jumps straight to Ch
74; a fixed-string search for the Ch 73 heading in that list returns nothing while the SAME search
shape hits "Ch 72 Naraka and Meru" and "Ch 74 Yggdrasil, the World-Tree Crossing" in the same list,
so the matcher demonstrably can match and the zero is real. What DOES exist, and is a different
artifact, is the per-chapter substrate section "Ch 73 — Annwn" further down the RCM — itself
flagged SUBSTRATE GAP, its body the displaced Realm-War unit, and its Tier 1/2/3 entries
cross-listing Ch 70-72's catalogs rather than listing the Four Branches of the Mabinogi. The realm's
real corpus is NAMED in that section's gap note per CVD §19.8 but is not ACQUISITION-LISTED
anywhere, which is precisely the layer Wave 3 repairs.
The ninth and last road node and THE canonical crossing into the fairy realm — the only myth-native
door. Re-based from the patriarch reveal to a STRUCTURAL KEYSTONE (the crossing plus the Thread-5
Tree of Life capstone). BE_0074 Níðhöggr at the root with a six-named serpent-brood add-class,
broken off the root and driven back, never slain; the winning verb is RE-ROOT. It now leaks
NOTHING — no Cassius, no House, no pattern, no thesis, no Grand-Sage hint, not even an
invoke-and-negate. Named care call: Álfheim is NOT the game's fairy realm. Intensity PEAK
(corrected 2026-07-28 against the live registry — the eighth-sitting Pattern A landing set
CH_74 peak with the R3 gate re-scoped in the same commit; this charter's "STANDARD" predated
it). Vril high to anchor. Proposed demand 9. Care: Norse, with the loudest Ariosophy, Nazi-occult and
völkisch firewall on the road, authorial-only, living inclusive Heathenry honored. Named blank: no
WGM geography; RCM substrate gap whose body is displaced Cassius-confrontation material belonging
at Ch 76, with the Eddas cross-listed at the Ch-70 fairy block and not yet listed here.
ELVEN KING at full sovereign capability, the only place in the arc he is playable. The vril CEILING
reference for the whole arc: demonstrated and FELT so the seventy-five-chapter climb back to it
lands, then stripped at the Ch-1 fall. BE_PROLOGUE, the First Rift — not an antagonist and not a
person but a dimensional rupture; verb SEAL, no fail-wall. Show-don't-tell is ABSOLUTE: the King
consults the Grand Sage, whose nature he has known ten thousand years, and says nothing; his
temporal sight gives no word, never as prophecy. Name-forbidden (HL_0099). RULED 2026-07-27,
INTEGRATION-PENDING — the authority is real and Josh-authored (DECISIONS_PENDING_JOSH: "RULED
2026-07-27 (third sitting) — THE AMULET PROVENANCE + THE PROLOGUE/CH-1 EMOTIONAL SPINE"), but it is
not yet canon in the repo's own sense, because docs/proposals/AMULET_PROVENANCE_BRIEF.md states of
itself: "Nothing here enters the spine, the registries, or the tier docs until the integration pass
runs with its own critic and its own reveal-discipline verification." So this program consumes it
as a RULED DIRECTION and never as a landed row, and the named blank below (open staging from the
amulet ruling) is the same fact stated as work. The ruling: the amulet was the mother's, passed at
the birth; she died unprotected in the
catastrophic rift-spawn event days-to-weeks later; the playable run-back; the King's
never-resentment; the first-rift set-pieces stage BEFORE the catastrophe so the ceiling is proved
and never beaten; the mother's death gets no author, which is reveal-safe. Intensity standard.
Vril CEILING. Care: game-canon invention, generation-legal on both board panels. Named blank: RCM
Prologue is 31 lines with no Earth-cultural corpus; WGM §2 entry is six bullets; no vril site row
exists for the fairy realm at the Prologue; region row fairy_realm 11 of 25 fields; open staging
from the amulet ruling.
selections are woven as childhood, not a menu; integrity initializes at zero. BE_01 the Fall, verb
ENDURE, ceiling STRIPPED on impact, no win and no fail-wall. The INVERTED CHANGELING is the design
key: a royal fairy child lost OUT with no swap — a loss that strips rather than a theft that
enriches. The fall reads as rupture and accident, never as curse or omen. Ch 1's end state pins the
Ch-2 start state exactly. Intensity VALLEY. Named blank: RCM declares Ch 1 out of corpus scope by
design, so the childhood has no substrate section at all; WGM §2 entry is seven bullets.
time since the Prologue, the single largest power beat in the game. The King's decades of grief
have CRYSTALLIZED INTO THE LAND; over and through it the realm's own dark mythology has risen and
turned. FOUR RATIFIED CHAMPIONS plus the trajectory-gated Last Coil — D-EMB-75 is titled "author
the four marquee champions to WS4 depth" and DECISIONS_PENDING_JOSH calls it "name the four
Realm-at-War champions," so the word ratified attaches to four, not five. The four: the Unlidded
Eye, the Westering Host under the Griever-at-the-Head (care-critical: terrible but pitiable, a real
grievance owed rest and denied), the Pitch-Court under the Sunless Jarl, and the Cinder-Advance
under the Kindler at the Wall. The fifth entity, the LAST COIL fused into the war, rides the F1
Vritra spare-branch ruling and is TRAJECTORY-CONDITIONAL — it exists only on the branch where Ch 72
spares Vritra, which is why this charter's own Ch 72 entry describes that branch as "seeding the
Ch-75 Last Coil." The live registry carries BE_0075_S1 through S5, so five ENTITIES is defensible
as a count of encounter rows; five RATIFIED CHAMPIONS is not, and the distinction is load-bearing
for any wave that sizes champion art against a ratified slate. Plus the ASSEMBLED TITAN made of her
own unfinished work. The load-bearing guardrail: the war
CONDENSES OUT OF THE REALM'S OWN SICKENED SUBSTRATE, condition-made-weather, never a marched cabal
army and never an invading people (HL_0109 / HL_0114). Intensity PEAK, mode direct. Vril CEILING.
Proposed demand 10. Named blank: substrate gap plus two unlanded re-home debts (the
childhood-familiar realm-return, and the full-legibility inscription payoff with INSC_77H).
the first time (HL_0035), the House and the 5,000-year pattern surfacing at their patriarch, in a
three-phase Social plus Combat plus Moral boss implementing Plato's tripartite soul and the
tyrannical inversion. The §17.1 / HL_0109 / HL_0114 guardrail at its absolute sharpest: evil is a
chosen soul-DIRECTION, never heritable or ethnic; banking is the instrument and never the secret;
Cassius is one endpoint and never history's omnipotent author; sophisticated and never monstrous.
Phase 2: the FATHER encounter, the moral north, three completely distinct path encounters including
canon's hardest fight in the game — grief-driven, never villainous. The father-silence bookend
begins here. Intensity PEAK, mode puzzle_routed. Proposed demand 11 (the ceiling of the ladder).
then the single canonical surfacing of the thesis, released exactly once and carried by the IMAGE —
the mirror-morph the player's own eye completes — never by a caption, a glyph, a spoken line or a
cut. The Grand Sage never speaks (HL_0046). The morph is the protagonist's own gender presentation
(HL_0115). No combat, no fail state, player-paced dwell, authored NEUTRAL with the three endings
colouring it. The waterfall is a canonical vril recharge-source (HL_0112) — the game's first
mechanic returned. Intensity standard, mode astral. Demand: EXEMPT by declaration, the release, not
a rung — and an exempt row in an ORDINAL column needs a declared sentinel or the curve tooth reads
it as zero and fires a false regression against Ch 76's ceiling. The sentinel is the repo's existing
marker, THE EMPTY STRING (the cited-or-empty convention), and T-RLM-CURVE skips empty
demand_ordinal rows explicitly rather than coercing them.
single variance is journal COMPLETENESS. The thesis is not re-declared; the Epilogue lives it
forward. Intensity VALLEY. Maturity B plus.
beat CH58_B10 reading ws.integrity_score, on the truly-good trajectory only. Canon is explicit
that this is the arc's deepest familiar payoff and is NOT the Grand-Sage reveal. Maturity A minus,
the highest grade on the axis, and the template for what typed-and-gated looks like.
Ability-canon only: Dimension Pocket is a Tier 6 Adeptus Minor Ether-Crown passive opening a
personal astral sub-space with 200-plus creature housing, growth-rate-free cultivation and
constraint-free construction, INTEGRITY-VISIBLE across the five bands. T1_Personal_Dimension_Spec
is TWO LINES. The recovered marriage and partnership system (INGEST_GDD §10) has no home and the
gap register names this spec as its natural one. Program relevance: it is the cheapest possible
demonstration of physics-blending-with-lore (a sub-space with no growth-rate and no
terrain-stability constraints is a physics ruleset the PLAYER authors), and it is the only realm
surface whose visual identity is a direct function of the integrity system. Coverage note per the
lens audit: its teachers are ALREADY on the roster in five places, so this row is DESIGNED-UNBUILT,
not un-lensed.
58 and the fairy-realm-is-a-real-place premise; the only contested astral space in canon is Ch 46,
where counter-projection watchers can follow her in. It has no biome, no geography, no site rows,
no visual language and no rules of its own — the connective tissue between every realm on this list
and the least designed thing on it. Thread 16 is the corpus's thinnest thread against a core
mechanic.
realm slot is the game's OWN invented otherworld — the rift-interior register this section's
RIFT INTERIORS row names as the open design space, framed by the ruled 20-rift-dungeon
economy — and is NEVER built against Aboriginal Dreaming substrate (the Dreaming keeps its
full Ch-57 maximum-care weight and its motif-level presence per the Ch-68 Rainbow Serpent
precedent). Josh's calibration is standing doctrine (memory invent-when-needed-fun-calibration:
"our fun and engaging take" — no blatant disrespect AND no lying, both ways; invent when
needed to keep things fun). Art boards for the invented substrate are UNBLOCKED under the
normal choice-board protocol; the care table's row-12 MAY-NOT stays binding for Dreaming
substrate specifically.
into flesh, occupying a vacated substrate-gap and wearing the eroded shape of something the
depletion took. Collect-verb CONTAIN, hazmat-and-exorcism grammar, never tamed or bonded, outside
the 33-legendary and 22-slot firewall by construction. The etymology closes the loop: cassus,
hollow. Reveal-refined — the surface reading is safe from first appearance, the depth reading stays
Layer-3 post-Ch-76. Three forks stand open (admission, in-world name, reveal window). Program
relevance: it is the negative-space counterpart to the realms and the most direct visual expression
of the vril-altered-physics doctrine available outside the realms themselves.
is DEAD — the twelfth sitting ruled TWENTY BOUNDED RIFT DUNGEONS (10 vril + 10 health, the
BotW-shrine register, full arc) plus randomly-spawning POP-UP dungeons (Hollow axis 2), all of
which are inside-the-rift content.] What survives of this row: rifts remain spontaneous quantum
events neither side creates intentionally; the authored STORY beats stay at the MEMBRANE; and
the rift-dungeon interiors' AESTHETIC/ruleset design is still genuinely open space directly
under the directive's remit — the one place a new otherworld register can be authored without
touching a living tradition. The 20-dungeon economy is the content frame that design fills.
Sub-rows worth carrying rather than promoting: Elysium and Tartarus at Ch 70; Patala, the seven
talas, the graded Narakas and Svarga at Ch 72; the six Houses and the ballcourt at Ch 71; the
glass-fort caers at Ch 73; the three wells at Ch 74; Álfheim as an explicitly non-fairy world; and
the Ch-44 metaverse as a digital otherworld entered by astral projection.
The live registry encodes an ALTERNATION, not a rise, and it does so because two hard lines require
it. Both are quoted verbatim from live T0_Hard_Lines rather than paraphrased, because this repo has
already caught and corrected this exact mis-attribution class once — STATUS.md records "HL_0048/0049
mode-variety mis-attribution across Ch 3-7 (both mode-distinctness clauses are HL_0049; HL_0048 is
intensity-topography/no-coasting) — corrected." A charter that briefs Josh on whether to touch these
lines cannot restate them from memory.
consecutive intensity-peaks. No three consecutive valleys. Mode variety per chapter per Pillar 11 —
no chapter coasts on one mode."
five-chapter window contains three or more distinct combat-mode dominant chapters."
So the mode constraint is TWO consecutive, not three, and it carries a second clause — the
five-chapter window must contain three or more distinct dominant modes. Both clauses are HL_0049;
HL_0048 owns intensity topography and the no-coasting rule.
harness/check_chapter_topography.py encodes the two HL_0049 clauses as R1 and R2 and the HL_0048
intensity rule as R3, so the split is machine-visible and not a matter of reading.
The alternation finding survives the correction A FORTIORI: the real constraint is STRICTER than the
charter's earlier paraphrase, so a road that already alternates under a three-consecutive reading
alternates at least as hard under a two-consecutive one. SUPERSEDED IN PART (2026-07-28): the
sentence that followed here read the crossing as standard with the apex at Ch 69 — that was true at
charter-write time and was overtaken the same day by the eighth-sitting Pattern A landing, which set
CH_74 PEAK (the crossing is now the road's apex) and CH_66 valley in the live registry with the R3
gate re-scoped in the same commit. Mechanical novelty also front-loads: school debuts SCH_010 through SCH_013 run Ch 66-67-68-69
and STOP, because ratified decision D-RR.5 has Ch 70-74 showcase and extend existing disciplines
rather than author new schools.
The resolution the program adopts is that felt intensity is DEMAND RELATIVE TO POWER, and the three
instruments are separable:
anchor at Ch 73-74, ceiling at Ch 75. This is already gated by the spine's own
ws.atmospheric_vril_density_modifier seams and must not be contradicted.
directive.
should never be tuned as difficulty.
power collapses and the answer stops being force. Xibalba, where the myth's own heroes win by wit,
is the correct floor for that trough, and canon already put it there.
One signature law per realm plus compounding reprises gives Ch 70-74 real mechanical novelty
without minting schools D-RR.5 ruled out. This is a synthesis finding neither analysis stated on
its own and it is the cleanest available answer to the front-loading problem.
What this resolution does NOT settle is whether the CROSSING should stay standard. See contradiction
C-1 and the Josh brief at 6.2.
---
Proposal-tier. The doctrine's owning authority is T1_Vril_and_Magic_System_Master (§1 already owns
"vril alters real physical law" and §10 owns depletion physics); the machine layer is the schema at
2.3. Nothing here mints until its own critic and the gate ritual run.
A realm is not a place where physics stops. A realm is a place where a small, DECLARED set of
physical couplings is changed, every undeclared coupling runs exactly as it does on Earth, and the
change is derived from that tradition's own cosmology rather than from a designer's taste for the
strange. The metaphysical law of a realm is not an overlay on the physics; it is a TERM IN the
physics — an oath that binds adds a real constraint, a name that has weight adds a real mass, a realm
where the dead walk changes a real collision channel. Harmony is the measurable property that the
metaphysical and the physical are the same event described twice. Weirdness that is not in the
declared contract is a defect, and the machine can see it.
weirdnesses. Each is one culture's partial view of one physics, and the fairy realm is the source
all nine partially express. This is the game's own convergence thesis (fifteen vril-parallels
describing one substrate from fifteen vantages) applied to physical law instead of perception.
Every bend is marked toward_source true or false: a bend moving local physics toward the fairy
realm's native state belongs to the gradient, a bend moving away is a local peculiarity needing a
stronger provenance case. This law is also the design-side justification for the build order.
The baseline is the human world from Ch 2 onward. Everything not named in the diff runs the shipped
simulation unchanged. This is what makes nine realms affordable in a solo build: nine data rows
over one physics, not nine physics.
the source tradition's own primary corpus. A delta without a citation is invention wearing a
realm's name, and the gate rejects it. This is the cited-or-cut firewall applied to physics, and it
is the anti-slop mechanism with the most teeth: generic fantasy weirdness cannot produce a citation.
physical quantity it writes, and that name must resolve to something real in the build — a gameplay
attribute, a worldstate key, an environment-grammar property, or a named subsystem. Prose does not
satisfy it. If the answer to "what does this law physically do" is "it glows," the law is an overlay
and the row fails. This single requirement converts "blends harmoniously" from a taste judgement
into an arithmetic one.
and familiars, fauna, props. A rule that only affects the player is a gimmick with a costume on.
Scope defaults to all_actors and any narrower scope declares its justification IN DATA, so the
exception is visible rather than silent.
the same realm, without a line of tutorial text. Each realm's threshold zone is its tutorial by
construction. This is the natural-voice doctrine applied to physics: the player learns the local law
by watching the world obey it.
regionally, or run at different rates; it never reverses the player's action order). The input
contract (what the player pressed is what the character attempted; realms change what an attempt
achieves, never what it was). Persistence of self (build, mastery tiers, gear, imprints, bonds and
record cross every threshold intact; a realm may close a usual answer, never confiscate the build).
Readability primitives (a telegraph is a telegraph; the counter-class vocabulary keeps its meaning;
realms restyle these, they do not invert them). And the CVD §17 hard-line floor, unchanged and not
this program's to touch.
Rising vril must never read as a power ramp. The existing difficulty doctrine already proves the
pattern — the density dial changes four world-demand levers and nothing on the enemy sheet.
Proposal-tier concept, not a mint. Recommended home: a new T0_Realm_Ruleset_Registry as the machine
layer with T1_Vril as the binding authority. Extending T0_Environment_Grammar_Registry was considered
and rejected — that registry is PER-FEATURE (a lava vent, a waterfall) and this is PER-WORLD, and
overloading it would make its key mean two things. Prose-only in the tier doc was rejected because no
tooth can read prose. Engine-versus-canon declaration, per the reuse standing order: the SCHEMA, the
gate, the conformance tests and the critic rubrics are ENGINE and transfer whole to game #2; the ROWS
are CANON and do not.
Eleven rows at completion: RLM_0000 the human-world baseline, RLM_0001 through RLM_0009 the road in
chapter order, RLM_0010 the fairy realm. A twelfth row for the Personal Dimension is proposed at 2.5.
CONFIRMED AGAINST THE HOLLOW EXPANSION, NOT RE-LITIGATED (integration record
docs/pipeline_review/HOLLOW_EXPANSION_INTEGRATION_2026-07-27.md §4.3). The eleven-row closed roster
and the closed physics vocabulary V both stand. The Hollow admission brief's locus row was the one
place its recommendation collided with this charter, and the collision resolves in this charter's
favour: the rift and deadzone LOCUS rows belong in T0_Environment_Grammar_Registry as per-feature
rows, never in T0_Realm_Ruleset_Registry. The confirmation is recorded here so the collision is not
re-opened by a later wave reading the admission brief alone.
Field families:
tradition_primary_corpus as a foreign key into T0_Research_Corpus_Registry so the provenance tooth
has a real target rather than a prose string; derivation, one of baseline, delta or source, with
exactly one baseline and exactly one source row asserted by the gate.
fluid, acoustic — each written in the existing environment-grammar micro-grammar so one parser
serves both registries. Then HOLDS and BENDS, two arrays over one closed physics vocabulary V,
where every term of V appears in exactly one of the two: a term in neither is an unanswered
question and fails, a term in both is a contradiction and fails. This partition is the single most
important property in the schema, because it turns "which earthly rules hold and which bend" from
an authoring intention into a machine-checkable completeness claim RELATIVE TO V. Plus toward_source
per bend.
V IS ITSELF A MINT, AND THE CLAIM CARRIES ITS BOUND. What T-RLM-PARTITION proves is completeness
over V — set union equals V, set intersection empty — and NOT completeness over physics. A realm
that bends something real but absent from V passes green, which is exactly the
a-search-that-cannot-match-reports-zero failure class this repo has been burned by four times and
which §2.3 invokes by name two paragraphs down for RLM_0000. The RLM_0000 control tests the DIFF
MACHINERY (does the conformance test fail on a mutated baseline); it does not test whether V is
WIDE ENOUGH. So three things are stated rather than assumed:
definition alongside the seven profile fields, minted in Wave 2 under its own fresh-context critic
exactly as the schema is, and versioned with it. It is NOT authored ad hoc per realm row, and no
realm row may introduce a term that is not in V — that is what makes the partition checkable at
all.
property-key set, which is the parser this schema already reuses. That set is TINY today: the
live T0_Environment_Grammar_Registry carries two seeded rows and four distinct property keys
across them (conductivity, mass, structural_integrity, thermal_state). Four terms cannot describe
nine cosmologies' worth of altered law, so V is a real authoring task in Wave 2 — expanded to
span the seven profile families (gravity, light, time, matter, thermodynamic, fluid, acoustic) at
the granularity the realms actually bend — and not a column that already exists under another
name. Anyone reading "closed vocabulary" as "already built" would ship the completeness claim on
a four-term list.
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. A vocabulary that has
never been probed for gaps is not armed, by exactly the reasoning this charter already applies to
mutations. Each blind spot the probe finds widens V; the probe re-runs until the remaining
outside-V bends are ones the program can defend leaving out. The probe's residue is recorded in
the schema doc, so the bound is visible rather than implied.
And the honest scope of the pre-hardware period: the vocabulary-free backstop is
Realm.NoUndeclaredDrift, which measures every NON-declared quantity in-sim against the baseline
realm and needs no vocabulary at all — but it is the RUNTIME half, and §5.3 ships the data half now
and rides the sim half with the build. So through every doctrine ratification, every schema mint,
every ruleset row authored at Wave 2 and every art bible, the completeness property rests on V
ALONE. That is survivable because V is probed rather than assumed, and it is the reason the probe
is a gate item rather than a nice-to-have.
for within-realm variation (Ch 72 runs low at the naga depths and anchor at the world axis in the
same chapter); recharge_law, depletion_law, ability_availability_delta.
signature (exactly one per realm); tradition_citation; trigger written in the ACTIVE
condition-expression grammar and nothing else so it is executable against harness/condition_expr.py;
physical_expression, mandatory and resolving to a real quantity (Law 3 armed in schema); scope
defaulting to all_actors; THE D-RR.5 BOUND recorded as a schema constraint rather than left to
authorial care — a signature law never grants a mastery gate and never confers school membership,
because a law that does either is a discipline wearing a realm's name and breaks the ratified
decision that Ch 70-74 "showcase and extend existing disciplines and author no new schools — the
13-school count and the 120-gate mastery scaffold are unchanged." T-RLM-COMPOUND arms it; taught_at as a site reference in the same chapter; EXPLOIT, how a clever
player turns the law into a tool rather than only suffering it, which is what keeps the realms from
reading as corridors of rules; reprise_of for compounding; violation_consequence, never a
progression stop.
(L1 action, L2 completionist, L3 connection), mechanical_form, honoring_reward, breaking_cost,
alternate_route (mandatory and non-empty so it chains into the existing no-dead-end gate rather
than duplicating it), care_tier, care_ref (WORDING CORRECTED 2026-07-28 per adjudication 10: the
realm ruleset's care column adopts the ratified 2-value care_tier — standard | elevated — with
the duty axis a separate column per the ratified care table; the gate-enforced 6-value care_band
on other registries stays untouched and is never this schema's key).
sovereign_distance.
silhouette_law, impossible_architecture_budget (a declared cap, because a budget makes
impossibility COMPOSED rather than random), signature_phenomenon, and the three concept-choice slots
the directive requires. Those slots are not free-form: the accurate option is a rendering of the
provenance columns, the enriched option is the delta budget spent for playability, and the
surrounding option is the taboo and exploit list turned into quests, easter eggs and adjacent
zones. The ruleset row is therefore UPSTREAM of the art wave, and the art wave inherits its
citations.
The baseline row is not optional. RLM_0000 with an empty bends array looks like a wasted row and is
the most load-bearing one in the registry: without it the conformance test has nothing to diff
against and would pass on any input — the exact "a search that cannot match reports zero" failure the
repo has been burned by four times. The baseline row is the positive control and the gate's must-fire
fixture is a mutated copy of it.
A taboo is the most efficient carrier the program has: simultaneously an action-layer constraint, a
cultural teaching, and a cross-realm thread.
return from Hades is a facing-cone check on a follower actor. Do not eat the otherworld's food in
Annwn is a consumable that binds a time cost. Do not sit when the Lords of Xibalba offer the bench
is a perception check against a false actor with no thermal or vril signature.
how it is honored. This is where the care doctrine's "care means weight and authenticity, never
omission" actually gets paid.
FELT, never declared. No narrator says cultures converge here.
Four rules keep taboos playable. A taboo never hard-gates progression (alternate_route mandatory,
chaining into check_no_dead_end.py). A taboo is opportunity-bearing in both directions (honoring pays
something concrete, breaking buys something concrete at a real price; a prohibition with only a
downside is a wall, and players route around walls rather than feeling them). A taboo drawn from a
living tradition carries a care_tier and the rename-or-reframe tool, never removal — and what the
care tier governs is not WHETHER the tradition's own antagonists are fought but HOW the living
community around it is portrayed. And a taboo never encodes a real-world fringe claim; three of the
nine realms carry a demonstrated fringe-uptake drift class (Atlantis and hyperdiffusionism, the
Apsu's drowned continents, Norse material and völkisch appropriation), and WGM §15.2's residual
Hyperborea-Thule reference that had to be resolved at v2.3 proves it is demonstrated rather than
hypothetical.
The strongest single instance, carried forward as the program's showcase: ATLANTIS'S OWN
METAPHYSICAL LAW CAN BE ITS FIREWALL. The realm refuses a claimant — the being who arrives asserting
descent from it, inheritance of it, or the right to speak for it is not recognized and the threshold
does not open. That is exactly the integrity-as-recognition frame the spine already rules for Ch 66,
and it converts an anti-hyperdiffusionist duty of care from a list of forbidden phrases into a
playable law the player experiences from the inside. Care discipline that becomes a mechanic is care
discipline that survives the build.
Candidates for the authoring wave, not ratified rows. Each is drawn from the tradition's own
cosmology and from what the spine already rules.
mastery-gated hand becomes true of the material it is carved on, which makes the game's own
inscription system the realm's physics. Bends matter, acoustics (concentric rings as a real annular
resonator), light. Exploit: carve to make a barrier light.
so the puzzle is research rather than guessing, mapping onto the archive Ch 67 already is. Bends
thermodynamics (heat leaves the body at a managed rate), light (polar night makes illumination a
resource), time (the realm runs on rates rather than states, which is what a Sampo that grinds
actually is). Taboo: singing what you do not know.
volume that makes permanent terrain — and the chapter's own crossing-function rules that the sea is
not slain, so the law is LEARNED AND THEN DECLINED. A metaphysical law used as a moral instrument
without a word of sermon.
is admissible and a lie has weight. Gravity is locally replaced by the balance: weight becomes moral
and the scale is a real instrument reading the accumulated record as a mass term. This is the
showcase row for the whole doctrine — the integrity system, abstract for the entire game, becomes
literally physical for one chapter.
follower first, and exploitable because some things follow only while unseen. Bends matter: the dead
have no mass, so they cannot be pushed or fought. That is the honest answer to the recorded
"no-enemy mythological realms" slop finding — the realm is not enemy-free, it is
enemy-INAPPLICABLE, and it says so mechanically.
check replaces a threat check, countered by the player's existing sight and thermal kit rather than
a new gimmick. Bends light (the torch and cigar economy is a literal burn-rate) and time (the houses
are timed). This is where the trough pays off: wit beats power because power is gone.
fluid event with permanent terrain consequence, which is also the climb-out canon rules. Bends
gravity (Meru is an axis; gravity points inward and the climb is a real reorientation across
concentric shells) and time (the yuga rungs run at different rates, which the existing clock tooth
already measures as dilation).
A famous, real, playable myth mechanic the chapter already carries. Bends time (the otherworld
year, dilated and deliberately unmeasurable from inside, which the natural-voice doctrine supports
because the clock is diegetic or absent), matter (the cauldron's plenty is real and unremovable),
light (perpetual soft day with no night cue, which is how the time loss actually lands).
enforced later. That is the myth's own norn-thread and also, precisely, the spine's existing
mark_write forward_track mechanism with its consumer-continuity gate. The tradition's law and the
repo's machinery are the same object — the strongest single harmony instance available, and it
sits on the highest-leverage node.
ambient, sovereign demand arriving with it.
PLAYER writes: no growth-rate constraint, no terrain-stability constraint, integrity-visible
substrate across the five bands. Including it costs one row and one gate exemption (it has no source
tradition, so the provenance tooth scopes to source-bearing rows), and it buys the cheapest possible
proof that the schema describes a real physics contract rather than a nine-item art brief.
Each realm introduces exactly ONE signature law and at most two supporting deltas, and from road
position three onward each realm reprises at least two laws already learned. By Ch 74 the player is
fluently juggling eight learned laws. The fairy realm then lands as sovereign not because it adds a
ninth novelty but because it is the place where every law the road taught is simply the local
weather. That payoff structure is what Law 0 predicts and it is measurable straight off the registry:
count signature laws per realm, count reprise_of references per realm. Unlimited novelty per realm
produces nine tech demos; one signature law plus reprises produces a road.
Deterministic arithmetic gates first, then local triage, then full-strength critics who open findings
and are the only tier permitted to close one. Critic reads stay ADVISORY and never gating.
Deterministic teeth, gating, proposed as harness/check_realm_ruleset.py, each shipping a must-fire
and a must-not-fire fixture because a rule that was never mutated was never armed: T-RLM-BASELINE
(exactly one baseline and one source row), T-RLM-PARTITION (holds/bends completeness OVER THE
DECLARED VOCABULARY V, never over physics — the bound is stated on the tooth itself so no report
inherits the unbounded claim, and the tooth ships alongside the §2.3 V-ADEQUACY PROBE that shows it
blind to a deliberately outside-V bend before V is ratified), T-RLM-PROVENANCE
(citations resolve into the corpus registry), T-RLM-EXPRESSION (physical_expression resolves to a real
quantity, never a prose sentence), T-RLM-TRIGGER (parses under condition_expr.py), T-RLM-FLOOR (no
realm bends a Law 6 invariant), T-RLM-UNIVERSAL (scope declared, narrow scopes justified),
T-RLM-LEGIBILITY (taught_at resolves to a same-chapter site with ambient_threat none or low),
T-RLM-NODEADEND (alternate_route non-empty, chaining into the existing gate), T-RLM-COMPOUND (one
signature per realm, two-plus reprises from position three, AND the D-RR.5 BOUND: a signature law
never grants a mastery gate and never confers school membership — a law that does either has become a
school by another name and breaks the ratified "no new schools at Ch 70-74" decision, which is the
one way the signature-law lane could silently violate the decision it was designed to work within),
T-RLM-CURVE (demand non-decreasing across road position, skipping declared-exempt rows by their
empty-string sentinel rather than reading empty as zero, AND declared power agreeing with the spine's
ruled density seams — the anti-drift tooth that stops the ruleset and the spine silently diverging), T-RLM-CARE (care_tier present; the
fringe-uptake firewall as a FIXED-STRING scan for retired coinages and fringe-tradition names on any
player-facing surface, never an escaped pattern), and T-RLM-REVEAL (no realm name, delta vocabulary
or law name on a player-facing surface before Ch 66, reusing check_reveal_discipline.py's mechanism
rather than building a second one).
Runtime conformance, gating at the sampled_invariant tier, three headless tests:
Realm.DeltaConformance (each declared quantity measured in-sim matches within tolerance),
Realm.NoUndeclaredDrift (every NON-declared quantity matches the baseline realm within tolerance —
the harmony property made arithmetic, and the most valuable test in the program because undeclared
weirdness is exactly what "blending harmoniously" excludes and this test sees it with no model in the
loop), and Realm.LawFires (each signature law's must-fire and must-not-fire fixtures assert its
physical expression actually writes the named quantity).
Build-space sweeps extended: sweep every canonical build archetype against every realm and assert
each archetype clears the realm by at least two distinct routes. The design rule this encodes is the
difference between a ruleset and a wall — a realm may INVERT a build's usual answer (that is the
whole point, and the trough realms depend on it) but may never close every answer a build owns. Your
usual answer is shut and another of your own answers opens. A realm only one build clears is a
balance defect exactly as a never-picked bonus is.
Critic rubrics, advisory, each keyed to a NAMED slop mode or doctrine clause rather than a generic
quality word: RB-RLM-LAWFUL (from ten unlabeled frames, state the realm's signature law — if the
critic cannot, Law 5 has failed on the art side even where the data passes), RB-RLM-HARMONY (name the
physical quantity that changed; "a glow" is an overlay finding against Law 3), RB-RLM-PROVENANCE
(name the source tradition from frames alone; a verdict of generic fantasy is the sharpest available
slop detector), RB-RLM-PRESENCE (name who opposes the player and what face they have — keyed directly
to the recorded "no-enemy mythological realms" and "faceless realm war" defects), RB-RLM-POSITION
(shown this realm beside the previous one, state which comes later on the road), and RB-RLM-CARE (the
living-tradition and anti-fringe read run TOGETHER with a devil's-advocate pass calibrated against
over-conservatism, so both failure directions are checked in the same wave).
The director re-read gate is unchanged and binding: before any realm play-quality claim enters a
commit message, a report or a canon doc, the director opens the deciding frames from the newest
session directory resolved through last_run.json and records what was seen. Realm work is unusually
exposed to the frame-claim failure mode because realm frames are the most beautiful and the most
likely to be described rather than looked at.
Realm names, delta vocabulary, law names and the nine-realm count are spoiler-tier and belong to no
player-facing surface before Ch 66; T-RLM-REVEAL arms this. The legitimate forward seeds already
exist and need no invention: the transformation chapters at Ch 38, 55 and 58, and the anchor-density
Component 11 sites, all show physics behaving strangely without naming a realm. The L3 cross-realm
convergences are FELT and never declared. Concept art for realms is spoiler-tier and does not enter
any pre-Ch-66 surface including marketing and gallery contexts. The Grand Sage silence gate is
unaffected and continues to bind everything here.
---
The unit of WORK is the ZONE, and the zone already exists as a typed row: T1_Region_Page_Template's
Zone Catalog carries terrain, gameplay functions, ambient_threat, vril_density, climactic_affordances,
protected, area_bounds and build_priority, and declares seven extension columns of which three —
care_tier, asset_routing, environment_grammar_refs — are this program's outputs. The pipeline needs
no spine of its own; it fills a table already waiting.
Their declared states differ, and the difference matters, so it is read off the declarations rather
than assumed uniform. Only ONE of the three is empty by declaration:
asset_category/asset_routing enum is ratified (an unratified enum value would be invented canon)."
This program does not own its unblocker; the enum ratification at §6.3 does.
flag and the page's explicit care-register overlay where one is ruled." The program RECONCILES with
that derivation and does not treat the column as a blank slate it owns. This is the same fact that
drives the §3.3 field split — a column with a ratified enum and a ratified derivation is not
available to be repurposed as the costume rule's key.
no defensible match exists (never a forced one)." Emptiness here is a defensible verdict, not a
waiting slot, so filling a cell requires a real EG_ match and leaving one empty stays legal.
The unit of DECISION is coarser. A realm gets ONE establishing board that rules its whole visual
grammar, and only after that board is picked do its zone boards run under the ratified grammar.
Ruling a realm zone-by-zone would produce a realm assembled from nine independently-good rooms that
do not read as one world, which is the exact failure the realm-stratum vision names.
The output of a pick is a RATIFIED ART BIBLE per realm. The bible is the only artifact the asset
factory reads. Boards are the deliberation record; the bible is the law.
Three panels, a marked recommendation on each, and a pick line. The board is a VERSIONED IN-REPO
ARTIFACT per the §6.3 ruling — a compact markdown doc plus its image files under
docs/art_boards/<realm>/, delivered as rendered files, self-contained and offline-safe by
construction and never published. The board's discipline mirrors the standing decision protocol
rather than inventing a new one: every panel carries options, a one-line rationale per option, a marked RECOMMENDATION with
its reasoning, and the STRONGEST OBJECTION to that recommendation. A board that presents images
without a recommendation is a bare question wearing pictures, and the protocol forbids it.
PANEL A — THE ACCURATE DEPICTION. What the source tradition itself says the place looks like. Three
options, each a plate sheet rather than a single image: silhouette and architecture, palette and
light, and one signature detail. Each option carries four things and no fifth — a one-line rationale
naming the tradition-specific reading it commits to ("the Duat as the Book of Gates renders it:
register-banded, procession-oriented, the hours as architecture" is a rationale; "Egyptian
underworld, mysterious" is not); a CITATION LINE naming the substrate rows behind it, where a
citation that cannot resolve is a defect and an empty one means the option is not an accurate
depiction at all but a guess with good lighting; a RIGHTS value per image; and TWO badges read off
the Zone Catalog columns rather than re-derived per board — the ratified care_tier badge (standard or
elevated, driving review depth) and the proposed generation_class badge (driving what may legally sit
on this panel at all). The bar is AUTHENTICITY, NOT
REVERENCE, and it binds in both directions: the panel must not scrub what the tradition actually says
(the Lords of Xibalba are the Popol Vuh's own antagonists and the board renders them as such), and it
must not composite across unrelated traditions to make a prettier plate.
PANEL B — THE ENHANCED VERSION. Three options, governed by one auditable rule: REALITY IS THE FLOOR.
Every enhanced option is a strict superset of an accurate option — it amplifies scale, light, density,
motion, verticality or legibility and contradicts nothing on Panel A. An option that changes what the
place IS has left the program and belongs in a canon brief. Each option is labeled with two explicit
fields, WHAT WAS AMPLIFIED and WHAT STAYED TRUE, which is the audit surface: a reviewer who cannot
read those two fields off the board cannot tell enhancement from invention, and neither can Josh.
Each names its PLAYABILITY CLAIM — the specific play the amplification buys (verticality that buys a
climb route, luminance banding that buys wayfinding without a marker, density that buys an ambush
lane). An amplification with no play claim is decoration, and decoration is the first symptom of
slop. And each states the ALTERED LAW in one clause, taken from the realm's ruleset row — which real
physical behavior vril bends here and how the bend reads on screen — so the physics critic can check
whether the image shows a coherent altered law or shows physics simply abandoned.
PANEL C — THE SURROUNDING RECOMMENDATIONS. Not images. A text card set, five slots, because Josh
asked for exactly five kinds of thing and a board that blends them into prose forces him to do the
sorting. STORY SUPPORT (which thread stations here, which reveal is held, what the place is doing for
its chapter). SIDE QUESTS as seeded rows minting into the existing quest-seed queue with its housing
ledger and check_seed_housing.py gate, never a new lane. EASTER EGGS including the completionist and
no-kill tier the natural-voice doctrine reserves hard puzzles for, where an easter egg that gates
progression is a defect by standing ruling. EXPLORATION, ADVENTURE AND ACTION HOOKS — every realm node
is combat-present by Pillar 3, so this slot always has content and a board returning an empty action
slot has found a design defect rather than a quiet place. NEARBY IMPORTANT ZONES, writing into the
Zone Catalog's neighbor-facing fields rather than into prose. Each card carries a provisional id in
the target registry's own id space so the pick is an APPLY, not a re-authoring pass.
THE PICK LINE. One line per panel: an option id for A, an option id for B, each with a free-text
amendment lane ("A2, but the light comes from under the water"); a comma list of card ids for C, with
the rest dropped or deferred by default. A fourth lane is named on the board so it is never
improvised — MORE LIKE THIS: Josh names a direction, the board regenerates, and the regeneration
cites the previous board so the reasoning chain survives. Nothing about the pick line asks him to
write a brief.
WHAT MAY NEVER APPEAR: no internal mode taxonomy or verb-kit vocabulary (a board is not a player
surface, but its captions are seed text for downstream surfaces and taxonomy that enters here leaks
outward); no unmarked recommendation and no more than one marked recommendation per panel, enforced
by the deterministic gate because a board with two recommendations is a board that failed to decide;
and no previz_only image on Panel A, ever.
primary, Panel B underlays. UNGATED, runs today. This is the lane the repo is best positioned for
and the one nobody has scheduled: BOOKS_OWNED records Josh's complete National Geographic run since
1888 with VISUAL REFERENCE named as its primary value, an owned reference corpus covering most of
the arc's geography and sitting unused by any visual lane, plus museum open-access collections,
which are the cleanest rights class available for the living-tradition realms.
structure. Panel B default. GATED-5090 for volume. Recommended over raw generation for a structural
rather than a stylistic reason: a paintover inherits the reference's proportions, so the enhanced
option is a superset of the accurate one BY CONSTRUCTION rather than by a reviewer's judgement,
which converts "reality is the floor" from a promise into a property of the process.
realm is wholly a literary or game-canon invention. GATED-5090.
THE COSTUME RULE, extended to the board and stated because it is the highest-value single rule in the
design. The existing photogrammetry-costume rule says a generated image run through a reconstruction
pipeline produces an asset wearing photogrammetry's credibility as a costume. The identical error is
available one step earlier and is easier to make: a generated image placed on PANEL A wears
ACCURACY's credibility as a costume. Josh reads Panel A as "this is what the place actually looks
like." An image that is a model's guess, sitting under that header, has told him something false in
the one place he is relying on the pipeline to be true. So the rule takes no judgement call: Panel A
is Lane R for every realm whose source tradition is living, and Lane G is permitted on Panel A only
where the realm is wholly invented. Atlantis and the fairy realm qualify. Nothing else on the road
does. The rule is keyed on a COLUMN rather than on reviewer memory, so the deterministic gate applies
it, not a person — but on WHICH column is a schema question the charter got wrong on its first pass,
and the fix is below.
THE RULE'S KEY IS TWO ORTHOGONAL FIELDS, NOT ONE. The charter originally keyed the costume rule on
care_tier. It cannot be. care_tier is a RATIFIED two-value enum and its declaration in
T1_Region_Page_Template is explicit: "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." It has no way to say "living," and it has a live derivation authority the
program must RECONCILE with rather than overwrite. Registers this charter's §1 uses descriptively —
Xibalba's LIVING, Naraka's ELEVATED-with-maximum-register-duties — are analyst prose readings of care WEIGHT, not values of that
column, and this charter asserts none of them as enum values. care_tier stays exactly as ratified and
this program does not widen it.
What the costume rule actually needs is a different axis. The two things being conflated are
orthogonal:
standard and elevated. Governs critic routing, review depth, and the §17.1-is-collective duty.
Owned by the care doctrine and by L-R3's named output, not by this program.
all. PROPOSED here as its OWN field. It is a legality axis about image PROVENANCE, and it is a
fact about whether a real tradition is being depicted, not a judgement about how carefully.
They are independent, and Atlantis is the proof: generation-legal (wholly invented, no living
tradition to misrepresent) AND elevated-duty (its RECEPTION history — Donnelly, Blavatsky root-races,
documented hyperdiffusionist racism). Those are compatible facts about different columns. That is
the whole of contradiction C-4: not a disagreement between two analyses, both of which are right, but
a schema defect in the charter's own keying. Split the fields and C-4 closes as a SCHEMA FIX, and
this charter records it as closed on those terms rather than carrying it forward as a fork.
THE PROPOSED FIELD. generation_class, a closed four-value enum, proposal-tier, riding the
registry-extension rail at §7 like every other column add in every other lane:
B. Atlantis and the fairy realm, and nothing else on the road.
(a hallucinated Sumerian deep is still a hallucination) and legal on Panel B over a Lane-R
underlay. The Apsu; Pohjola and Hades pending L-R3.
must carry a Lane-R underlay. Xibalba, Naraka and Meru, the Duat; Annwn and Yggdrasil pending
L-R3.
living, because the failure direction of guessing wrong on this axis is asymmetric: a wrongly
restrictive call costs a wave of reference assembly, a wrongly permissive one ships a model's guess
about a living people's cosmology under a header that says it is accurate. Annwn sits here today,
which is C-3 stated as data instead of as a disagreement.
Assignment is not a care judgement and not reviewer memory: it is the factual question L-R3 answers
(does this tradition have practitioners today), which is exactly why L-R3's named output GATES the
first board of any realm it settles. Until that output lands, unsettled is the default and the gate
is strict.
WHAT THE GATE READS. read_boards.py's costume tooth reads generation_class, never care_tier: for any
realm whose generation_class is living or unsettled, every Panel A image must carry a rights value in
the reference set. care_tier continues to drive critic routing and review depth, which is what it was
ratified to do. Two fields, two jobs, one tooth each, and neither column asked to mean two things —
the same reasoning that keeps the per-world ruleset row out of the per-feature environment-grammar
registry at C-5.
Where this lands in the plan: the generation_class mint is a Wave-2 item under §5.3 with its own
critic and its own baseline refresh, and the field-set decision is a DIRECTOR call listed at §6.3,
not a Josh brief — it is schema, not world canon.
THE NEVER-RAW-GENERATE RULE EXTENDS FROM STRUCTURES TO ICONOGRAPHIC SYSTEMS. The existing constraint
covers real cultural ARCHITECTURE. The realms are made of cosmological grammars — the Duat's
registers and gates, Naraka's strata, Xibalba's houses, Meru's axis — which are living traditions'
own pictures of their own cosmos, with their own internal logic, and which a model will confidently
hallucinate. That carries the same class of error as a hallucinated temple: it gets a living
tradition's picture of its own cosmos wrong, at AAA fidelity, in a shipped product. This is a genuine
extension of a ruled constraint into new territory and the realm program is the first work that hits
it.
Three documents already name the same hole and none closes it: the AssetGen doc opens the
capture-rights gate explicitly ("provenance and permission, not just realness"), the drop-in contract
raises it as an open item, and the end-to-end build plan states it as MISSING outright — no
reference_image_ref, capture_rights, reference_source or use_class field exists anywhere in the asset
lane, so the agent cannot distinguish a rights-cleared photo from a prohibited one or from
AI-generated input. The concept lane is where it becomes urgent because concept art is where
reference imagery ENTERS the pipeline; closing it here closes it upstream of every consumer. The
precedent to copy rather than invent is the book-substrate lane's use_class field: a per-row enum
recording what may legally be done with this source, checked by a gate, refusing to let an
unpermitted class flow into an output.
The vocabulary — every image on every board carries exactly one value: own_capture (Josh's own
photography or footage, the cleanest class); museum_open_access (the best available class for the
living-tradition realms and the reason Lane R is viable for the Duat at all); public_domain (with the
note that public domain in the image does not by itself clear a site-access or cultural-protocol
question, so this class still routes through the cultural-authenticity critic); licensed (with the
licence identified); generated_local (legal on Panel B, and on Panel A only for invented realms);
previz_only (any territory-capped, MAU-capped, non-commercial or closed-SaaS route — usable
internally for thinking, NEVER promotable to an art bible and never on Panel A).
LANDED 2026-08-02 — the field this section asked for now exists. "licensed (with the licence
identified)" had nowhere to identify it: docs/art_boards/PLATES.csv's only licence column was
generation_licence_ref, which is the C-12 MODEL licence and reads n_a_lane_r on every Lane-R
plate because no model ran. The first board shipped two Panel A plates routed licensed with no
licence named at all, one of them citing an owned National Geographic run as though ownership were a
licence — docs/BOOKS_OWNED.md §8 names that run a production art-reference asset and says nothing
about reproduction rights. Both rows are re-routed, and PLATES.csv now carries image_use_class,
built on the precedent this section names rather than on a fresh one: a per-row closed enum of
uppercase tokens recording what may legally be done, gated by T-IMAGE-LICENCE in
harness/read_boards.py, and carrying REFERENCE over verbatim from the book-substrate lane because
that token is exactly the case that was mis-stated. SITE-level rights remain the separate,
still-unbuilt second field this section also names.
Two rules make the vocabulary do work: PROMOTION IS RIGHTS-GATED (an image moves from board to bible
only if its rights value is in the promotable set), and SITE-LEVEL RIGHTS ARE SEPARATE FROM
IMAGE-LEVEL RIGHTS (some heritage sites and museums prohibit commercial capture or commercial 3D
reconstruction regardless of who took the photo; the per-site check is a second field, not a
substitute, and the realm program is the natural place to populate it because realm boards are the
first work that systematically assembles real-site reference at scale).
describes — architecture, palette, light, materials, scale, inhabitants, the place's own internal
logic. Grounded in the region-filtered substrate plus the spine node plus the RCM section. Output
is a cited research pack both Panel A and Panel B draw from. The largest single body of work in the
program, and entirely UNGATED.
rather than three renderings of one idea. The standing protocol's requirement to include the
already-applied option has a direct analog: where a realm already has an authored visual
description in its spine node, that description is an option on the board, named as the incumbent.
and Josh never sees a board that failed a floor. The four are the CULTURAL-AUTHENTICITY critic
(tradition's own iconography or generic fantasy wearing its name; compositing across unrelated
traditions; a monster defanged, where Xibalba's Lords, Naraka's wardens and the Apsu's sea are
marquee antagonist material and rendering them as scenery is the defect the rated-R ruling exists to
catch; deity portrayal within the ruled limits; AND the counterweight, is any option TIMID —
care-tier inflation is a defect by ruling and a critic that only ever pushes toward restraint
produces nine reverent empty realms); the PHYSICS-COHERENCE critic (name the altered law, name the
mechanism, check the consequences — if buoyancy inverts, things that are not the hero object also
fall upward and shadows agree — and check it against the mechanics, because a law that cannot become
an affordance is a law the player can only look at); the AAAAA DE-SLOP critic (run the
interchangeable-otherworld test LITERALLY by presenting plates unlabeled and asking the critic to
assign them; check the monotone solemn register, where Annwn is an Otherworld of PLENTY and
rendering it as another dim underworld is a canon error and an art error at once; check the empty
realm and the depopulated establishing shot; check symmetry-and-centrality as a substitute for
composition, the most common generative-image tell); and the REVEAL-DISCIPLINE critic (which guards
the DOWNSTREAM ROUTING of ratified art rather than the boards themselves — an art bible for a
Ch-75-or-later node may not be referenced by any zone whose chapter set contains a chapter before
66, and a board that renders the realm's script legibly in an early-chapter context breaks a hard
line with a picture, which no text gate would catch).
that moment, with the pick line QUOTED VERBATIM so the ruling is never paraphrased. Per the standing
lesson that a subagent writing a user ruling into canon reads as fabricated authority, the
ratification write is a DIRECTOR-level apply, never a subagent one.
Deterministic pre-critic before any model reads a board: harness/read_boards.py, built on
harness/read_captures.py's interface contract rather than a fresh one, gating PRESENCE and SHAPE and
never CONTENT. It asserts that every promised plate exists as a real non-empty file (a
promised-but-missing plate is a structural FAIL, an expectation with no plate is a WARN); that every
image carries exactly one rights value; that every Panel A image on a realm whose generation_class is
living or unsettled carries a rights value in the reference set (the mechanical form of the costume
rule and the most valuable tooth in the tool — and it reads generation_class, never care_tier, per
the §3.3 field split); that exactly one recommendation is marked per panel; and that every citation line
resolves to a real substrate row. It copies read_captures.py's --self-test mode with must-fire and
must-not-fire fixtures and exit code 3 for broken teeth, because a tool that was never mutated is not
armed.
One per realm at docs/art_bibles/REALM_<node>.md, the ONLY artifact the asset factory reads, which is
what keeps the boards from becoming a second source of truth. Contents: PALETTE (named colors with
values AND the rule for where each is allowed to appear, because a palette without placement rules
produces nine realms that use their palettes the same way); LIGHT DOCTRINE (source, direction, color
temperature, and the IMPOSSIBLE-LIGHT clause stated as an altered law rather than an effect);
MATERIAL VOCABULARY (mapped where possible to real material behavior so the physics critic has
something to check); SILHOUETTE AND ARCHITECTURAL GRAMMAR (the field that most determines whether a
realm reads as itself from a hundred meters away); SCALE RULE (tied to the Zone Catalog's
meters_per_vertex rung so the art call and the geometry budget agree); THE VRIL-PHYSICS SIGNATURE
(the realm's altered law with its visible read and its mechanical read — the field carrying the
directive's core sentence, and the field that hands to T0_Environment_Grammar_Registry); MOTION
GRAMMAR (underused in most art bibles and load-bearing for an otherworld, since motion is where "this
place obeys different rules" is FELT rather than seen); THE FORBIDDEN LIST (the de-slop clause per
realm, written as specific prohibitions rather than a quality aspiration); CITATION APPENDIX with
rights values; and THE INTENSITY SLOT showing where this realm sits on the climb and what the two
adjacent realms do, so Josh rules the ladder as he rules the rooms.
Downstream consumers, each already existing, so nothing orphans: T0_Environment_Grammar_Registry
(the realm's features become EG_ rows carrying physical_properties_array, elemental_affinity,
tactical_affordance_class, attack_verbs, enemy_usable, protected_care_ref — which is what stops nine
gorgeous realms from having ordinary combat in front of them; an art-bible field that cannot become
an EG_ row is a field describing scenery); the Zone Catalog's three output columns, per their
differing declared states at §3.1;
the region page's architecture and UE-build sections; mesh_tags.style at generation time (concept
decides, geometry executes); and the image-critic expectation manifest, so that when the realm is
built the critics judge frames against the RATIFIED BIBLE rather than a reviewer's memory.
A pre-5090 board should be SHIPPED rather than deferred, with Panel B as a REFERENCE MOODBOARD rather
than a generated concept set — existing rights-cleared art and photography demonstrating the
amplification DIRECTION, with the same amplified and stayed-true labels. It is weaker than a generated
concept and the board says so on its face. What it buys is that the palette, light doctrine,
silhouette grammar, altered law and the whole of Panel C are ruled BEFORE hardware arrives, so the
5090's image time is spent rendering ratified directions rather than exploring open ones. That is the
correct order of operations regardless of hardware, and the constraint is doing the program a favor.
Two, because the road has two genuinely different problems and piloting only the easy one validates
half the machine.
both generation lanes, the pick loop, the bible and the scatter with no capture-rights blocking.
It also carries a real anti-fringe duty attached to its RECEPTION history rather than its content,
which exercises the cultural-authenticity critic in its most interesting mode.
and iconographic corpus, elevated care, and a Panel A that must be assembled from rights-cleared
reference rather than generated. If the machine works here it works everywhere on the road. Ch 69
is also the materialized-argument node, so its board must serve a boss class whose shared visual
grammar is already RULED (inscription-made-flesh, per-boss regional condemned-register skin, shared
hollow-core tell, and the bound that a construct never wears a culture's sacred beings) — a live
constraint and a worked example of what an art bible looks like done right.
Engine versus canon, declared up front so game #2 harvests it: ENGINE is the board format and schema,
read_boards.py and its fixtures, the rights vocabulary and its promotion gate, the four critic rubric
SHAPES, the bible TEMPLATE, and the pick-to-ratify-to-scatter loop. CANON is the nine realms and
their research packs, every ratified bible, the vril-physics signatures, the care-tier assignments,
and the reveal-window constraints.
---
The realm surface is the single most under-lensed marquee surface in the game, and the roster
half-knows it in a shape that HIDES most of the gap. Cell 14 names the gap as
"otherworld-as-distinct-ruleset" and bundles it with the Personal Dimension, whose teachers are
already on the roster in five places — so the roster reports a covered surface as uncovered while
diluting the real gap. Cell 15 claims a slice of the realm road too, so the surface is touched by two
cells and OWNED BY NEITHER, which under the QA program's aspect_id join contract will report the
realm rubrics as ROSTER-INCOMPLETE indefinitely. The only lens whose charter names a realm at all is
scoped to roguelike-run properties. And god-of-war — the closest AAA analog that exists, a shipped
nine-realm Norse cosmology with a world-tree hub — RAN, and produced four survivors, none of which is
realm adaptation. The closest teacher we have was spent on companions and presentation.
Decomposed against what the built chapters actually contain, the realm surface is TWENTY aspects
(R-A1 through R-A20), of which three have a real teacher, four have a partial or adjacent one, and
thirteen have NONE. Five of the nine realms are death or judgment realms sitting consecutively at
nodes four through eight, against one partial register teacher and a slop mode our own review already
named.
The realms build after the earthly regions and the fairy realm builds last — at slice-end under the
2026-07-27 ruling, see the banner — so a naive reading puts every realm lens at P2 or later. That
reading is wrong for exactly three of them, and the reason is the freeze, not the build order. What freezes soon and is expensive to unfreeze: the realm ruleset
SCHEMA, the HARMONY DOCTRINE, and the DR-2 tiered-asset request contract. Anything shaping those runs
before them.
DOCTRINE WAVE — runs before the schema and the doctrine freeze, and before the first art-concept
fan-out:
· Shin Megami Tensei and Persona's mythology roster · Hades's Greek layer only, handed over so two
lenses do not collide. Teaches source-selection discipline (Ragnarok ships nine realms from one
tradition and had to decide what each realm IS as a place, not as a myth-summary); the
myth-literacy ladder (Okami rewards Shinto knowledge without ever requiring it — our exact L2/L3
contract); insider versus outsider adaptation; and the mass-roster failure mode where SMT and
Persona flatten cosmologies into collections, which is instructive precisely where our audit-only
assessor rule is the dodge. Aspects R-A2, R-A3, R-A11, R-A13, R-A15.
Depths and Sky layers as a widen of the in-flight botw-totk lens · Antichamber or Superliminal as a
perception-rule minor cluster. Control is the medium's best answer to the directive's own sentence —
the Oldest House is a building whose PHYSICS ARE ITS LORE, documented in-world as science by a
bureau that studies it, which is vril-altered physics at realm scale, shipped. Manifold Garden
teaches gravity-as-per-world-axis and how impossible architecture stays navigable. Echo teaches ONE
rule pursued to its floor rather than a bag of gimmicks. The Depths teach a distinct-ruleset LAYER
inside a shipped open world. Aspects R-A1, R-A3, R-A4, R-A7a (the experiential threshold,
explicitly NOT the conveyance, which the live transitions fork already owns and which a realm lens
must not re-litigate). This lens IS the schema's design input.
Origins' Curse of the Pharaohs Duat and Aaru zones plus Discovery Tour · Raji · Mulaka · Pentiment.
Never Alone is the collaborative gold standard with elders' documentary material as the unlock
reward, a structural model for our L2 layer. AC Origins is a direct precedent never examined: a AAA
studio shipped a playable Egyptian afterlife with the Duat and the Field of Reeds alongside a mode
that separates the game from the scholarship — the exact separation our audit-only rule performs.
NAMED OUTPUT, and the reason this lens GATES: the per-realm care tiering, delivered as a table the
art-concept fan-out reads before generating anything. Aspects R-A2, R-A9, R-A11, R-A15 as a BOUND
on art concepts, R-A20.
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 the
CORRUPTION-AS-CONDITION slice of L-R7, because the guardrail binds every fairy-realm design
decision from the first concept forward.
That is three full lenses plus two slices, inside the roster's five-to-six-per-wave cap, matching the
existing wave shape exactly: one read-only agent per lens, model set explicitly to the opus tier
alias, compact returns with full detail to scratchpad.
P2 WITH THE FACTORY:
Colossus · Death Stranding · Ico · Sable. Teaches traversal as the emotional content rather than
the connective tissue, the staged vista and arrival beat, wordless procession, scale as feeling.
The unique lesson: the Ch 76 Approach is canonically a no-combat no-fail-state VALLEY that must
still be the second-most-important playable sequence in the game, and nothing on the roster teaches
how a walk carries that weight. Aspects R-A5, R-A17, R-A4, R-A13. If a P1-C slot frees up this
belongs there, because its findings feed the presentation contract.
Finch · Hades's register layer · Kentucky Route Zero as an optional minor. Spiritfarer is grief as
an entire game's register and its ferrying structure is the closest analog to the Ch 75
grief-geography encounter (grief crystallised into terrain, addressed by restraint and consent
rather than damage). Grim Fandango proves a death realm need not be solemn and that register is a
CHOICE per realm. Aspects R-A6, R-A12, R-A18.
shift · Little Nightmares · Alice: Madness Returns as a minor. Psychonauts 2 is the best in the
medium: each mental world has its own physics, art language and emotional thesis, and they cohere
because the FRAME is consistent even when the rules are not — exactly the
nine-realms-must-not-become-nine-minigames problem. Silent Hill 2's shift teaches the crossing as a
RE-RULING of a space you already know, the cheapest and most powerful realm-crossing move available.
Merge note: L-R2 teaches RULE-per-space and L-R6 teaches THEME-per-space; if wave budget is tight,
run them as one lens with two title-clusters. They should not both be dropped.
WITH THE FAIRY BUILD, LAST:
Princess's Twilight Realm and its exiled sovereign · Elden Ring's Nokron and Deeproot as a widen.
Hallownest is a broken kingdom with a ruined king, a court, a fading folk and an INFECTION spreading
through the kingdom's own substrate — structurally our fairy realm almost line for line, including
the corruption-as-condition framing our guardrail requires. A teacher that shipped
corruption-as-condition well is load-bearing for the single sharpest care line in the finale.
Aspects R-A14, R-A12, R-A15, R-A6.
the world-tree hub, realm-travel presentation, per-realm identity, and how a shipped AAA held nine
distinct otherworlds together. Cheaper than any new lens and higher-yield than most.
L-R1 and its register layer to L-R5, so three lenses do not each half-read the same title.
is unlensed is the roster's own principle failing on a title it already owns.
other question.
rather than minting it inside L-R2.
CEILING-SETTING art surfaces and the tiered-asset contract must carry a realm tier above the region
tier — plus R-A16 per-realm audio identity claimed here, since T1_Audio_Spec is an empty stub.
This one is time-critical: that lens is P1-C precisely because it must precede the DR-2 contract
FREEZE, and if the ceiling requirement is not in the charter when it runs, the asset contract
freezes against earthly-region assumptions and the realms inherit a tier never sized for them.
as the same register.
Dimension — route to the five existing sandbox lenses, with its finding landing as DESIGNED-UNBUILT
rather than MISSING).
honest current teacher and verdict class, so the QA program's aspect_id join stops reporting the
realm surface as roster-incomplete.
Net roster change: 40 lenses to 47, of which only three run early.
---
Every wave runs the standing method: deep-dive fan-out, alternatives analysis, adversarial critics
across all lenses, model set EXPLICITLY on every agent to the opus tier alias, one read-only agent
per unit, compact returns with full detail to scratchpad, deterministic scripts wherever a script
suffices. The director orchestrates, adjudicates contested findings, re-reads deciding artifacts, and
performs final integration and every canon-write. No wave is a ceremony: the fan-out fires when the
work touches more than one target, makes a claim needing independent verification, or is volume
authoring.
AMENDMENT IS DONE, not to-do — verified in docs/proposals/COMPARATOR_LENS_PROGRAM.md at WAVE P1-C:
"presentation-craft-anim-audio (precedes the DR-2 asset contract — AMENDED 2026-07-27 per the Realm
Design Program: the MYTHOLOGICAL REALMS + the fairy realm are declared CEILING-SETTING art surfaces
for this lens; the DR-2 tiered-asset request contract must reserve the realm tier's headroom before
it freezes)." The text matches this charter's requirement exactly. Applied in the working tree and
pending commit at the time of writing, so the only remaining action is that it LANDS in a commit —
not that it gets authored. Recorded this way so it is neither done twice nor reported as
outstanding.
(2026-07-28, register v1.3 entry 70): THERE ARE NO REALM ZONE ROWS TO POPULATE.** The zone
catalog is earth-geodata-bound (the four live pages, 27 earthly zones); --populate-empty over
an empty realm row set succeeds and writes nothing, so running this step as scheduled would
no-op green and report a prerequisite met that is not. The discriminator is now landed — the
Zone Catalog carries zone_space_class enum{earth_geo | realm_authored}, default earth_geo,
applied to all 27 live zones 2026-07-28 through the sanctioned writer — and this step UNBLOCKS
only when realm zone rows exist (zone_space_class=realm_authored, realm_id +
authored_bounds mandatory, the earth-geodata fields empty; the Wave-2/realm-page work mints
them; the T-RLM-ZONE tooth — every mythic_realm region has ≥1 realm_authored zone row — arms
with the mint). When it unblocks, the populate runs THROUGH THE SINGLE SANCTIONED WRITER —
harness/apply_zone_catalog_columns.py --populate-empty, --dry-run then --apply, never by hand —
per the ruled Zone Catalog constraint at §7, values from the ratified two-value enum (standard ·
elevated) reconciled against the column's own declared derivation, with L-R3's care table
(docs/pipeline_review/REALM_CARE_TIERING_2026-07-28.md) authoritative. The generation_class
axis is NOT written in this wave: it does not exist as a column until the Wave-2 mint, and until
then every realm reads unsettled by default, which is the strict path.
the spine site rows use baseline, low, moderate, high, anchor. These are the same five ranks under
two spellings. The new gate has to pick one anyway, and picking silently would create a third.
fourteen deferred Thread-5 refs, ready for Wave 3.
L-R1, L-R2, L-R3 plus the register-variety and corruption-as-condition slices. Five units, inside the
per-wave cap. L-R3's named output — the per-realm care tiering — is the gating deliverable for every
later art wave and settles contradiction C-3. No doctrine lens launches against a schema already
frozen; a return landing after its schema froze becomes a fix-forward ticket with the miss recorded.
returns, under its own fresh-context critic.
refreshed in the same commit.
run_gates.py, and NOT counted as armed until a mutation has been shown to fail it.
build).
same critic, seeded from the environment-grammar property keys and widened to span the seven
profile families — followed by THE V-ADEQUACY PROBE at §2.3, which is a gate item and not optional:
nine deliberately-outside-V bends authored as fixtures, T-RLM-PARTITION shown blind to each, V
widened, re-run. V is not ratified until the probe's residue is recorded.
living · unsettled), declared on the registry-extension rail with this program as its single owning
system, with docs/registry_extensions.json and docs/fidelity_baseline.json updated and
check_registry_extensions.py passing in the same commit. Explicitly NOT a widening of the ratified
care_tier enum, which this program leaves exactly as declared. Batched with the other column adds
per the rail's batching instruction. This is the item that unblocks the costume rule's tooth, which
unblocks Pilot 2.
generation_class enum shape, the teeth and the fixtures are ENGINE; the per-realm values are CANON.
The five structural blanks and the drift, ranked by what they cost the program if unfixed before
authoring starts:
from the ruleset rows.
displaced pre-re-architecture body with a do-not-use instruction, and each needing its real corpus
listed (Homer, Hesiod and Virgil; the Popol Vuh; the Puranic and Vedic tradition; the Four Branches
of the Mabinogi; the Poetic and Prose Eddas). Note that the spine entries DID run build-time
deep-research passes with adversarially-cited agents, so the DESIGN is grounded even where the
corpus registry is not; the gap is in the acquisition and citation layer.
CONFIRMED per §1.1 (blocks present for Ch 66-72 and Ch 74, no Ch 73 block; the same search shape
hits Ch 72 and Ch 74, so the zero is real). The repair adds the missing block with the Four
Branches of the Mabinogi as its Tier 1 anchor (the Davies 2007 Oxford translation, T0_RC_01324,
already in the corpus registry and already carrying Annwn in its own description, so the fix is a
ch_cross append plus a block, not an acquisition). Distinct from the substrate-gap repair above:
the per-chapter "Ch 73 — Annwn" section DOES exist and needs its displaced Realm-War body replaced,
while the acquisition list needs a block created from nothing.
movement and permanence column empty, and three primary keys still carrying the retired coinages.
spine's better descriptive labels in the ruleset row rather than by minting enum values.
Ch 76, only the waterfall row.
needing no hardware.
fixtures, the art bible template, and the four critic rubrics.
hole three documents have been carrying open.
Geographic corpus. This produces REAL Panel A content today.
explicitly so no reader infers otherwise: PANEL C CARDS LAND PROPOSAL-TIER, and a card's provisional
id in the target registry's own id space is a RESERVATION, not a housed seed. Nothing enters the
ratified quest-seed queue until the pick, and check_seed_housing.py is the gate that keeps the
distinction real rather than conventional. This wave authorises card AUTHORING, never seed housing.
since a visual crescendo that fights the pacing topography will lose to it in play.
Pilot 1 Atlantis, then Pilot 2 the Duat, with Panel B as a reference moodboard and the board saying
so on its face. Picks ratify into two art bibles, which scatter. The machine is proven on both the
easy and the hard case before any of the remaining seven realms consumes a wave.
beyond the realm establishing board. Gated on the 5090.
rather than scheduled separately, since that pre-flight is already specified as a cheap GPU-free
licence re-check.
real-world-architecture test whose stated purpose is to CONFIRM the never-raw-generate call rather
than to greenlight it.
AAAAA-polished slice comes first and Josh plays nothing before it). **The slice is PROLOGUE→CH 13 and
the gate fires twice** (Gate 1 at the Ch 2-13 exit, Gate 2 at the full-slice exit).
"last, after all nine mythological realms"). The nine mythological realms stay post-slice; the Ch 75-77
fairy-realm return extends this build rather than preceding it.
Law 0's reasoning, which the re-pointing leaves intact.
Desk work now and all of it: doctrine, lenses, schema, world repair, concept research, rights, boards
with reference panels. Generation when the hardware lands. Builds after the earthly regions; under the
2026-07-27 ruling the fairy realm and the origin pair land at SLICE-END, after Ch 2-13, with the nine
mythological realms following in the post-slice factory era.
---
The choice boards ARE the Josh surface, and the pick line is deliberately three short lines. Nine
realm establishing boards plus the fairy fold, plus zone boards under each ratified grammar. His
input is the PICK, with a free-text amendment lane and a MORE LIKE THIS lane so a redirect never
requires writing a brief. Everything upstream of the board — research, alternatives, critics,
regeneration on a NO-GO — is automated, and he never sees a board that failed a floor.
Per the standing protocol, none of these goes to Josh as a bare question. Each gets alternatives,
deep-dive, adversarial review, a recommendation and the strongest objection. The recommendations
below are the charter's opening positions, not the briefs.
sitting). The INTENSITY half was superseded by the eighth-sitting Pattern A ruling: CH_74 is
PEAK in the live registry (the crossing IS the apex; CH_66 valley), landed with the R3 gate
re-scope — the lift this fork weighed already happened by Josh's own arc ruling, inside the
hard lines. The CHARACTER half Josh ruled at the fourteenth sitting: option (b) — PER-REALM
crossing character (each tradition's own threshold: the Duat's gates are not Yggdrasil's
roots) inside an intensity envelope bound to the Pattern-A terrace, so valleys stay valleys.
The flag asymmetry rider: every realm's hard-line flag state derives EXPLICITLY from the
ratified care table (docs/pipeline_review/REALM_CARE_TIERING_2026-07-28.md) — none by
omission. The pre-ruling brief text below is historical.
The original fork text: the directive asks the realms to strongly build intensity toward the
fairy realm. The live registry (at charter-write time) had Ch 73 Annwn and Ch 74 Yggdrasil —
the last approach and the DOOR — both at standard, with the road's emotional apex at Ch 69,
node four of nine. The two hard
lines a lift would touch, verbatim from live T0_Hard_Lines so the brief carries the constraint
rather than a paraphrase of it — HL_0048: "Mountain-range topography across the 77-chapter arc is
canonical. No three consecutive intensity-peaks. No three consecutive valleys. Mode variety per
chapter per Pillar 11 — no chapter coasts on one mode." HL_0049: "Two consecutive chapters never
share the dominant combat type. Every five-chapter window contains three or more distinct
combat-mode dominant chapters." What that means for THIS fork, stated precisely: an intensity lift
at Ch 73 or Ch 74 is an HL_0048 question only — HL_0049 governs combat MODE, not intensity, and
the road's modes at Ch 72-73-74 (vril_management, direct, environmental) already satisfy both
HL_0049 clauses and are untouched by any intensity change. The live intensity string across Ch
70-77 is standard, peak, peak, standard, standard, peak, peak, standard; raising Ch 73 and Ch 74
to peak would put peak at Ch 71, 72, 73, 74, 75, 76 and break HL_0048's no-three-consecutive-peaks
clause outright, so the fork is not "lift the door" but "which ONE of the two, if either, and what
drops to pay for it." Also touched: the reviewed topography baseline, which is structural canon
under the cardinal rule. Opening recommendation: do NOT change the intensity column; carry the
directive on the DEMAND and AWE axes instead, and give Ch 74 the road's highest awe ordinal and its
strongest signature law (the carved debt, which is also the Thread-5 capstone and the spine's own
mark_write mechanism). Strongest objection: a standard-rated crossing may simply read as an
anticlimax no matter what the awe column says, and a hard line that produces an anticlimactic door
is a hard line worth a ruling rather than a workaround.
nodes, but Ch 70, 71, 72, 73 and 75 are NOT in that set. That is a real asymmetry inside a single
ratified road, and canonical_hard_line_flag is structural canon the program should decide about
deliberately rather than inherit. Opening recommendation: brief it as a deliberate ruling either
way rather than letting the program's realm-parity work quietly imply a change.
in-world name, the reveal window). The realm program is the natural forcing function because the
Hollow is the negative-space counterpart to the realms and the most direct visual expression of the
vril-altered-physics doctrine outside them. Opening recommendation: carry the existing
recommendations forward — admit as an OBSERVE/CONTAIN class outside the 33 and the 22, "the Hollow"
common with "substrate-echo" as the class term, pure atmosphere pre-Ch-76.
sittings 2026-07-27; ruling record docs/spine/DECISIONS_PENDING_JOSH.md; integration record
docs/pipeline_review/HOLLOW_EXPANSION_INTEGRATION_2026-07-27.md §4.3). The fork is closed and the
bullet above is preserved for provenance only. Admission is RULED and expanded into a content pillar
across eight axes — realm-traversal cleansing, rift dungeons, the visibility law, affliction
cleansing, evil-path hollow abilities, the event layer, the world's cleansing meter, and the vimana
traversal-encounter layer — in TWO TIERS, whisper and content, and the class is FOUGHT and DEFEATED.
"Substrate-echo" is RETIRED; "the Hollow" is authorial and registry only, with the player-facing
vocabulary attaching to the RIFT, which is promoted to the durable activity and marketing noun. And
the reveal is a FIVE-GATE ladder over a three-rung perception ladder, not "pure atmosphere
pre-Ch-76." Read the integration record for the live position; this bullet governs nothing.
is budgeted per realm node against the Pattern A terrace clauses and the ruled power curve, which is
the tightest budget on the board because a density that flattens the Ch 70-71 trough would spend the
road's best intensity instrument. Realm-road cleansing advances the world-restoration worldstate
immediately (WS_024, not the completionist meter) and BANKS its power yield to the ruled climb
points and the ceiling. No cleansing may write a vril density that contradicts a spine node's
declared entry or exit seam, and no cleansing may write demand_ordinal: cleansing yields power,
never demand relief. Realm-traversal cleansing was not named in Josh's meter-scope ruling and
tracks through worldstate rather than the completionist meter.
Board delivery medium was carried here in an earlier draft and has been REMOVED from this list. It is
a creative-within-vision workflow choice, not canon, and the standing protocol's remedy for that is
to apply the recommendation and show the path — not to spend one of Josh's three genuine forks on it.
It is ruled at §6.3.
Per the delegated care calibration and the creative-within-vision rule: the per-realm care tiering
(after L-R3's named output, which settles the Annwn disagreement); the register-variety assignments
across the five death realms; every signature law and delta; the schema's field set and home; the
asset_routing enum ratification (schema, not world canon — and it currently blocks a declared column
from ever being populated); the enum-drift resolution at Ch 70 and 71; whether the Personal Dimension
takes a ruleset row; whether the concept lane becomes its own translation doc (opening recommendation
yes, docs/translation/T99_Translation_ConceptArt.md, sibling to the others, because the AssetGen
doc's authority is over MESHES and a concept lane inheriting mesh constraints by implication is
exactly the unstated inheritance that rots); the SCHEMA FIELD-SET decisions, which now explicitly
include the proposed generation_class enum at §3.3 (schema, not world canon — and note the companion
non-decision: the ratified care_tier enum is NOT widened by this program, so there is nothing to rule
there); options-per-panel (opening recommendation three, with an optional wildcard); and every one of
the Wave-3 repair items, which are grounded structure-fill inside a locked vision.
BOARD DELIVERY MEDIUM — RULED, and the path shown rather than surfaced. Boards land as VERSIONED
IN-REPO ARTIFACTS: a compact markdown board doc plus its image files under docs/art_boards/<realm>/,
presented to Josh as RENDERED FILES at each choice point. The reasoning, and it is the same reasoning
that governs everything else in this project: the repo is the memory, and every pick is diffable. A
board that lives in the repo carries its own provenance — which research pack it drew from, which
critics cleared it, what the previous board looked like when a MORE LIKE THIS regeneration cites it,
and exactly what changed between v1 and v2 of a panel. A board that lives anywhere else is a
screenshot of a decision with no history behind it, and the pick line that ratifies an art bible then
quotes a source that cannot be diffed. It also settles the spoiler question by construction rather
than by compromise, which is why C-7 closes with this ruling: nothing is published, so the
fairy-realm boards need no special handling and no separate medium — the most spoiler-dense artifacts
the project will produce sit behind the same repo boundary as everything else. Josh's own consumption
preference is arguably his to state, and the counter-consideration is recorded fairly: this concerns
how he personally reads the output. The protocol's answer is that he redirects a ruling he dislikes
in one line, which costs less than a brief he never asked for. If he wants a richer rendered surface
later, the markdown-plus-files shape is the correct SOURCE for one either way, so nothing is
foreclosed.
---
The standing rule is that every artifact needs a CONSUMER and a DOC_MAP row, and a doc is not done
until it is listed there.
registry, check_realm_ruleset.py, read_boards.py, the art bible template and each ratified bible,
the concept-art translation doc if adopted, and the rights vocabulary. Each row names its consumers
and its status tier.
through 5 enter as ranks there, so the realm program competes for window time on the same ledger as
everything else rather than running beside it. The gated items enter
docs/ROADMAP_POST_5090_TO_SHIP.md instead.
the seven lenses to the proposed-new list, add the nine widen notes to the affected charters, and
add a realm block to the execution plan naming the doctrine wave and its freeze gate.
fold with honest teachers and verdict classes; update entry 22 (the doc-stub cluster) and entry 23
(the homeless marriage and partnership system) as the Personal Dimension row is decided.
architecture, R-A5 pilgrimage traversal, R-A15 per-realm art direction, R-A17 the awe beat) gain
RUBRIC ROWS at its landing, per its own §8 aspect_id join contract. The six RB-RLM rubrics at 2.7
register there as advisory reads, and the three runtime conformance tests register at the
sampled_invariant gating tier.
partnership system, and for the twelfth ruleset row if adopted.
resolved.
nine gorgeous realms from having ordinary combat in front of them.
asset_routing — populated by this program and by nothing else, through the sanctioned writer, and
each per its OWN declared state at §3.1 (asset_routing declared-empty pending its enum
ratification; care_tier reconciled against its ruled derivation; environment_grammar_refs
cited-or-empty, where empty stays a legal verdict). Plus the proposed generation_class column once
Wave 2 mints it, declared on the extension rail with this program as its single owning system.
the world it sits in are joined rather than merely adjacent.
the proposed generation_class field, the optional realm_ruleset_ref on the spine site grammar, and
every Zone Catalog extension column it populates — rides the existing rail and does not rebuild it.
docs/BUILD_PLAN_END_TO_END.md states the rule: "The registry-extension rail (ALREADY BUILT — use
it, do not rebuild it): docs/registry_extensions.json single-system column ownership +
docs/fidelity_baseline.json added_columns/extension_hashes + harness/check_registry_extensions.py +
registry_fidelity.py --emit-baseline ... Every pending column add in every lane should ride this
rail, and they should be BATCHED into as few re-baseline commits as possible." Concretely: each new
column is declared in docs/registry_extensions.json attributed to the ONE system that owns it (this
program owns generation_class and the three Zone Catalog outputs), lands in
docs/fidelity_baseline.json as an added_column with its extension hash, and is enforced by
harness/check_registry_extensions.py, which fails both directions — a manifest column that is not a
real added_column, and an added_column owned by no system or by two. The batching instruction binds
too: this program's column adds go in as few re-baseline commits as the wave structure allows,
not one commit per column.
value-encoding grammar rules that extension columns may be appended "ONLY by the single sanctioned
writer harness/apply_zone_catalog_columns.py (never by hand)." Every Zone Catalog write this
program performs — care_tier, environment_grammar_refs, asset_routing, and generation_class once
minted — goes through that writer. A hand-populate violates a ruled constraint and trips the
header/row lockstep check, which hard-fails the reader for every downstream consumer, not just for
the edited page. The writer's --populate-empty mode is the correct vector for filling an EXISTING
column: it is all-or-nothing per pack, refuses a nonexistent column, refuses to overwrite a
non-empty cell, refuses an unknown zone id, refuses any header change, and re-running the same pack
is a byte-for-byte no-op. Run --dry-run before --apply, always.
trusted. Any registry edit refreshes docs/fidelity_baseline.json in the SAME commit or the gates go
red.
---
These are not resolved here. Each is a real disagreement between the four analyses or between an
analysis and live canon, and each is named so the program cannot inherit it silently.
against the directive and says that if the program lifts one thing first it is the crossing. The
harmony analysis reads the ruled shape as GOOD — a plateau, a trough, a climb-out, a ceiling — and
says a naive make-each-realm-bigger architecture would contradict ruled canon. Both are right about
different axes; the disagreement is one of POSTURE, and it is briefed at 6.2.
monotone axis canon already gives the program, stated at Ch 66 and repeated at Ch 72. The harmony
analysis reads the ruled per-site values and finds they are explicitly NOT monotonic — Ch 70 and 71
drop every site to low. Both readings cite canon. The likely reconciliation is that the doctrine is
a NARRATIVE frame and the site values are the LOCAL truth, but the program should not assume that;
it is a real tension inside the spine and the ruleset's T-RLM-CURVE tooth is where it gets settled.
art pipeline places Annwn in the historical-or-literary column while noting it sits closer to
elevated than to Hades; the lens audit places it in the LIVING tier with practitioners today. This
changes whether Annwn's Panel A may be generated at all. The underlying question is FACTUAL — does
the Welsh Otherworld tradition have practitioners today — not a judgement call, so it needs no
Josh input. L-R3's named output is the designated settler and it must run before Annwn's first
board. Until it lands, the disagreement is carried as DATA rather than as prose: Annwn's
generation_class is unsettled, which the §3.3 gate treats as living, so the strict path is the
default and no board can be generated against an open question by accident.
routes Atlantis as wholly invented and therefore Lane-G-legal on both panels. The inventory and the
lens audit both record an ELEVATED anti-fringe duty on it. This was carried as an analyst
disagreement and it never was one: BOTH ARE RIGHT, and they are facts about DIFFERENT THINGS. The
defect was in this charter's own keying — one ratified two-value column asked to carry a care band
and a generation-legality axis at once, which it cannot, and widening a ratified column to make it
fit would have been a schema mint dressed as a bugfix. Resolved at §3.3 by splitting the key: the
ratified care_tier (standard · elevated) stays untouched and keeps its declared derivation, and the
legality axis becomes its own proposed closed-enum field, generation_class, on the extension rail.
Atlantis is then generation_class invented AND care_tier elevated with no contradiction, and the
costume tooth reads the second column, not the first. Recorded here as CLOSED rather than deleted,
so the reasoning survives for the schema mint at Wave 2.
explicitly REJECTS extending the per-feature environment-grammar registry. The art pipeline names
that same environment-grammar registry as the sharpest fit in the repo for the physics doctrine and
says an art-bible field that cannot become an EG_ row is a field describing scenery. The
reconciliation this charter adopts is a chain rather than a choice — the per-world ruleset row is
the authority and it SCATTERS into per-feature EG_ rows — but the two analyses did place the
load-bearing home differently and the schema mint must settle it explicitly.
machine on the easy and the hard case. The inventory names Ch 74 Yggdrasil the highest-leverage node
in the whole program and says lift the crossing first. This charter takes both (pilot the machine at
66 and 69; give 74 the strongest signature law and the highest awe ordinal), but the two documents
genuinely disagree about which realm gets the program's first serious attention.
contradiction to a director call. The art pipeline argues boards are not player surfaces, so a
fairy-realm board built during the road leaks nothing, and recommended published pages for
image-density. The harmony analysis states flatly that realm concept art is spoiler-tier and does
not enter any pre-Ch-66 surface including marketing and gallery contexts. An earlier draft split
the difference — pages for the road, local files for the fairy realm — which was a compromise
rather than a resolution and left the road's own boards on the publishable side of a line the
harmony analysis draws around all realm concept art. The ruling dissolves it: boards are versioned
in-repo artifacts, nothing is published, so the harmony analysis's constraint is satisfied for
EVERY realm and not just the fairy fold, and the art pipeline's image-density concern is answered
by rendered files rather than by a published surface. No per-realm exception is needed and no
spoiler boundary is drawn twice.
relevance as the cheapest demonstration of the doctrine. The lens audit says it is ALREADY covered
by five rostered lenses and that reporting it as a gap conceals coverage. Both are true of different
layers (design coverage versus spec population), and the charter's proposed twelfth ruleset row is a
bet that the design half is worth carrying inside this program rather than deferring to the
post-book-processing spec schedule.
inventory flags that school debuts stop at Ch 69 by ratified decision, so mechanical novelty
front-loads on the road. The harmony analysis proposes one signature law per realm across all nine
without noticing the decision. The charter's synthesis at 1.4 — the signature-law lane REPLACES the
school-debut lane for Ch 70-74 — was an inference needing verification, and it has now been checked
against D-RR.5 as it actually reads at CVD line 1395: "The SCH_010-013 threshold-school debuts stay
in order at Ch 66-67-68-69 ... The five added realms (Ch 70-74) showcase and extend existing
disciplines and author no new schools — the 13-school count and the 120-gate mastery scaffold are
unchanged (ratified D-RR.5)." The synthesis is COMPATIBLE: a signature law is a realm-local physics
rule, not a discipline, and it touches neither the 13-school count nor the 120-gate scaffold. The
bound that keeps it compatible is now recorded IN THE SCHEMA and armed by T-RLM-COMPOUND — a
signature law never grants a mastery gate and never confers school membership — because that is the
single way this lane could drift into minting schools by another name.
---
THE OBJECTION. The realms build LAST — after the earthly regions, after the Ch 2-5 AAAAA slice Josh
is waiting on, after the 5090. This charter proposes seven new lenses, a new registry, two new
harness tools, a new translation doc, a rights vocabulary, four critic rubrics, an art bible per
realm, and a nine-realm repair wave, for a surface nobody will build for a long time. That is a
program competing with the only work Josh can actually play, and the roster already carries
thirty-one unrun lenses.
THE ANSWER, in four parts, and the objection is right about one of them.
all the DR-2 tiered-asset request contract freeze soon, and the tiered-asset contract is gated by a
presentation lens that did not know the realms are the ceiling-setting art surfaces. If that
charter amendment missed, the asset contract would freeze against earthly-region assumptions and
the realms would inherit a tier never sized for them. That amendment WAS the highest-urgency item
in this charter, it cost nothing, and it is now DONE — applied verbatim at
docs/proposals/COMPARATOR_LENS_PROGRAM.md WAVE P1-C, per §5.1, and awaiting only its commit. The
argument stands as the reason the freeze axis is the right one to plan against; the item itself is
discharged, and the highest-urgency REMAINING items are the doctrine wave and the care tiering.
the Duat without it means generating images of living traditions' cosmologies against an unwritten
standard and critiquing them after they exist. Bounding a prompt costs nothing; regenerating a
realm's art costs a wave.
edits, the vocabulary reconciliation, the Wave-3 world repair, the rights vocabulary, and Lane R
plate assembly all run on local compute and desk time. The Wave-3 repair in particular is not realm
work at all — it is nine substrate gaps, a missing index block, four stale artifacts asserting a
retired structure, and ten region rows at a third of their fields, all of which are canon-integrity
debt that will be paid eventually and is cheaper to pay before authoring than after.
compete with the Ch 2-5 slice. They ride waves that are already scheduled, and if a window is tight
the correct cut is the P2 craft lenses and the seven non-pilot boards — never the doctrine wave, the
freeze-critical charter amendment, or the care tiering, because those three are the ones whose cost
goes up if they are deferred.