pipelines/LANE_world-terrain-interiors.md
Verbatim from workflow wf_d0ca605f-891. Each finding carries either its own register entry id or
the cross-lane merge that absorbed it. This file is the evidence of record the compact register
entries cite.
T1_Region_Page_Template [ACTIVE v2.0] §Section 17 (L252-253) declares the required fields as "environment areas, interiors, vegetation palette, NPC population…". docs/ROADMAP_POST_5090_TO_SHIP.md L489 (the factory machinery block) makes it a per-chapter obligation: "the AMPLIFICATION doctrine applied (reality is the floor: under-sized settlements expanded and interiors deepened, guided by the culture's own architectural grammar)" — ruled by Josh, docs/PRE_5090_BUILD_PLAN.md L2069-2086 ledger 2026-07-25k. CVD L25 makes architecture one of the documented materials "surfaced through gameplay rather than narration".area_bounds cell — ethiopia the_village_debtera "the mourning household interior 14×10"; south_india wootz_forge_and_herbalists_room "two workable interiors ~14×10 each" and kerala_kalari "treatment room and oil kitchen ~60×45"; swahili_coast kilwa_kisiwani "market and Sufi-quarter interiors 220×180" and dhow_open_sea "deck set 45×14 with the rigging traverse above it"; west_africa abomey_royal_court "the night-court interior 45×35" and ile_ife_bronze_city "foundry and diviner's-compound interiors 180×140"; south_africa witwatersrand_kimberley_workings "labor floor 180×120, superintendent's office deck 40×30"; bali besakih_mother_temple "lontar library interior". Section 17 prose on the newer pages already names reusable KITS (ethiopia: "One reusable interior kit on those three names covers the eleven Lalibela churches, the Aksum precinct, and the village church"; south_india: "The kalari is the region's one hero interior and it is fully specified"; west_africa: "Needs the workshop interior kit, the diviner's compound… the barracoon and ledger-house interiors"). Nothing structures any of it: a header scan of all 62 registries returns ZERO columns matching interior/indoor/room/building/architect — POSITIVE CONTROL: the same scan did match impossible_architecture_budget (T0_Realm_Ruleset_Registry), architecture (T0_Research_Corpus_Registry) and weather_cycle_profile/terrain_feature_class_ids (T0_Region_Index), so substring matching worked. docs/ASSET_DROPIN_CONTRACT.md §2 enumerates exactly eleven asset classes and the only environment row is "Vril-site / prop / environment mesh → T0_Vril_Site_Registry.mesh_id_ref" — one static mesh per site, no architecture or interior class. Content/Maps/ holds 9 .umap files, all region or arena (AngkorKhmer, Bali, BorobudurMerapi, FloresIsland, GreyboxArena, LiangBua_TerrainSmoke, PalembangMusi, SamosirToba) — no interior level or sublevel exists. docs/FACTORY_CONTRACT.md FR-033 is the only rider that mentions the doctrine and its source column is doctrine_only:docs/ROADMAP_POST_5090_TO_SHIP.md with verified-by NONE(the slope assertion and navmesh continuity teeth are game-repo side).docs/PRE_5090_BUILD_PLAN.md L50-111) owns interiors; a full grep for interior across PRE_5090_BUILD_PLAN_VOL2, ROADMAP_POST_5090_TO_SHIP, 5090_SETUP_RUNBOOK, FACTORY_CONTRACT and MILESTONE_BUILD_PIPELINE returns exactly ONE hit, the doctrine sentence above. The register's GTA row lists "Enterable interior density" as DIVERGENT-BY-DESIGN (docs/DESIGN_GAP_REGISTER.md L214) — but that is a bare one-line verdict about GTA-style AMBIENT urban density, and it is being read as cover for the authored interiors the pages already require.asset_concept/morphology/asset_category/mesh_tags across the 15-node slice) must classify assets in a taxonomy that has no interior class; and the Ch 6-13 per-chapter STAGING waves (PRE_5090_BUILD_PLAN_VOL2 §WS-4+) reach the first interior-bearing region at Ch 7 (wootz forge, kalari) and Ch 8 (Kilwa) with nothing to build against. Also blocks the P2 generative asset waves, which dispatch off registry rows that do not exist for this class.T0_Interior_Space_Registry (interior_id · parent zone_id FK · culture/build-grammar ref · footprint + storey count · enterable|referenced · room/segment count · care_tier · kit_id · the standard assetgen block) + add interior_authored to the zone_space_class enum + a twelfth asset-class row in ASSET_DROPIN_CONTRACT §2 + one FR rider with a real verified-by; the engine contract belongs in a new docs/translation/T99_Translation_Interiors.md or a §Interiors block in T99_Translation_UE5_Build.docs/PRE_5090_BUILD_PLAN.md L2082-2083 (the AMPLIFICATION doctrine): "caverns/ruins accurate at the mouth then vaster and more complex below." CVD §3 Component 6 (L92) makes interior volume constitutional: "Internal structures harness vril through resonant geometric-substrate mechanisms", with the canonical anchors including "Ch 58 Antarctica (underground facility)". docs/translation/T99_Translation_Terrain.md §5 item 1 (D-MPV, Josh 2026-07-21) reserves the 1 m/vertex hero tier explicitly for "caves, boss pockets, ritual grounds".docs/pipeline_review/tech_research/GEODATA_TERRAIN_CHAIN.md L215 states the same: mesh terrain "can represent overhangs/caves/arches", and L508 defers it. docs/ENGINE_OPTIMIZATION_DOCTRINE.md mentions "cave systems" exactly ONCE (L1020) and it is a lighting aside. T0_Vril_Site_Registry (42 rows) carries 15 rows whose site_class is inherently an interior — pyramid 6, megalith 5, tomb 3, underground_facility 1 — and has ZERO spatial columns (28 columns: site_id…notes plus the assetgen block; its only geometry field is a single mesh_id_ref). POSITIVE CONTROL on the zero: the same registry read did return populated site_class, vril_function and mesh_id_ref values, so the columns were readable. terrain=cave_system appears on 1 of 84 zone rows (flores liang_bua_cave). And the tooth that certifies space is DEM-only: Humanity/QA/SPACE_AUDIT_FLORES_POSTSCULPT.md L22 grades liang_bua_cave | Ch2 | 220x180x60 karst interior | 0.0 | 0.0 | 100% | OK after the sculpt pass, while L13 discloses "COP30 is a ~31 m ground sample… a sub-31 m cliff or a graded pad is invisible here" — the pass measures the slope of the ground ABOVE the cave, and docs/PRE_5090_BUILD_PLAN.md L2020 records the fix as "Liang Bua 21.3°→0.0° with a level cave-mouth terrace replacing the 27° opening".terrain_type→PCG-template routing (plan L2871, BUILD-NOW) will need a cave_system template, and a PCG scatter over a heightfield cannot produce an enclosed volume. Checked the queue: no ranked item, no Vol2 item, and no FR rider owns below-grade space; the only cave hits in either plan volume are the Ch-2 CAVE_BEAR tutorial (Vol2 U11.27) and the terrace sculpt note above.CLAUDE.md model-seats block) — Liang Bua cannot be finished as a terrace. Immediately upstream of that, the geometry fork (Mesh Terrain re-target vs authored mesh-kit/level-instance interiors) must be ruled BEFORE the 5090 asset waves start dispatching architecture, because the two answers imply different asset pipelines. Also blocks Ch 12/13 blombos_cave and the Ch-58 underground facility.below_grade_depth/vertical_extent/storey_count on the new interior registry + spatial columns on T0_Vril_Site_Registry, plus a W-SPACE tooth extension: any zone with terrain=cave_system or an interior row must assert an interior volume exists, and DEM slope must not be accepted as its evidence.dhow_open_sea = "open-water arena ~3 km run, deck set 45×14 with the rigging traverse above it | P1 (the boss set-…)" and coral_reef_hazard_zones = "reef-crest run ~2.6 km… warded-threshold pocket 120×9…", gameplay functions "combat, puzzle, vril_management", ambient_threat HIGH; angkor_khmer west_baray = "West Baray reservoir ~7.8×2.1 km (earthen dikes ~12 m) + West Mebon island-temple pocket ~100×100 + reversing-flo…". Ranked item 27 (docs/PRE_5090_BUILD_PLAN.md L108) is a LIVE rank whose deliverable is "CH_05 sub-boss real mechanics (ledger decode, chain meter, baray-flow strip)". 11 of 84 zone rows carry terrain=water_body.Humanity/TODO_CANON.md L395-398 — "FloresIsland.umap now carries the FloresOcean placeholder plane… the real water pass (actual water system/material, shore interaction) remains open"; L638-642 — "MV-5 — drowning NOT modelled… there is no air/breath/drowning clock… A drowning model + the real water system/material/shore-interaction (the 'real water pass') ride a later item." That later item does not exist: a grep for "real water pass" and "water system" across all of docs/ returns only Vol2's MV-5 absence line (L554) and one unrelated SIGNIFICANCE_LENSES mention of Harappan municipal water systems. No Water plugin is enabled (Plugins/ holds only NeoStackAI). No registry anywhere carries a water column — POSITIVE CONTROL: the 62-registry header scan that returned zero for water/ocean/river/lake/tide/depth in the same pass matched weather_cycle_profile and terrain_permanence_default, so the scan was live. SEAM_TYPING_SPEC covers water only as T2 conveyance ("the player cannot steer", §0.2), which is inter-chapter travel and explicitly not within-chapter play.*_wc_water.png) and a swim volume, and as nothing else: no water-body rows, no depth/current/navigability/shore/seasonal-flood data, no breath model, no underwater traversal or camera, no reef/flood-pulse state. A named LIVE rank (27) is scheduled to build a baray-flow mechanic on a reservoir that has no water representation, and the Ch-8 boss set-piece is an arena on open water.dhow_open_sea) and the reef's high-threat combat/puzzle zone hit it in the Ch 6-13 staging wave. The absent breath/drowning clock also leaves MV-5 shipped in a state where the player swims indefinitely.T0_Water_Body_Registry (water_class · depth_band · current · navigability · seasonal_state · shore_type · FK zone_id), open a NAMED plan item for the real water pass in the W4/W-MOVE lane (it is currently owned by a TODO sentence), and route the breath/drowning rule into the ruled death-choice spec (docs/proposals/systems/DEATH_CHOICE_RUNTIME_SPEC.md).T0_Dungeon_Class_Reference CLASS A roster_cardinality = "exactly 20 (10 vril pieces + 10 health pieces)", per_instance_registry_row = TRUE (Josh, twelfth sitting — docs/PRE_5090_BUILD_PLAN.md L1386-1391). docs/proposals/systems/RIFT_DUNGEON_SCAFFOLDING_SPEC.md §3.2 declares "The rift interior is the invented eleventh register" and §9 arms T-RIFT-8 "the interior answerability tooth".registries/ for DG_nnn|DUN_nnn|RD_nnn|RIFT_nnn returns 0, while the identical scan shape for VS_CHnn_nnn|BE_nnnn returns 982 hits — the pattern form works, the dungeon id space is unminted. The registry that should hold them is an unresolved fork by its own admission: CLASS A instance_home = "the rift instance run-economy row (F-1's home)", and §6.2/§8 carry F-1 as an open DIRECTOR fork ("STEP 2, director: rule F-1, F-2 and F-7"). And the spec contains no spatial grammar: a grep of the full spec for room/layout/floor/chamber/corridor/footprint/geometry/size returns only unrelated uses (the rarity FLOOR, the answerability FLOOR, the chapter FLOOR per locus token) — there is no dungeon footprint, no room or segment count, no depth, no entrance geometry, no art kit.docs/DESIGN_GAP_REGISTER.md L499-518). None of the three asks for an instance row or a spatial spec, and none is a ranked plan item with an owner.[canon-author·L]) hits it first and soon: Fork C ruled the important crossings become real staged or open DUNGEONS (SEAM_TYPING_SPEC §0.2 L50), and the slice-scope reconcile scales rank 21 to "eleven seams instead of two, including the Indian Ocean crossing" (docs/pipeline_review/SLICE_SCOPE_RECONCILE_2026-07-27.md L408-409) — each typed against a dungeon class whose instances have no rows and no size.docs/PRE_5090_BUILD_PLAN.md L2081-2083): "under-sized settlements EXPANDED into fuller living places guided by their own cultures' architectural grammar", carried into the per-chapter factory at ROADMAP_POST_5090_TO_SHIP.md L489. docs/proposals/systems/base-building.md §3 (L77) sets the module BOM: "Trade-crafted modules are the building blocks — walls, roofs, floors, stations, defenses, decor — crafted from the §3.2 material economy."T0_Dwelling_Registry is a FALSE FRIEND at 2 rows, and both are PLAYER home classes: DW_0001 "Trade-Home (per trade hub)" and DW_0002 "Dwelling-Home (world-home)", source_anchor = base-building.md §trade-homes / §dwelling-homes, notes = "DRAFT anchor". Its build_grammar_ref is EMPTY on both rows. The grammar pointer is circular: T0_Trade_Registry.shared_build_grammar_ref = shared_trade_module_grammar on 2 of 10 trades (TRADE_01, TRADE_03), and docs/proposals/systems/WS7_WIRING_BUILD_PLAN.md L207/L234 declares that token resolves to T0_Dwelling_Registry.build_grammar_ref — the empty column. No module, structure, or settlement registry exists (POSITIVE CONTROL: the 62-registry header scan that found impossible_architecture_budget and architecture found no structure/module/building/settlement column). And the village FK is a phantom: T0_Chapter_Index.seed_village_ids reads [POPULATE→T0_Seed_Registry] on 59 of 79 chapter rows and NONE_DEFERRED on the other 20 — there is no T0_Seed_Registry among the 62 (only T0_Quest_Seed_Registry, 0 rows), the same defect class already documented for T0_Folklore_Entity_Registry in docs/proposals/PHANTOM_REGISTRY_ROUTING_FINDING.md.urban + 8 agricultural zones need populated settlements) and ranked item 11 (T0_Persona_Index + personas on every slice entity, P1 EXIT) which will author villagers whose homes have no type; LW-8 ambient-NPC presence/schedule (Vol2 W-LW3) needs the same houses to schedule people into.T0_Structure_Registry (structure_id · culture/region FK · structure_class dwelling|workshop|temple|monument|civic · rank/wealth variant · count-per-settlement · interior FK · assetgen block), give shared_trade_module_grammar a real artifact target, and repair seed_village_ids to point at a registry that exists in the same commit as its --emit-baseline.docs/proposals/systems/environment-combat.md §4 promotes weather to a first-class combat input — Model B is "the elemental-affinity matrix (ENHANCE / WEAKEN by element × terrain × weather)" (L24, L81), §4.3 "Weather as an arena/region cycle", and §290 asks to "Author the 69 Region Affinity Profiles… weather tables from climate rows". CVD L728 carries weather in the Tier-C world state. Ruled valleys-never-boring and the L1.5 ambient lane make time-of-day part of the felt world (docs/PRE_5090_BUILD_PLAN.md final-ascent rulings).T0_Region_Index.weather_cycle_profile is populated on 4 of 76 rows (dry;ash;clear · monsoon;rain;clear · monsoon;ash;clear · monsoon;flood_pulse;clear), all four with env_profile_status=stub; the 9-value vocabulary is declared only in WS7_WIRING_BUILD_PLAN.md L116 (rain|storm|fog|sandstorm|ash|monsoon|dry|celestial_window|clear). There is NO effect table: T0_Element_Effectiveness is 124 rows over exactly 5 columns — attacker_element, defender_element, effectiveness, attacker_axis, source — with no environment or weather axis, and ranked item 17 ("Complete + import the element matrix; lift the 10-element cap; per-enemy weaknesses") does not add one. And there is no per-beat or per-scene authority: the beat row grammar carries no time or weather field (docs/spine/CH_08.md L103 — "CH08_B01 · seq 1 · role hook · mode traversal · site dhow_open_sea · boss none · fork none"), and T0_Scene_Spec_Registry's 31 columns contain no time_of_day, lighting or weather field (POSITIVE CONTROL: the same read returned its real audio_priority_layer, mesh_refs and cinematic_status columns). T0_Chapter_Index.environmental_params is real prose on 20 of 79 rows and [POPULATE→T8_Region_Pages Section 17…] on 59.time_of_day and weather_required fields added to the SPINE_SCHEMA v3 beat grammar (§3b/§3d) and to T0_Scene_Spec_Registry, so the drivers and the capture critics both have an authored target.docs/BUILD_PLAN_END_TO_END.md §Top-3 (L46, L75) — one table add "flips FIVE lanes from BLOCKED to READY"; docs/PRE_5090_BUILD_PLAN.md T0.2 (L2773) records it DONE for "7 columns × 27 zones, cited-or-empty"; W4.8/RM.1 (Vol2 L1183) is the consumer — "worldcover_biome from the Zone Catalog routes the template; explicit fallback so an unseen biome never silently no-ops".worldcover_biome 0/84 filled, asset_routing 0/84, environment_grammar_refs 0/84, zone_space_class present on only 27/84 rows (the 8 Ch 6-13 pages carry a 17-column header without it) and every one of those 27 reads earth_geo; zone_bounds_wgs84 blank on 7 rows (central_africa_congo 5 of 8, rift_valley 2 of 8). POSITIVE CONTROL: the same parse returned populated meters_per_vertex, footfall_surface and care_tier on those rows, so the reader was working. harness/check_build_readiness.py L277-278 fails the UE5 lane on exactly this column — "Zone-Catalog worldcover_biome column (PCG Biome Core / auto-material key)". The WorldCover rasters themselves exist for the 4 built regions (Saved/Terrain/*_wc_forest.png, _wc_water.png, …), and docs/DOC_MAP.md L320 records the mechanism (zone_pack --populate-empty fills worldcover_biome from the W4.5 fetch) — it was never run.DESIGN_GAP_REGISTER entry 70 records the zone_space_class discriminator as "landed on all 27 live zones" (L679) while 84 zone rows are live, so the realm/space discriminator and its T-RLM-ZONE tooth are blind to two-thirds of the slice.derive_region_bbox→terrain fetch for those congo and rift_valley zones.harness/apply_zone_catalog_columns.py --populate-empty pass over all 12 pages (worldcover_biome from the existing rasters; zone_space_class extended to the 8 newer pages; the 7 bboxes cited) with --emit-baseline in the same commit, and correct entry 70's 27-of-84 arithmetic in place.docs/FACTORY_CONTRACT.md §1 states the design: "The checklist is DERIVED. A rider added here reaches every future chapter automatically… A landing that makes a per-chapter promise declares an OWNER: line and a rider row in the same commit. Gate 30 fires the day a promise lands uncited." The per-chapter machinery in ROADMAP_POST_5090_TO_SHIP.md L477-495 promises interiors deepened, settlements expanded and the amplification doctrine applied on every chapter.doctrine_only:docs/ROADMAP_POST_5090_TO_SHIP.md, verified-by NONE(...)), FR-039 (festival windows + ambient/SFX rows). POSITIVE CONTROL that the row set was fully read: the same listing returned FR-001…FR-070 including the class-scoped rows (class:boss, class:hard_line, class:rr_node). No rider keys interiors, dwellings, water, weather/time-of-day, or dungeon space; the only artifact reference for the amplification half is a doctrine URL, and the only executable teeth named are the game-side slope and navmesh assertions — both of which are DEM-surface measurements.DESIGN_GAP_REGISTER L511-518); the space classes above are not in it.harness/emit_chapter_checklist.py derives that node's checklist from the contract, so the first Ch 6-13 build silently inherits no interior, water or weather obligation. Every subsequent node repeats it.factory_contract.json mirror (interiors authored per chapter; the region's water bodies typed; weather/time-of-day authored per beat; dungeon space rows for any bounded interior the node opens), each with a real path:/registry: source and a named tooth rather than doctrine_only, landed with gate 30's R1/R2/R3 scan re-run.docs/ENGINE_OPTIMIZATION_DOCTRINE.md §3.2 decision 1 (L1014-1020) rules the consequence and names the deadline: MegaLights changes the authoring rule from "few lights, each cheap" to "as many lights as you like, but keep per-pixel light overlap low", and "it must be written into the art bible before the first region is lit… This is a design unlock for a canon dense with lantern-lit settlements, ritual grounds, vril sites and cave systems." L1918 repeats the clause.docs/ returns only forward-looking references (this doctrine's two clauses, the realm-wave head-of-art-bible notes in DESIGN_GAP_REGISTER L1334/L1746, and the continuo handoff), and a filename search for *ART*BIBLE*/art_bible returns nothing. Meanwhile four regions are already lit (FloresIsland, Bali, AngkorKhmer, SumatraJava maps with sun/sky/skylight per the W4 ledger, docs/PRE_5090_BUILD_PLAN.md L2521-2600). The doctrine's streaming and soak architecture is open-world only: §6.4's soak route is "one full traversal of the largest authored region… plus one combat set-piece with the densest authored light and VFX load, plus one dwell in the densest settlement" — no enclosed-space leg — and the World Partition / FastGeo rules (L984, RULE WP-5/WP-7) address actor streaming across terrain, never level instances or interior cells.T1_UI_UX_Spec + T1_Audio_Spec from the shipped surfaces; there is no equivalent ranked item for the lighting/art rule set.route_id to the §6.4 soak route. Boundary note: the visual/reference half of an art bible belongs to the concept-art lane — this row claims only the enclosed-space lighting and streaming rule.