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.
SCRIPT-EMITTED by harness/emit_region_consumption_manifest.py from the typed status fields
of the audited EXR records. Do NOT hand-edit. Regenerate instead.
CONSUMABLE (status queued) — the audit advanced this row. Authors MAY use it.
HELD (status proposed) — a critic MAJOR holds this row. Authors MUST NOT use it.
CUT (status rejected) — the audit cut this row. Authors MUST NOT use it.
Consuming a HELD or CUT row is a hard authoring defect, not a judgement call.
EXR_SRC_00157
audit verdict: RATIFY-WITH-CUTS
extraction_status: deep_read_audited
[CONSUMABLE] STORY SEED EXR_SRC_00157/S01 — Reading the keris — the three measurements
[CONSUMABLE] STORY SEED EXR_SRC_00157/S02 — Tumpek Landep — the day steel is blessed
[CONSUMABLE] STORY SEED EXR_SRC_00157/S03 — The auspicious-day layer (Pawukon dewasa)
[CONSUMABLE] STORY SEED EXR_SRC_00157/S04 — The Kajeng-Keliwon night when the leyak walk
[CONSUMABLE] STORY SEED EXR_SRC_00157/S05 — Straight-line demons and the aling-aling geometry
[CONSUMABLE] STORY SEED EXR_SRC_00157/S06 — The net-tent over spilled blood
[CONSUMABLE] STORY SEED EXR_SRC_00157/S07 — The caru — scaling an appeasement to the trouble
[CONSUMABLE] STORY SEED EXR_SRC_00157/S08 — The balian's branching craft (a healer skill-web)
[CONSUMABLE] STORY SEED EXR_SRC_00157/S09 — Consulting the taksu medium
[CONSUMABLE] STORY SEED EXR_SRC_00157/S10 — The freed Dalang, staged as the gods' own remedy
[HELD] STORY SEED EXR_SRC_00157/S11 — The balance-trial as a fragment of the calonarang
[CONSUMABLE] STORY SEED EXR_SRC_00157/S12 — Barong's protection — the trance where the blade will not bite
[CONSUMABLE] STORY SEED EXR_SRC_00157/S13 — Tirta Empul and the poisoner's spring
[CONSUMABLE] STORY SEED EXR_SRC_00157/S14 — The kanda-empat guardians (and how a leyak is made)
[CONSUMABLE] STORY SEED EXR_SRC_00157/S15 — Folding the canang and raising the banten-tegeh
[CONSUMABLE] STORY SEED EXR_SRC_00157/S16 — The bajra-uter, the bell that sings only by accident
[CONSUMABLE] STORY SEED EXR_SRC_00157/S17 — Filing the six enemies (matatah coming-of-age)
[CONSUMABLE] STORY SEED EXR_SRC_00157/S18 — The Uluwatu cliff — thieving monkeys and the winged gates
[CONSUMABLE] STORY SEED EXR_SRC_00157/S19 — Nyepi eve — wake the demons, then play dead
[CONSUMABLE] STORY SEED EXR_SRC_00157/S20 — The cremation that whirls the spirit lost
[CONSUMABLE] STORY SEED EXR_SRC_00157/S21 — Kotekan interlock and the shimmering ombak
[CONSUMABLE] WEAVE LINK EXR_SRC_00157/W01 — Bali's origin-of-holy-water myth is the Indic churning carried in the Mahabharata itself, not a parallel in...
[CONSUMABLE] WEAVE LINK EXR_SRC_00157/W02 — Kala Rau is Rahu by descent, the name and the beheading-by-the-discus story carried intact into Bali; a rec...
[CONSUMABLE] WEAVE LINK EXR_SRC_00157/W03 — Rangda is explicitly Durga carried into Bali and localised through the calonarang; naming the descent prote...
[CUT] WEAVE LINK EXR_SRC_00157/W04 — A shared cosmological category (earth borne on a turtle), useful as a resonance rather than a diffusion cla...
[CUT] WEAVE LINK EXR_SRC_00157/W05 — Two independent traditions cast an archer-god as the same double agent of disease and cure; a clean paralle...
[CUT] WEAVE LINK EXR_SRC_00157/W06 — Bali's religion arrived as an already-fused Siwaist-Mahayana import from Java; the link makes the Ch-5 west...
[CUT] WEAVE LINK EXR_SRC_00157/W07 — Lamination-for-pattern is an independently-arrived metallurgical art the player can compare across blade cu...
[CUT] WEAVE LINK EXR_SRC_00157/W08 — A shared category of egalitarian reform-against-caste, and a ready hook for the Mayadenawa reframe: the sec...
[CUT] WEAVE LINK EXR_SRC_00157/W09 — A shared personal-guardian-spirit category the arc can resonate against; here it is the honest substrate fo...
[CONSUMABLE] ENTITY EXR_SRC_00157/EC_01 — Bedawang Nala (the world-turtle)
[CONSUMABLE] ENTITY EXR_SRC_00157/EC_02 — Naga Basuki and Naga Anantaboga
[CONSUMABLE] ENTITY EXR_SRC_00157/EC_03 — Kala Rau (Kala Rauh)
[CONSUMABLE] ENTITY EXR_SRC_00157/EC_04 — Rangda (Uma / Durga / Kali in widow-form)
[CONSUMABLE] ENTITY EXR_SRC_00157/EC_05 — Barong ket (Banaspati Raja / Wisnu-form)
[HELD] ENTITY EXR_SRC_00157/EC_06 — Rarung (Kalika) and Jero Luh (Anak Lingsir)
[CONSUMABLE] ENTITY EXR_SRC_00157/EC_07 — Leyak (the night-adept)
[CONSUMABLE] ENTITY EXR_SRC_00157/EC_16 — The pusaka keris line (legendary blade)
[CONSUMABLE] L3 CONNECTION EXR_SRC_00157/L3_01 — The whole island runs on rwa bineda / Tri Hita Karana — the three interdependent forces (order, chaos, and ...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00157/L3_02 — The subak water-temple system is documented cooperative irrigation engineering and collective-action govern...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00157/L3_03 — Pamor keris-making is real materials science: crucible reduction of iron, carbon content within a half-perc...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00157/L3_04 — The Pawukon is a genuine combinatorial calendar — ten concurrent weeks (1..10 days) generating 55 day-names...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00157/L3_05 — The gamelan's shimmering ombak is real acoustics — paired instruments deliberately tuned a hair apart produ...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00157/L3_06 — The Kala Rau eclipse-myth encodes the real mechanism of eclipses — the sun or moon periodically 'swallowed'...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00157/L3_07 — The book explicitly distinguishes the historical Aryan cattle-herding migration into India from 'the Aryans...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00157/L3_08 — The kanda empat (the buried placenta and the three vanished birth-fluids) is the one substrate power that b...
[REVIEW] 8 civilizational_marks carry NO status field. Read this
record's audit_summary.care_findings in full BEFORE consuming any mark — a mark
cut by the audit lives there and nowhere else.
audit care_findings (18) — binding on every author of this record
[MAJOR] weave_links W04, W05, W07, W08, W09 — Five of nine links fail the 4.1 Frame-A bar that both ends be NAMED and CITED. Four (W04, W07, W08, W09) target culture cross_cultural with a motif abstraction (a widely independent creation motif; pattern-welded / Damascus blade-craft; reform movements rejecting birth-caste; the attendant guardian-spirit / genius / familiar); W05 names a culture (greece_aegean, Apollo) but carries no citation for that end - p.250 is Eiseman on Rudra and never mentions Apollo. A link to an abstraction cannot be checked for hyperdiffusion because it has no second end to check. All five are BLOCKED at the weave lane; only W01, W02, W03 advance. Each needs one named counterpart culture with its own citation. Two additionally mis-typed independence and were retyped catalog-convergence: W05 (two Indo-European traditions with a live shared-inheritance hypothesis) and W08 (Bujangga Wesnawe, Buddhist and bhakti currents all share the Indic religious field, so they are not independent arrivals).
[MAJOR] weave_links W06 (Siwa-Buddha alloy to java_buddhist) — Chronology and direction defect. The target anchor reads the Majapahit Siwaist-Buddhist fusion carried to the Borobudur/Java crossing, conflating strata roughly 450 years apart: Borobudur is 9th-century Sailendra Mahayana, while the Siwa-Buddha fusion and the Suthasoma's bhineka tunggal ika are 14th-century Majapahit (Mpu Tantular, c.1365). The direction is also inverted - the record's own why_it_matters says Bali's religion arrived as an already-fused import FROM Java, yet Bali is authored as the source end; the source corroborates independently at print p.152, dating Majapahit's arrival in Bali to about a century after the 13th-century Panataran reliefs. Third consequence: bali.md fixes the present frame at ~1200 CE with the Majapahit stratum reachable only through forward time-fluctuation windows, so a Ch-5 crossing meeting the Majapahit fusion is out of frame as written. Fix: Java as source end, a Majapahit-era target, Borobudur routed as a separate earlier Sailendra link.
[MAJOR] entity_candidates EC_10 (Indra), EC_11 (Rudra / Siwa Mahakala), EC_13 (Dewa Ruci) — Three revered deities of a LIVING religion are routed to target_registry T0_Boss_Encounter_Registry while their own portrayal notes describe a quest-giver or rescuer, an honored ambivalent patron never flattened into a monster, and a mystery-school revelation figure. The record contradicts itself: EC_04 (Rangda), EC_05 (Barong) and EC_15 (Sanghyang Kumara) route to T0_Creature_Roster with target_species_class deity. The v0.2.3 enum correction normalized the registry token without questioning whether a boss registry was the right destination. If merged as-is, Indra and Dewa Ruci become boss rows - deities are portrayable and usable per canon, but these three are authored as allies and patrons, not combat targets. Re-route to T0_Creature_Roster with target_species_class deity before merge. EC_03 (Kala Rau) and EC_09 (Mayadenawa) are correctly in the Boss registry and must stay.
[MAJOR] entity_candidates EC_09 (Mayadenawa) portrayal_note and care_flag — The 17.1 reframe instructs: route the villainy through the defaming institution, not the man - where the institution named is a jealous rival priesthood. Executed literally this stages a living Balinese priestly lineage (the Brahmana pedanda) as the villain, which is precisely what 17.1 COLLECTIVE protection forbids, and it collides with this chapter's protected ally Pedanda Ida Bagus Mantra (CHAR_0023) and with bali.md Section 4's posture that the pedanda are protected. The motive is sound (a living reform sect's revered ascetic should not be flattened into a demon-king), but the fix as written protects one living community by defaming another. Recommended: contested-memory framing - two living traditions remember the same figure differently, both rendered with weight, the ambiguity held - while villainy stays on the fictional certification apparatus (NW_0002) the chapter already provides as its institutional proxy. He remains stageable as a boss.
[MINOR] story_seeds S12, S16, S20 (seed-grain care_flag) — Three seeds touching living rites or consecrated implements carry care_flag none while the record-level care_summary flags exactly this material: the babuten trance keris-to-the-chest (care_summary: sacred trance-invulnerability with taboo-consequence, never gore-spectacle), the bajra prayer-bell (the pedanda's consecrated ritual implement, framed by S16 as a player-earned masterwork/tool), and the ngaben cremation rite (a living death-rite of real families). The seed is the unit that travels into docs/proposals/quest_seed_queue/bali.md, so record-level flags do not follow it. Copy the handling notes down to seed grain; for S16, frame the bell as a smith's masterwork held in the rite rather than a looted power-item.
[MINOR] l3_connections L3_01 — The row welds two distinct named concepts of a living tradition into one: rwa bineda / Tri Hita Karana - the three interdependent forces (order, chaos, and the balancing centre). The three-force gloss is source-faithful for RWA BINEDA (verified at print p.2, where Eiseman describes the polarity as including a third centre position balancing the other two), but Tri Hita Karana is a different framework - the three causes of well-being, the relations to the divine, to other people and to the land - as bali.md Section 4 correctly states and as the UNESCO subak dossier (SRC_00251) uses it. Split the two names before any consumer reads the row, or the wrong definition of a living, UNESCO-recognized framework propagates through the design layer.
[MINOR] entity_candidates EC_12 (Empu Pradah) descriptive_substrate — The substrate says the sage turns her own black magic against her to restore balance, leaning on the literary Calon Arang ending in which the widow is defeated and released - in tension with the record's own never-slain Rangda guardrail asserted at EC_04 and S11 and ratified at bali.md Section 5 and BE_0004_S3. The source's own Barong-drama fragment (verified at print p.296-298) is the no-death version in which the two make up. Frame the sage's counter as restoration of balance, never a defeat-and-kill, so the two rows cannot be read against each other downstream.
[MINOR] story_seeds S18 (seed_kind) — seed_kind is vehicle_or_mount but the content is a cliff agility/traversal course, a monkey-troop theft hazard and a temple-gate architectural read - no vehicle and no mount. The stray trigger is the phrase a potential taming-line encounter about a crab-eating macaque. seed_kind is a typed enum that routes the seed downstream, so a mis-kind mis-files a traversal set-piece into the mount framework. Re-kind at merge (encounter, or a traversal/immersion kind); the audit does not own the field.
[MINOR] entity_candidates EC_01, EC_02 (entity_kind) — Bedawang Nala and the two cosmic nagas are typed entity_kind creature, the T0_Creature_Roster's real-animal kind, though they are cosmological beings. bali.md Section 3 explicitly routes folkloric beings to Creature_Roster.folklore_attribution, and the record itself uses folkloric_being for the leyak and bhuta-kala. Recommend folkloric_being at merge. Schema 11.2 interaction worth recording: as creature they would be culture-exempt, so a mis-kind here silently exempts a culturally-owned being from the culture gate - I filled culture on both regardless, which 11.2 permits on any kind.
[MINOR] weave_links W08 and entity_candidates EC_09 (scholarly basis for living-community claims) — The claim that the Bujangga Wesnawe read Mayadenawa as their own maligned ascetic is a statement about a LIVING sect's self-understanding, and the caste material is a statement about a living social system - both resting on a single 1990 trade-press ethnography with no academic apparatus. The record's care_summary already flags the caste handling. These two claims should carry a scholarly anchor before they harden into registry rows or drive a chapter's antagonist framing; they are the highest-consequence claims in the record for 17.1 purposes.
[MINOR] story_seeds S01, S02, entity EC_16, weave W07 (keris material vs the ~1200 CE present frame) — Verified at print p.152: the source dates the earliest firm records of the kris to 13th-century relief carvings at Candi Panataran in East Java. bali.md fixes the present frame at ~1200 CE, so the keris substrate sits at or just past the frame's edge on the source's own evidence. This is not an assertion of anachronism - keris-class daggers plainly predate the earliest dated relief - but the region page's Section 6 keris claim now has a datable boundary attached to it, and the deep-read's own evidence should be recorded rather than lost, since four rows depend on it.
[NOTE] l3_connections L3_02 (phrasing) — The row carries an island ordering itself without a single hand forcing it, quoted from bali.md Section 7 where it describes decentralized subak governance. It is not a Velheim reference and the row is design-layer, but it is lexically the banned who-frame construction (a negated agent-noun), the exact pattern a prior Range Review caught leaking into spine prose. Keep it out of any player-facing surface text.
[NOTE] weave_links W02 (missing canon link) — The Balinese Kala Rau telling pursues Dewi Ratih specifically, and bali.md Section 15 already claims a Moon-Goddess link toward Ch 7 (LINK_0135). The weave omits the pairing the region page has already made, losing a free reciprocation.
[NOTE] entity_candidates EC_12 (Pande clan origin) — The identification of the calonarang sage with the Pande clan's proclaimed first head and sole object of worship is a clan-origin claim about a LIVING clan (the record cites the Pande's own lontar, Prasasti Sira Pande Empu). Verify against that lontar or a specialist source before it becomes a T0_Character_Index row; a wrong ancestor attributed to a living clan is a 17.1 misattribution, and he must never be villainized or played for comedy.
[NOTE] story_seeds S17 (matatah) — Tooth-filing is a Balinese Hindu rite of passage for members of the community. The assimilation-integrity mechanic is ruled canon, so the player undergoing it is not out of bounds, but the honest default surface is witness-and-serve for a peer, with undergoing gated on genuine acceptance into a household - which also makes the integrity re-weighting mean more. The seed already offers both; make witness the default.
[NOTE] story_seeds S18 and bali.md Section 7 (Uluwatu) — Verified at print p.250 that Pura Luhur Uluwatu is one of the sad kahyangan, the six most important temples in Bali - a living sacred site of the highest rank. bali.md Section 7 already binds it to referenced composition rather than replica; keep the agility course on the cliff traverse and the approach thresholds, never vaulting consecrated shrines, and let the winged split-gates be read rather than clambered.
[NOTE] record-wide (scholarly basis) — All nine weaves are classified primary_ethnography: Eiseman is a first-hand, long-residence participant-observer and the book is the region keystone, but it is a 1990 trade-press essay collection with no peer review. Descriptive Balinese practice is well-supported; the comparative FAR ends (W01 Mahabharata, W02 Rahu, W03 Durga, W06 Majapahit) are the author relaying general Indological knowledge outside his fieldwork. bali.md Section 16 already names the right anchors - Belo (the 17.1-critical Rangda-ambivalence keystone), Lansing, Endicott - and Section 20 already gates build-readiness on the Belo acquisition. Those links should not harden before it lands.
[NOTE] record-wide (citation verification method) — Citations were verified against the source PDF, not assumed. The offset is print page = PDF page minus 13, and eight anchors across all four row classes resolve exactly: p.2 (rwa bineda), p.8 (kori agung, bhoma, candi bentar), p.62 (Kala Rauh, the churning, Tirtha Amertha), p.147 (balian practice), p.152 (kris metallurgy), p.250 (Uluwatu, Rudra), p.296 (Barong part II, calonarang, rangda=widow), p.307 (the panasar). Anchors are also topically monotonic against the book's own four-part contents (Religion; Ritual and Magic; Anniversaries and Temples; Performing Arts) and consistent across rows citing the same material. The record's page_coverage remains null.
mint_flags (14) — these are RULINGS-PENDING, never invented into the page
S01: A minigame/contest quest_type (blade-measurement and pamor-reading), plus a pande-forge ZONE absent from bali.md's Zone Catalog.
S02: A festival-or-daily-immersion quest_type (recurring observance with an upkeep effect).
S03: A planning/scheduling minigame quest_type (the Pawukon dewasa layer).
S04: A folklore_combat quest_type (the recurring Kajeng-Keliwon night event).
S05: A folklore_combat / environmental-tactic quest_type (aling-aling line-of-sight and the offering-trap).
S06: A ward/defense-objective quest_type (guard the blood-site net-tent three nights).
S07: A ritual-puzzle quest_type (caru scaling and the nawa-sanga colour-direction wheel).
S09: A divination-consult / regional_npc_dialogue quest_type (the taksu medium, consulted until readings agree).
S16: A trade-expression quest_type (the once-in-a-playthrough bajra-uter masterwork).
S17: A rite-of-passage / character-beat quest_type (matatah, the six enemies).
S18: A traversal-course quest_type (the Uluwatu agility run), plus the seed_kind re-kind noted in the findings.
S19: A recurring world-event quest_type (Nyepi eve and the day of stillness).
S20: An ambient-immersion quest_type, plus a cremation-ground / Pura Dalem ZONE absent from bali.md's Zone Catalog.
S21: A performance/rhythm-duet minigame quest_type (kotekan interlock against a partner).
EXR_SRC_00251
audit verdict: RATIFY-WITH-CUTS
extraction_status: deep_read_audited
[HELD] STORY SEED EXR_SRC_00251/S01 — Mayadenawa and the Poisoned Spring (Tirta Empul purification boss)
[CONSUMABLE] STORY SEED EXR_SRC_00251/S02 — The Subak water-allocation puzzle (proportional dividers + fallow timing)
[CONSUMABLE] STORY SEED EXR_SRC_00251/S03 — The water-rights arbitration
[CONSUMABLE] STORY SEED EXR_SRC_00251/S04 — The Rice Goddess harvest (ani-ani blade and the cili rice-mother)
[CONSUMABLE] STORY SEED EXR_SRC_00251/S05 — The tunnel-masters of Klungkung (Engineering seed)
[CONSUMABLE] STORY SEED EXR_SRC_00251/S06 — The rock-cut gorge of Gunung Kawi
[CONSUMABLE] STORY SEED EXR_SRC_00251/S07 — The self-ordering island: nested water-temple hierarchy + upstream/downstream water
[HELD] STORY SEED EXR_SRC_00251/S08 — The lava that stopped at the gate
[CONSUMABLE] STORY SEED EXR_SRC_00251/S09 — The stone that fell from the sky (forgotten-language literacy)
[CONSUMABLE] STORY SEED EXR_SRC_00251/S10 — Dewi Danu, Goddess of the Lake (disguised benefactor)
[CONSUMABLE] WEAVE LINK EXR_SRC_00251/W01 — Bali's sacred volcano-axis is genuinely the Indic Meru cosmology received and localized onto real volcanoes...
[CUT] WEAVE LINK EXR_SRC_00251/W02 — The local Tirta Empul myth is the Indic god Indra carried into a Balinese landscape legend; naming the rela...
[HELD] WEAVE LINK EXR_SRC_00251/W03 — Dewi Sri is a shared Javanese-Balinese agrarian deity, so the rice-goddess rite reciprocates directly into ...
[HELD] WEAVE LINK EXR_SRC_00251/W04 — The dossier itself frames Bali against the 'elsewhere in Asia the power of princes was associated with cont...
[CONSUMABLE] WEAVE LINK EXR_SRC_00251/W05 — The dossier states the candi 'show a strong link to the Classical states of Java'; the monument style is Ja...
[CONSUMABLE] ENTITY EXR_SRC_00251/EC_03 — Dewi Danu (Goddess of the Lake)
[CONSUMABLE] ENTITY EXR_SRC_00251/EC_04 — Dewi Sri (Rice Goddess)
[CONSUMABLE] ENTITY EXR_SRC_00251/EC_05 — Ida Sesuhunan Petali (the Forest God)
[CONSUMABLE] ENTITY EXR_SRC_00251/EC_06 — Bhatara Empelan (Deity of the Weir) / Ida Bhatara Pura Ulun Swi
[CONSUMABLE] ENTITY EXR_SRC_00251/EC_07 — Ida Batara Tengahing ring Segara (the Sea God of pestilence)
[CONSUMABLE] ENTITY EXR_SRC_00251/EC_08 — The Fallen Sky-Stone (Prasasti Manukaya)
[CONSUMABLE] ENTITY EXR_SRC_00251/EC_09 — Old Balinese / Kawi (Old Javanese)
[CUT] ENTITY EXR_SRC_00251/EC_10 — The 'Jalu' keris (River-of-the-Dagger blade)
[CONSUMABLE] L3 CONNECTION EXR_SRC_00251/L3_01 — Bali's water-temple network is a real-world case of emergent self-organization: with no central controller,...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00251/L3_02 — The Balinese 'ricebowl' is real geochemistry: the Batur caldera eruption (~23,670 BP) laid a mineral-rich i...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00251/L3_03 — You can date the spread of the subaks with DNA: a Y-chromosome study of 587 Balinese men shows the furthest...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00251/L3_04 — Regional pest control without pesticides: by synchronizing fallow-and-flood windows across large contiguous...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00251/L3_05 — A rite can outlive its own literacy: the Manukaya community washed the sky-stone on the full moon of the fo...
[REVIEW] 7 civilizational_marks carry NO status field. Read this
record's audit_summary.care_findings in full BEFORE consuming any mark — a mark
cut by the audit lives there and nowhere else.
audit care_findings (12) — binding on every author of this record
[MAJOR] weave_links W04 — target_anchor and scholarly_basis_classification — HELD from the weave lane pending a target-anchor rewrite. The source says only, at II-5: 'Elsewhere in Asia, the power of princes was often associated with control of irrigation. But in Bali, water is regarded as the gift of a Goddess...'. The record's target anchor instead names 'riverine hydraulic states (Egyptian / Mesopotamian model)' and 'state-controlled irrigation as an instrument of princely/despotic power' — which the source never says and no page anchor supports. As written the link imports Wittfogel's hydraulic-despotism thesis, which is scholarly contested and extensively critiqued as orientalist, and it casts Egyptian and Mesopotamian civilizations — which the arc renders at Ch 15-17 under the same 17.1 obligations Bali gets — as a despotic foil that makes Bali look good. That is a defamatory-trope risk pointed at a third party, and it is easy to miss because it reads as a compliment to the region under audit. Fix: (a) retarget to what the source states, the princely-control-of-irrigation pattern elsewhere in Asia; (b) hold the contrast at INSTITUTIONAL-FORM grain (cooperative water order vs state-administered irrigation), never civilizational character; (c) require a real page anchor from an Egypt/Mesopotamia source before the link queues at that end; (d) keep scholarly_basis_classification=contested so the graph carries the evidence tier honestly. The Balinese half is solid and independently attested (II-71: Raffles' 1811 surprise that the rajahs did not claim all land; the 1887 Liefrinck study quoting Montesquieu on the freedom of those who till the soil).
[MAJOR] entity_candidates EC_10 — the 'Jalu' keris — HELD from the weapon lane. The record proposes a named legendary blade the source does not attest. II-33 and II-65 give only the rock-cut inscription 'haji lumah ing jalu...rwa anak ira' — 'the King who passed away in the Jalu...and his two children' (the king identified as Udayana) — plus the gloss that the word 'Jalu' means 'Keris', which is why the river is the Sungai Pakerisan, River of the Dagger. 'Jalu' is the place where a king died; the keris connection is an etymology of a river name. Building a named artifact out of it does two things at once: it invents naming canon, which is Josh's authority under the cardinal do-not-invent rule, and it credits Balinese tradition with a pusaka it does not hold — a mis-credit dressed as enrichment. Fix: demote to unnamed pusaka-keris substrate feeding the region page's existing Section 6 legendary line (which already reserves 'a named, lineage-carried blade' for the build pass), and raise the naming to Josh if a named blade is genuinely wanted. The two genuine substrate facts survive the demotion intact: the river of the dagger as a lineage-death memorial, and Indra's kris as the instrument that creates holy water (II-64).
[MINOR] citations across S02, S03, S05, S06, S07, W04, L3_01, civilizational_marks[4] — Anchor drift of 1-2 print pages on six citations — PDF-page numbering read as print-page numbering. Every claim IS attested in the source; the anchor just points one to two pages late, so this is correctable provenance hygiene, not fabrication. Corrections, verified chunk-by-chunk against the indexed source: the 'all farmers who benefit from a particular flow of water share an obligation', the multi-subak congregations, and the 250+ congregation all sit at II-5 (cited as II-7 in S02 cite4, S03 cite3, S07 cite1, civ-mark 5 cite1; the 250+ redistributive-centre framing is also at II-20); the Bebetin A I AD 896 undagi aungan tunnel engineers sit at II-54 (cited as II-56 in S05 cite2 — the civ_marks row cites it correctly); the 'haji lumah ing jalu' inscription sits at II-65 (cited as II-67 in S06 cite3); Raffles and the Liefrinck/Montesquieu quotation sit at II-71 (cited as II-72/II-73 in W04 cite2); the Lansing and Miller 2005 footnote is footnote 5 on II-5 (cited as 'II-7 footnote' in L3_01). Fix the anchors before any row is appended, since the anchor is the provenance the registry inherits.
[MINOR] story_seeds S02 — the term 'tektek' — The seed names the proportional flow-dividers 'tektek'. Neither 'tektek' nor 'tembuku' appears anywhere in the source (verified: zero occurrences across all 114 chunks); the dossier calls them only 'proportional irrigation dividers, which make it possible to tell at a glance how much water is flowing into each canal' (II-5). Supplying an unattested vernacular term inside a living-tradition seed is precisely the authenticity failure 17.1 guards against — it reads as authoritative cultural detail while being uncited, and it would flow straight into the localization glossary the region page Section 17 mandates. Fix: drop the term, or carry it only with a source that actually attests it.
[MINOR] story_seeds S08 and S10 — game_surface_routing to besakih_mother_temple — Both seeds carry material documented at Pura Ulun Danu Batur but route to Besakih because the bali.md Zone Catalog has no Batur zone. The lava-halt is a documented account of the 1905 Batur eruption at the Batur temple (II-58, the Moojen quotation: 'The glowing lava stream was stopped just at the main entrance in an inexplicable way!'), and Dewi Danu's temple is Pura Ulun Danu Batur on the crater rim (II-19/20). Staging one living community's documented sacred history and its own goddess at a different community's temple is a 17.1 misattribution — the collective-care obligation includes not reassigning a community's sacred property to its neighbours. I set zone_or_site_hint to null on both rather than force a wrong zone. Fix: mint a Batur crater-rim water-temple zone in the Zone Catalog (mint flag raised). Secondary note on S08: the event is 1905, against the chapter's ~1200 CE present frame — the seed's own time-fluctuation/atmospheric framing is the right handling and must be kept, never asserted as present-frame history.
[MINOR] story_seeds S09 / entity_candidates EC_08 / l3_connections L3_05 — the decipherment framing — The source's own sentence is the hazard: 'In the 1920s, the Dutch archaeologist Wilhelm Stutterheim deciphered the inscription and reported his findings to the local Balinese, who had no knowledge of its contents' (II-62). Reproducing that colonial-era register in a player-facing beat would make an outsider the recoverer of a living community's own meaning, and would frame the community by what it lacked rather than by what it kept. The record's care_flag correctly handles the anti-fringe half (real archaeology, not lost-tech) but not this half. Fix: make the community's own thousand-year faithful transmission the achievement the beat honours — they performed the rite correctly for a millennium — route the recovery through Manukaya's own priests and custodians with the protagonist assisting rather than revealing, and keep Stutterheim as authorial/design-layer provenance. Related: S09 is tagged proposed_layer L3 though the player acts in it; the playable half belongs at L2, with L3_05 as its design twin.
[MINOR] entity_candidates EC_09 — One language_script candidate bundles two distinct languages owned by two distinct peoples (Old Balinese; Kawi/Old Javanese) plus a script variety (Kadiri Quadrate). culture is a GATED key for language_script under schema 11.2, and no single honest value exists for the row as constituted — the compound value I supplied is a stopgap. Fix: split into Old Balinese (Balinese; the Manukaya stone, II-62) and Kawi/Old Javanese (Javanese; II-31, with Kadiri Quadrate as its script form, II-33).
[MINOR] entity_candidates EC_02, EC_03, EC_04, EC_05, EC_06, EC_07 — target_registry routing — The five deity rows carry target_registry T0_Character_Index, which the contract's 2026-07-22 correction supersedes: deity and folkloric_being both resolve to T0_Creature_Roster, deities under species_class=deity (the discriminator carrying what the table name does not). The record predates the ruling (generated 2026-07-21) and is internally inconsistent already — EC_02 carries target_species_class=deity alongside T0_Character_Index. Additionally EC_06 is kinded folkloric_being with species_class=folklore though the source calls both beings bhatara / 'Deity of' (II-45); re-kind to deity. Mechanical, and harness/rag/exr_conform.py's RETIRED_REGISTRY_MAP + KIND_SPECIES_CLASS should apply it — but it must be applied at merge or six rows land in the wrong registry.
[MINOR] story_seeds S01 / entity_candidates EC_01 — the shape-changer attribute — The boss design leans on Mayadenawa as 'the shape-changing demon-king', but II-64 attests only 'the demonic king Mayadenawa' who 'creates a poison spring', with Indra plunging a kris or banner into the earth to make Tirtha Empul and revive his followers. The shape-changing comes from the wider Usana Bali tradition, which EC_01's substrate acknowledges but S01 does not. This is an evidence gap on a boss's core mechanic, not a care failure. Fix: carry a second cited source for the shape-changing, or build the encounter on what the dossier attests — the poisoner of the spring, fought across the purification court. Either way the rated-M no-defang posture is correct and should not soften.
[NOTE] care_summary.care_tier = elevated — Checked for care-tier INFLATION, which is itself a defect, and concluded elevated is the honest value — no change. Bali appears in neither the named high-care set (Sami, Haiti, Australia) nor the named elevated set (Congo, East African, South Africa, Mesoamerica, Amazon), which would default it to standard on a literal reading of 2.6. But the material is a living religion with actively worshipped World Heritage sacred sites, living priesthoods, and communities named in the source as present-day custodians, and the region page already binds 17.1 living-tradition handling throughout. Routing up on a non-negated authorial invocation is documented as 8.8-safe. The binding_disciplines list is specific and operational rather than ceremonial, which is the real test of whether a tier is doing work.
[NOTE] the whole record — reveal discipline — Re-verified row by row across all 10 seeds, 5 weaves, 10 entities, 5 L3 connections, 7 civilizational marks and the care summary. No House-of-Velheim name, family, or pattern; no agent-noun who-frame or invoke-and-negate construction; no Grand Sage hint; no thesis declaration; no Corr/Astrid/Cassius. Nothing in this source touches the antagonist apparatus at all — it is an engineering and religion dossier — so the reveal surface is naturally clean rather than narrowly avoided. The single L3-visibility weave (W04) is design-layer as typed, and its content is institutional history rather than arc-truth, so there is no L3-truth-on-an-L2-row leak.
[NOTE] the whole record — defanging and mis-crediting check — No defanging found, and the record is actively good on this axis. Mayadenawa is staged as a fightable boss rather than protected into scenery. The five deities are portrayed as real and usable, honoured as good deities, with the deity-name-on-a-mortal-NPC limit explicitly held on every one. No living Balinese person, priest, pedanda, balian, pekaseh, or subak farmer is routed anywhere near a combat or villain frame — the antagonism in the chapter stays with the myth-monster and the institutional-proxy certification apparatus, exactly as the region page requires. No modern Western cryptozoology construct is credited as Indigenous tradition (nothing of that class appears in this source). The vril/Nazi-occult firewall is relevant only at L3_03 and is correctly pre-empted there. The one mis-credit in the record is EC_10, handled above.
mint_flags (4) — these are RULINGS-PENDING, never invented into the page
EC_10: A named legendary keris for the Bali seat of the 72-weapon roster (the record proposes 'the Jalu keris'; the region page Section 6 reserves 'a named, lineage-carried blade' for the build pass).
S06, S08, S10: Two additions to the bali.md Zone Catalog: a Batur crater-rim water-temple zone (Pura Ulun Danu Batur + Lake Batur) and a Pakerisan-gorge zone (Pura Gunung Kawi, the rock-cut candi and meditation niches).
W01, W02, W03, W04, W05: A scholarly_basis_classification enum member for institutional grey literature — a state-party World Heritage nomination dossier, expert-compiled and ICOMOS-evaluated, citing peer-reviewed work throughout.
W04: A frame_a_type enum member for contrast/inversion links.
EXR_SRC_00482
audit verdict: RATIFY
extraction_status: deep_read_audited
[CONSUMABLE] STORY SEED EXR_SRC_00482/S01 — Track the Island's Own Face (the Bali tiger OBSERVE loop)
[CONSUMABLE] STORY SEED EXR_SRC_00482/S02 — The Tiger at the Forest Edge
[CONSUMABLE] STORY SEED EXR_SRC_00482/S03 — What the Barong Outlived (modern-strata vanished-tiger beat)
[HELD] WEAVE LINK EXR_SRC_00482/W01 — The two Sunda island tiger subspecies are a genuine sister pair with a shared fate; as the player crosses t...
[HELD] WEAVE LINK EXR_SRC_00482/W02 — Both are Lesser-Sunda / Wallacea islands whose endemic fauna is shaped and endangered by the same island-bi...
[CONSUMABLE] ENTITY EXR_SRC_00482/EC_01 — Bali tiger (Panthera tigris balica)
[CONSUMABLE] L3 CONNECTION EXR_SRC_00482/L3_01 — Island endemic megafauna are fragile by the mathematics of isolation, not by chance: small insular populati...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00482/L3_02 — The reason Bali could carry both a dense wet-rice civilization AND a wild tiger is the same volcanic geolog...
[REVIEW] 1 civilizational_marks carry NO status field. Read this
record's audit_summary.care_findings in full BEFORE consuming any mark — a mark
cut by the audit lives there and nowhere else.
audit care_findings (11) — binding on every author of this record
[MAJOR] weave_links W02 - why_it_matters, and the target_anchor motif '(Wallacea)' — FACTUAL ERROR, contradicted by the record's own source. The rationale states 'Both are Lesser-Sunda / Wallacea islands'. Bali is NOT a Wallacea island: the Wallace Line runs between Bali and Lombok, so Bali sits on the Sunda Shelf (Sundaland), west of the line, while Flores is east of it in Wallacea proper. The cited source says so directly on printed p. 7 - 'the Sunda Islands (Sumatra, Java, Bali and Borneo)' - and the sibling shelf record EXR_SRC_00496 grounds the same point genetically (Sunda-Shelf land bridges joining Bali, Java and Sumatra at the LGM). This is exactly the gate-invisible class of error the audit exists to catch: it would ride into a Cross_Cultural_Link_Graph rationale and then into bali.md Section 15 prose. DO NOT CUT THE LINK - corrected, it gets stronger and truer: Bali (Sunda shelf, west of the Wallace Line) and Flores (Wallacea, east of it) are two biogeographically DISTINCT island systems governed by the same island-biogeographic logic - isolation, small effective population, absent corridors - which is a genuinely parallel-independent shared scientific category and, as a bonus, makes the most famous line in island biogeography a real L2 rabbit hole for the Flores-to-Bali crossing. Suggested rewrite of the rationale clause: 'Bali sits on the Sunda shelf and Flores in Wallacea, on opposite sides of the Wallace Line - two separate island systems whose endemic fauna is shaped and endangered by the same island-biogeography logic, never a claim one influenced the other.'
[MAJOR] weave_links W01 - link_type 'parallel-independent' (and the same typing on EXR_SRC_00496 W01) — MIS-TYPED RELATION. The Bali and Javan tigers are not an independent parallel - they are conspecific sister subspecies of one recent lineage. The sibling shelf record EXR_SRC_00496 (Xue et al. 2015, J. Heredity 106(3):247-257) places them in a monophyletic Sunda-Islands clade differing by only 1-2 nucleotides, joined by repeated Sunda-Shelf land bridges. Calling that 'parallel-independent' asserts an independence that is factually false, which is the mirror-image failure of hyperdiffusion rather than an instance of it. The Frame-A intent is still met (no diffusion claim, no culture-taught-culture claim, and it is not a cultural link at all), so frame_a_ok stands and the row is NOT cut; I have set frame_a_type to catalog-convergence, the only honest member of the three-value Frame-A enum - the two are documented co-members of a catalogued clade (the source's own pairing of the two extirpated Sunda island subspecies), not independent inventors of the same thing. ACTION for the link-graph ratification lane: link_type should become 'continuity' or 'catalog-membership' from the live T0_Cross_Cultural_Link_Graph vocabulary. Note this is a SHELF-WIDE pattern, not a one-off - SRC_00496's W01 carries the identical mis-typing on the identical relation, so fix both together or the graph will carry two rows asserting the same false independence.
[MINOR] citations across weave_links W01 and entity_candidates EC_01, EC_02 - every 'p. 2' anchor — MIXED PAGE-ANCHOR CONVENTION. The record's _pilot_provenance states 'Page anchors use the bulletin's printed page numbers', and the p. 5 / p. 6 / p. 7 anchors do exactly that - I verified each verbatim against the page-anchored chunks. The p. 2 anchors do not: they are PDF page indices. The article opens on PDF page 2, which is PRINTED page 3, and that is where both binomials are named ('the extirpation of Bali tiger (Panthera tigris balica) and Javan tiger (Panthera tigris sondaica)'). CORRECTION: W01's 'p. 2-3' should be 'p. 3-4' (p. 3 names both binomials in the Introduction; printed p. 4 carries the depleted-or-extinct subspecies list that makes them a pair); EC_01's 'p. 2 (Panthera tigris balica)' and EC_02's 'p. 2 (Panthera tigris sondaica)' should both be 'p. 3'. The content genuinely exists in the source, so this is an anchor repair, not a cited-or-cut failure - but it must be applied before append, because the contract's citation discipline is that a row's anchor resolves to a real page-anchored chunk, and a p. 2 lookup on this source returns the bulletin cover.
[MINOR] story_seeds S01 - citation p. 5 ('rich volcanic slopes ... monsoon forest'), and the same phrasing in EC_01 descriptive_substrate — SPLICED QUOTE ACROSS TWO PAGES AND TWO ISLANDS. The ellipsis joins two things that are not one sentence and not one subject: 'The rich volcanic slopes, with their superb drainage and climate' is printed p. 5 and is about IRRIGATION, not tiger habitat; 'monsoon forest' appears on printed p. 6 and refers to EAST JAVA ('Much of the extensive monsoon forest areas where tigers had lived in east Java had been converted to teak plantations'). This source never describes Bali's tiger habitat as monsoon forest. What it DOES say about Bali tiger habitat, on p. 5, is 'the large blocks of forested land that was the home of tigers in the island' - cite that instead. The design framing is unaffected and stays legitimate: the Agung-watershed monsoon forest is bali.md Section 1's own terrain authority ('monsoon forest in the watershed belts'), so the seed keeps its setting; it just must not be sourced to this book. Same repair applies to EC_01's 'volcanic-slope monsoon forest' phrasing.
[MINOR] entity_candidates EC_02 - region_ref 'bali' on a Java-target entity — MERGE MISROUTE RISK. EC_02 is explicitly a cross-pipeline hand-off ('do not author into the Bali chapter; flag for the Java region build', owner_lane cross_pipeline), but region_ref is bali because schema 5 derives it from primary_region. The deep-read merge names its stream views deep_read_<registry>_<region>.json, so the Javan tiger will land inside the Bali creature stream where the Bali region-page build consumes it. owner_lane and the care_flag carry the signal but region_ref is the routing key. Recommend the merge honour owner_lane == cross_pipeline by holding the row out of the primary region's view (or emitting it into a hand-off queue for the sumatra_java owners), rather than the audit rewriting region_ref - the region field is derived, not audit-owned, and the honest fix belongs at the merge.
[MINOR] entity_candidates EC_01 vs EXR_SRC_00496 EC_01 - dedup anchor — CROSS-RECORD DEDUP RISK on the same real animal. Both bali-shelf records propose the same entity against the same existing roster row CR_0017, but under different names: 'Bali tiger (Panthera tigris balica)' here, 'Bali tiger (Panthera tigris balica) -- the samong' in SRC_00496. Schema 10's anchor normalization is paren-strip plus first /-alternate, which leaves 'Bali tiger' against 'Bali tiger -- the samong' - they will NOT union, so the merge is likely to emit two candidate rows enriching one CR_0017. Culture is null on both (correctly), so schema 10's own warning applies: the culture arm of the Q3 anchor is unexercised and region_ref alone is doing the discriminating. Recommend the merge normalize a trailing ' -- ' epithet as well, or that the two records be deduped by hand at the CR_0017 append - this is a two-row collision, not a content conflict; both descriptions agree on every fact.
[NOTE] care_summary.care_tier 'elevated' (vs 'standard' on the sibling record EXR_SRC_00496) — CARE-TIER CHECK - judged DEFENSIBLE, not inflation, but recorded for the 8.8 batch pass because two records on the same animal, on the same shelf, disagree. The base subject is a natural-science bulletin about a real animal, which is standard; the escalation is history-handling for the colonial extirpation regime plus the 17.1 collective-protection discipline that keeps the loss off the living Balinese, and the care_tier_note says exactly that ('standard-with-elevated-history-handling'). Per schema 9 an affirmative qualifier legitimately routes UP. The reason this matters and is not bookkeeping: elevated must not be allowed to trigger omission behaviour downstream. The colonial hunting regime, the fragmentation, and the extinction are to be rendered with WEIGHT - S03's stilled empty forest is the record doing this correctly. Care here means the loss is shown and correctly attributed, never softened away and never re-attributed to the Balinese community.
[NOTE] record-wide taxonomy - 'Panthera tigris balica' / 'Panthera tigris sondaica' as two subspecies — DATED TAXONOMY, design-safe but flagged so the two bali-shelf records do not fight. This 2006 bulletin uses the classical eight/nine-subspecies scheme. Modern genetics - including the sibling shelf source SRC_00496 (Xue et al. 2015), which finds the Sunda tigers differ by only 1-2 nucleotides in a single monophyletic clade - supports the two-subspecies revision in which all Sunda-island tigers are P. t. sondaica. Nothing in the game breaks: 'Bali tiger' as the island's own extinct apex, and even the balica binomial as the historically-used name, are entirely legitimate for a ~1200 CE frame and for CR_0017. The rule to record is only that any TAXONOMIC claim in the harvest should be governed by SRC_00496, not by this bulletin, and that L3 substrate should not present the separate-subspecies status as current science.
[NOTE] cross-check - bali.md Section 3 ('extinct in the real world by 1937') vs this record / source ('gone from the wild forever' by the end of the 1940s) — DATE DIVERGENCE between the region page and the harvest, reconcilable rather than contradictory: 1937 is the last confirmed specimen (the standard last-record date, echoed by SRC_00496's 'last recorded in the 1930s'), while this source's 'By the end of the 1940s, the Bali tiger was gone from the wild forever' is functional extirpation. Both are true of different events. Recommend the region page phrase it as both ('last confirmed 1937; gone from the wild by the end of the 1940s') so a later reader does not treat one of the two as an error and 'correct' it - the flores lane already lost time to exactly this class of silent disagreement between a page and its source.
[NOTE] story_seeds S01 - zone_or_site_hint left null; bali.md Zone Catalog — REGION-PAGE GAP, recorded rather than papered over. S01 plays in the highland monsoon forest of the Agung watershed above Besakih, and the Zone Catalog has no forest or watershed zone - its seven rows are besakih_mother_temple, village_balian_ward, tirta_empul_springs, jatiluwih_subak_terraces, wayang_performance_ground, sanur_reef and uluwatu_seacliff_temple. besakih_mother_temple is the nearest anchor (it is on the Agung flank) but it is a protected living sacred site whose zone is the meru-tower complex and lontar library; routing a tiger-tracking loop into it would misplace the loop inside a protected temple precinct, which is worse than leaving the field honest. NULL is therefore the correct value, and the page owes either a highland-forest zone row or an explicit region-ambient routing for the CR_0017 OBSERVE loop. bali.md Section 3 already stages the fauna, so this is a Zone Catalog omission, not a design gap.
[NOTE] weave_links W01/W02 - scholarly_basis_classification, and W02's target-end citation — TWO SMALL VOCABULARY/EVIDENCE ITEMS. (a) The classification enum offered for this audit - peer_reviewed | scholarly_monograph | primary_ethnography | popular_or_dated | contested - has no member for grey literature. Tigerpaper is a non-peer-reviewed FAO/RAP regional quarterly bulletin: not peer-reviewed, not a monograph, not ethnography, and not contested (its factual account is uncontroversial and matches the specialist literature). I set popular_or_dated as the honest floor - it is the accurate half of that label, since the taxonomy IS dated - but the record should not be read as calling the article unreliable. If the enum is ever revised, 'technical_bulletin' or 'grey_literature' is the missing member. Note also the contract 2.3 list (frazer, eliade, campbell, doniger, peer_reviewed, primary_anthropological_field, regional_specialist) differs from the audit enum; the two want reconciling. (b) W02's far end (Flores/Wallacea island endemism) carries no citation - unavoidable in a book-first record, since this source never mentions Flores, but it means both_ends_named_and_cited is honestly false and the Flores end must pick up its anchor from the flores-indonesia records at the link-graph append.
mint_flags (2) — these are RULINGS-PENDING, never invented into the page
S01: No attested quest_type fits a chapter-persistent, non-quest fauna OBSERVE loop with an accumulating observation threshold (the CR_0017 7-observation counter granting Barong insight).
S03: No attested quest_type fits an authorial ambient world-state beat that is region-page prose plus a modern-revisit texture pass, with no player objective at all.
EXR_SRC_00496
audit verdict: RATIFY-WITH-CUTS
extraction_status: deep_read_audited
[HELD] STORY SEED EXR_SRC_00496/S01 — The Empty Forest (forward-window tiger track)
[CONSUMABLE] STORY SEED EXR_SRC_00496/S02 — Reading the samong -- real tiger-sign tracking
[CONSUMABLE] STORY SEED EXR_SRC_00496/S03 — The tiger came over the old land
[CONSUMABLE] STORY SEED EXR_SRC_00496/S04 — Modern-revisit: the apex the world state kept or lost
[CUT] WEAVE LINK EXR_SRC_00496/W01 — The Bali, Javan, and Sumatran tigers are one recent Sunda lineage (1-2 nucleotides apart), so any Bali<->Ja...
[CONSUMABLE] ENTITY EXR_SRC_00496/EC_01 — Bali tiger (Panthera tigris balica) -- the samong
[CONSUMABLE] L3 CONNECTION EXR_SRC_00496/L3_01 — During the Ice Age, low sea levels repeatedly joined Bali, Java, and Sumatra across the exposed Sunda Shelf...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00496/L3_02 — All modern tigers share a very recent common ancestor (~72,000-108,000 years ago; ~94,500 in this study), a...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00496/L3_03 — The Bali tiger our world walks in the 1200 CE frame is a real animal that went extinct within living memory...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00496/L3_04 — The Sumatran tiger is the closest living relative of the extinct Bali and Javan tigers and has been propose...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00496/L3_05 — The last physical Bali tigers survive only as museum skulls, bones, teeth, and pelt scattered across Europe...
audit care_findings (11) — binding on every author of this record
[MAJOR] weave_links W01 — link_type / frame_a_type — The row is a fauna common-descent relation typed as a cross-CULTURAL parallel-independent link, and no legal frame_a_type honestly applies — so frame_a_type is left NULL and the weave does not advance. Two separate problems. (1) FACTUAL INVERSION: 'parallel-independent' asserts two ends that arose independently, and the paper's central finding is the exact opposite — the Bali, Javan and Sumatran tigers are ONE recent lineage differing by 1-2 nucleotides, forming a monophyletic assemblage (p.247, p.252, p.254). The tigers are not a parallel; they are a common descent. (2) CATEGORY ERROR: all three frame_a_type classes (parallel-independent, catalog-convergence, documented-transmission) are culture-relation classes for the Cross_Cultural_Link_Graph, and this row's ends are 'Bali (Sunda apex fauna)' and 'Java + Sumatra (Sunda apex fauna)' — no motif at either end. The record's own care_flag concedes it ('This is a fauna/ecology link, not a culture-motif transmission'). Appending it as parallel-independent is actively harmful downstream: bali.md Section 15 prints each link WITH its type and states that 'only the parallel-independent links are a shared category', so this row would render as a Balinese/Javanese cultural parallel the source never claims. RECOMMENDED FIX (director-level, not a Josh canon ruling): do NOT append to Cross_Cultural_Link_Graph. The content is already carried honestly and completely by L3_01 and by the region page's Section 3 fauna register plus Section 15's existing Bali->Sumatra-Java lines. If the graph is later ruled to accept biogeographic-descent edges, re-author with link_type continuity (the enum already carries common-descent classes — LINK_0029 is linguistic-cognate) and resolve target_region_id to the existing sumatra_java page; that is a link_type-vocabulary question for the contract's critic-verified-demand path, not a record defect to paper over here. Everything else about the row is sound: both ends named, both page-cited (p.247, p.254 Fig 2 caption verified verbatim), scholarly basis peer_reviewed, and it makes no hyperdiffusion or one-culture-taught-another claim.
[MAJOR] l3_connections L3_05 — the museum diaspora list — Mis-attribution of scope against the record's own cited table. The row says the last physical BALI tigers survive across 'Stuttgart, Amsterdam, Leiden, Shanghai, St Petersburg, and a private New York collection'. Table 1 (p.249, verified) shows only Stuttgart 18922 and 18923 are voucher Bali specimens; Amsterdam 1827/1829/9183 and Leiden 45097 carry the BAL haplotype but are recorded as Indonesian/unknown origin and assigned 'Javan?' by morphology; Shanghai 1356 and St Petersburg 5737 carry SON (Javan); the S. Pivar New York specimen carries SUM6/7/8 (Sumatran). Three of the six named holdings hold no Bali tiger material at all. FIX before the L3 note lands: restate as the two Stuttgart vouchers plus four BAL-haplotype specimens of uncertain Indonesian origin at Amsterdam and Leiden (noting the 2010 Amsterdam-to-Leiden transfer), and drop Shanghai / St Petersburg / Pivar from the Bali claim. Do not cut the row — the rabbit hole is real and worth keeping.
[MINOR] entity_candidates EC_01 — descriptive_substrate 'solitary' — Cited-or-cut on a field that will land in T0_Creature_Roster. 'Solitary' is true of tigers generally but is not attested anywhere in this source; the paper supports smaller-than-Javan body size (p.252), ~5-year generation time (p.255 Fig 3 caption) and ungulate-prey dependence (p.255, 'suitable habitats containing ungulate prey'). Either drop the descriptor or attach a tiger-ecology source at append. Low risk of factual harm — recorded because the roster row is canon-bearing and every other substrate claim in this record is properly anchored.
[MINOR] story_seeds S02 — citations vs. mechanic content — The seed's substance is a real-field-sign tracking loop (pugmarks, scrapes, cached ungulate kills, scent-marks), but its two anchors cover only body size (p.252) and generation time / prey dependence (p.255). This genetics paper attests none of the tracking behaviour. The seed PASSES the five gates and should advance — the underlying tiger sign is well documented and the loop is already live canon via CR_0017 — but the minigame spec must be grounded in a tiger-ecology source before it is built, or the rewards-intelligence pillar is being served by unsourced detail in the one place the game promises real science.
[MINOR] entity_candidates EC_01 vs. the live CR_0017 row — extinction date — The record says 'extinct in the real world by the 1940s' (correct per p.247) with 'last recorded in the 1930s' (correct per p.255). The live registry row carries extinction_threat 'T4 (extinct 1937 historical anchor)' and bali.md Section 3 says 'extinct in the real world by 1937'. These are reconcilable — 1937 is the last confirmed kill, the 1940s is when the subspecies is treated as gone — but the merge must carry BOTH rather than silently overwriting the existing 1937 anchor with '1940s'. Recommended roster wording: last confirmed record 1937 (1930s); subspecies considered extinct by the 1940s.
[MINOR] l3_connections L3_05 — grounding_doc — grounding_doc is ESOTERIC_GROUNDING.md for a museum-specimen provenance trail, which is documented history and belongs in SCIENCE_GROUNDING.md. (use_class 'lore_substrate' is legal per schema 6 and is NOT a defect — only the grounding-doc routing is wrong.) Re-route at append so the esoteric layer does not accumulate ordinary archival history.
[NOTE] cross-record dedup — EXR_SRC_00482 (same bali shelf) — The sibling record for SRC_00482 (Ashraf, 'The Extirpation of Bali and Javan Tiger', Tigerpaper 2006) carries EC_01 'Bali tiger (Panthera tigris balica)' and EC_02 'Javan tiger (Panthera tigris sondaica)' — the same two entities — plus an overlapping OBSERVE-loop seed and a vanished-tiger modern-strata beat. These must UNION at the merge (the anchor-name normalization handles the paren-strip) and must not create duplicate T0_Creature_Roster rows or two competing enrichments of CR_0017. The two sources are complementary, not redundant: this one is phylogeny and restoration genetics, that one is range, habitat and the historical extinction drivers.
[NOTE] story_seeds S01 — proposed_layer — S01 is typed proposed_layer L3 but describes a scene the player walks through (the forward window, the stopped pugmarks, the silent forest). That reads L1/L1_5 with an L3 resonance rather than pure substrate. Not a gate failure and not this audit's key to re-type — flagged for the 7.5 Section-13 typing, which sees the seed in chapter context. No reveal risk either way: the 'L3 truth' involved is ecological extinction, which is not a reveal-ladder item.
[NOTE] l3_connections L3_02 — the Toba/Agung design bridge — The Toba super-eruption is Sumatra and the arc's live-volcanism atmosphere is Agung on Bali. The row uses Toba to ground a volcanic-catastrophe REGISTER, which is legitimate design-layer resonance, but it must never harden into an assertion that Agung is Toba-class or that Toba is a Balinese event. Consumption note for the region page, not a record defect.
[NOTE] care_summary — tier and 17.1 posture (verified, no inflation) — care_tier 'standard' is correct and is NOT care-tier inflation in either direction. Checked deliberately because the shelf is Bali and Balinese Hinduism is a living tradition with living sacred sites: this source is Western conservation genetics about a real animal, it carries no living-culture collective-representation load, and the record honestly leaves civilizational_marks empty rather than manufacturing a Balinese achievement claim out of a genetics paper. The tiger-Barong (samong) sacred link is correctly deferred to the region page's own grain, where SRC_00157 owns it. Reveal discipline verified across all five classes: no House-of-Velheim name, family or pattern; no agent-noun who-frame; no Grand Sage; no thesis; the extinction hooks are ecological grief, explicitly fenced from the reveal ladder by the S01 care_flag. Rangda, the leyak and the calonarang material are untouched by this record, so the do-not-defang bar is not engaged here — and OBSERVE-primary on a real animal is live registry canon, not softening.
[NOTE] weave_links W01 — target_anchor.target_region_id — target_region_id is null while the canonical sumatra_java region page exists (_source/02_Tier_2_Region_Pages/sumatra_java.md). Not a blocker at this audit because the weave does not advance on the MAJOR above, but if the row is ever re-authored under a corrected link_type, the foreign end resolves cleanly and both ends should be fully typed.
mint_flags (3) — these are RULINGS-PENDING, never invented into the page
S02: A minigame / field-skill quest_type (naturalist tracking and observation loop)
S01: An ambient / discovery-encounter quest_type (an optional in-world encounter that is neither folklore combat nor a materialized argument)
S03: Possibly NO quest_type at all — a lore/naturalist beat carried as region-page prose and observation-loop delivery
Totals
CONSUMABLE: 85
HELD: 11
CUT: 9
UNKNOWN: 0
Generated by harness/site/structure_site.py — the URL path is the repo path. review root