[HELD] ENTITY EXR_SRC_00158/EC_04 — Devata (aggressive spirits of the pustaha diagrams)
[CONSUMABLE] ENTITY EXR_SRC_00158/EC_05 — The datu's ritual arsenal — magic staff (tunggal panaluan) and medicine horn (naga morsarang)
[CONSUMABLE] ENTITY EXR_SRC_00158/EC_06 — The datu (ritual-specialist persona substrate)
[CONSUMABLE] ENTITY EXR_SRC_00158/EC_07 — Sisingamangaraja XII (Batak priest-king, hero of a pustaha-referenced legend)
[CONSUMABLE] L3 CONNECTION EXR_SRC_00158/L3_01 — The pustaha is made from the inner bark of Aquilaria malaccensis — the very tree that yields agarwood (gaha...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00158/L3_02 — The porhalaan reckons a schematic 360-day year (twelve months of thirty days) from the conjunction of the S...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00158/L3_03 — Surat Batak belongs to the Brahmi → Pallava family that seeded the whole archipelago's scripts (Javanese, B...
[CONSUMABLE] L3 CONNECTION EXR_SRC_00158/L3_04 — The colonial record cast the Batak as illiterate, cannibal 'primitives'; the manuscript evidence shows a br...
[REVIEW] 4 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] record-level - care_summary, story_seeds, and l3_connections (the harvest as a whole) — UPHELD BY THE FINALIZER ON INDEPENDENT EVIDENCE. THE SOURCE'S OWN CENTRAL FRAME IS MISSING FROM THE HARVEST: colonial collection provenance. This article is not primarily about pustaha in Batak hands - it is about seven pustaha located in four POLISH institutions. I read the opening pages myself: print p.164 frames the manuscripts as 'transgressive objects' functioning in parallel in Western and Eastern societies, print p.165 states that from colonial times the pustaha 'often been brought to Europe as souvenirs and kept in national or private collections' and records roughly 2,000 in worldwide collections, and Herman Neubronner van der Tuuk - the colonial-era linguist whose collecting and analysis produced the sub-Toba classification the record cites - is named at print p.169. Every plate in the record's own citation trail is an accession number in Warsaw, Poznan, Amsterdam, Hamburg or Indianapolis. Across 8 seeds, 4 weaves, 7 entities and 4 L3s, not one row touches the fact that the datu's sacral books are in European museums and the Batak are not the ones holding them, and the D4 colonialism-and-displacement lens is used on no row anywhere. I RE-RAN THE DOWNSTREAM NEGATIVE CHECK MYSELF rather than adopting it: _source/02_Tier_2_Region_Pages/sumatra_java.md returns 0 hits for 'van der Tuuk', and its only 'Collection' hits are the Fred and Rita Richman Collection book title - a source name, not collection-and-removal history. So the gap is real and not covered downstream. This is precisely the true history the doctrine says is never hidden or scrubbed, and omitting it is the more costly error, not the safer one. It is also live substrate left on the table: a pustaha-recovery questline whose ground truth is that most pustaha are elsewhere. Devil's-advocate pass run against over-conservatism points the SAME direction - the timid instinct is to leave collection history out as awkward, and that instinct is the defect. REMEDY: add the collection-and-removal history as an L3 rabbit-hole (cited p.164-165, van der Tuuk at p.169, the museum accession numbers throughout), show Batak agency in it rather than only loss, and route any antagonism through the collecting institution or apparatus, never through a community. Additive - no row is held by this finding, no new node, no chapter-count change.
[MAJOR] story_seeds S07 - the closing clause of description — UPHELD, ROW HELD. Read directly out of the record: S07's description ends '... and it is a test the Batak community itself applies.' That is an unsupported ethnographic attribution to a living people, in player-facing text. Print p.169 presents the three-arts division as Rozwadowska's own third-person classification of pustaha CONTENT; print p.166 carries only the datu's role and the Zollo positive-equilibrium quote. Neither cited page attributes a corruption-test to the Batak community. The original audit identified this and still passed the row five_gate PASS, which under exr_audit_apply.py writes seed_status='queued' and lands the clause - a firewall stored in a note. The claim does load-bearing moral work, making a real tradition's own register the community's villainy marker, so it is not cosmetic. Devil's-advocate pass against over-conservatism: holding is not a scrub - the three-arts taxonomy itself ships whole and remains the chapter's moral instrument, exactly as S04 already does it correctly. Remedy is a one-clause reframe, not a cut: the taxonomy is legible in the recovered pages the player reads, and the community's judgement of Siahaan stays canon-side. Re-audit to advance.
[MAJOR] weave_links W02 - target_anchor.culture and target_anchor.motif — UPHELD, ROW HELD. The only surviving weave would advance with target_anchor culture 'Tamil/South India' and motif 'the Pallava (Grantha-lineage) script'. The source's sole statement is 'Batak handwriting comes from Old Sumatran, derived from the Indian Palava system' (p.168, fn 19). My own full-text search across all 16 pages including the bibliography returns 0 hits for 'Tamil', 0 for 'Grantha' and 0 for 'Pallava' (the article spells it 'Palava'). The Pallava script is not the Tamil script, so the row attaches a real living people to a script they did not write - the misattribution class - and advancing it sends that into T0_Cross_Cultural_Link_Graph where it reads as canon and is FK-checked as such. The original audit named the correction and advanced anyway because the field is not audit-owned; not-audit-owned is the reason to HOLD. This is accuracy, not conservatism: the fix restores a true attribution rather than removing content. Correct to 'Pallava (South India)', drop the Grantha-lineage claim, re-audit. Everything else on the row - T17 verified by name, documented-transmission, the local-authorship credit that makes Frame A hold, scholarly_monograph - is correct and carries forward.
[MAJOR] entity_candidates EC_04 (Devata / Naga Padoha) - target_species_class 'folklore' — UPHELD, ROW HELD. DEEP_BOOK_PROCESSING_CONTRACT §2.6c is explicit that any row at species_class = folklore MUST carry a non-empty cultural_care_register, that this is a required-non-empty append condition rather than a note, and that until the gate check exists the §8.8 audit enforces it per-row. I inspected EC_04's keys directly in the record: target_species_class is 'folklore' and there is no cultural_care_register key at all - the same shape as the failure §2.6c was ruled to fix, where 9 of 12 folklore-class rows including Ebu Gogo and Mokele-mbembe carried the field empty and a firewall stored in an empty field protected nothing. The original audit wrote the correct enforcement text and then passed the row, which advances it to queued with the field still absent; recording enforcement in a payload the apply script cannot write is not enforcing it, so the row HOLDS. ENFORCEMENT TEXT for the landing row: 'Living Batak (Toba) tradition, attested in a museum-held pustaha spirit-diagram (Asia and Pacific Museum MAP2553) and in Voorhoeve 1951; NOT a modern Western cryptozoology construct. The devata are the source's own aggressive spirits and Naga Padoha the tradition's own great serpent - render at full malevolent weight, boss-eligible, attributed to Batak tradition and never to cryptozoological literature.' Devil's-advocate pass against over-conservatism: the hold DEFANGS NOTHING - 'aggressive spirits' is preserved verbatim and Naga Padoha stays boss-eligible; the hold exists to make the authenticity attribution real rather than notional. A second defect rides the same row: descriptive_substrate hardens p.170's 'the wizard (probably the datu himself)' into 'the wizard-figure (the datu himself)' - restore the hedge.
[MINOR] weave_links W03 - the p.166 citation, and the audit's record-level 'zero anchor drift' claim — UPHELD. W03 cites ['p.170','p.166'] and the original audit stated that p.166 carries the Batak genealogy / Indic religious-worship remark. My own per-page regex sweep places 'genealogy', 'Hinduism' and 'lexical' on print p.165 (PDF p2) and on NO other page. Print p.166 (PDF p3) carries the tribe-variety line, the isolation passage, the missionary 'primitives' and cannibalism-rumour sentences, the datu description, the Zollo quote and hadatuhon, and nothing about genealogy, Hinduism or lexical remains. So one of the record's 23 citations does not resolve, and the audit's 'all 23 citations resolve with zero anchor drift' is WITHDRAWN - the auditor asserted a resolution it did not perform, which is the bali failure class in miniature. W03's verdict is unchanged and in fact strengthened: its only Indic anchor is both an abstraction and mis-paged. All 22 other citations were placement-checked against the extracted source and do resolve.
[MINOR] entity_candidates EC_07 (Sisingamangaraja XII) - portrayal_note and care_flag — UPHELD AND RE-DERIVED. CARE-TIER CALIBRATION, NOT A VIOLATION. The hard constraints are correctly held: no villain casting, no deity-name on a mortal NPC, forward-stratum only. But the instruction 'keep to legend-reference' together with the soft gloss 'a revered Indonesian national resistance figure' risks flattening documented history into vague legend. Sisingamangaraja XII led sustained resistance to Dutch colonial expansion in the Batak lands and was killed by Dutch colonial forces in 1907; he is an Indonesian National Hero for that, not for legend. Under the bolder calibration, a forward time-fluctuation glimpse should carry that resistance honestly and specifically, with Batak agency shown rather than only loss - exactly what the war-and-resistance lens provides for. Devil's-advocate pass run against over-conservatism STRENGTHENS this finding: softening a real anti-colonial death into 'hero of legend' is the scrub the doctrine warns about, and the source itself calls him 'the Sumatran hero' without hedging. REMEDY is a reframe, never a removal - keep the reference register and the no-villain rule, and let the glimpse state what he actually did and what was done to him.
[MINOR] care_summary binding_disciplines and flags — UPHELD AND RE-DERIVED. AN OVER-ASSERTION IN THE PROTECTIVE DIRECTION. I read the record's care_summary directly: binding_disciplines says the cannibalism rumour 'is never depicted as real' and flags says it is 'present nowhere as fact'. The source says only that the missionary and traveler reports were 'greatly exaggerated' (p.166, citing Andaya 2002), which I verified in place. 'Greatly exaggerated' is not 'never happened', and the scholarship - Andaya and Reid, both in this source's own bibliography - treats Batak ritual consumption as a contested, narrowly-scoped juridical practice rather than a pure fabrication. Authoring a flat negative the source does not support is itself a departure from the record; it just points the other way, and it teaches the player something the evidence does not carry. The build does not need to adjudicate the question at all: the corrective instrument is showing the literacy, the book-craft and the calendrical science, which L3_04 and S08 already do well. REMEDY: soften the binding discipline from 'never depicted as real' to 'the sensational traveler accounts are shown as the exaggeration the record calls them; the build makes no counter-assertion the sources do not support', and keep the AUTHORIAL-ONLY reframe note exactly as written. Scope correction to the original audit: L3_04's own text is accurate and this finding does not touch it.
[MINOR] entity_candidates EC_05 - name field and citation trail — UPHELD, NOT A HOLD. Neither 'tunggal panaluan' nor 'naga morsarang' appears anywhere in this article - the finalizer's own search returns 0 hits for 'tunggal', 'panaluan' and 'morsarang' across all 16 pages including the bibliography; the plate caption at bracketed print [176] attests only 'Datu with pustaha, medicine horn and magic staff'. Devil's-advocate pass against over-conservatism says do not cut or hold: the names are real and abundantly attested in the wider Batak literature, the row is Thread-20 sacred-object substrate rather than combat loot, and the row's own descriptive_substrate DISCLOSES the gap in a typed field - the opposite of EC_04, where the required field is absent. Advance, but the landing must not let SRC_00158 stand as the citation for those names in T0_Weapon_Registry, and a second source must be attached first. Separately: entity_kind 'weapon' is a poor fit for a medicine horn and a ritual staff the row itself says are not combat acquisitions; record that in target_registry_note.
[MINOR] EC_02, S08, L3_02, S05, EC_01 - hedge and precision drops — UPHELD ON ALL FIVE COUNTS, NOT HOLDS. A recurring habit of hardening the source's hedges, the same habit that produced the EC_04 defect. Print p.169 says the string 'was necessary to carry and store (probably open) the manuscript, hung on the house wall'; EC_02 renders it 'hung open on the house wall' and S08 'pustaha hang open on house walls'. fn 26 at p.170 says '12 or 13 horizontal lines'; L3_02 glosses 'twelve months of thirty days', dropping the very alternation the rabbit-hole is about. fn 21 at p.168 says 'from nineteen to twenty horizontal basic signs ... varies by dialect'; S05 and EC_01 flatten to nineteen (defensible against the script's own self-naming surat na sampulu sia, but the range is what the source supports). Each phrase was placement-verified by the finalizer. Fix at landing; none is a gate failure.
[NOTE] care_summary care_tier ('standard') — CHECKED FOR INFLATION BY THE AUDITOR, THE CRITIC AND THE FINALIZER, AND STANDS. Devil's-advocate pass run explicitly against over-conservatism: the Batak are a living Indigenous people so §17.1 collective protection binds fully, but they are not on the high-care Sami/Haiti/Australia list nor in the elevated set, and nothing in the record's content - datu craft, script, calendar, divination, ward-making - carries a POSITIVE recorded restriction that would justify elevation; silence in one source is neither evidence of closure nor grounds to raise a tier. The one genuinely elevated element is the single real revered figure at EC_07, already handled at row level rather than by dragging the whole record up a tier, which is the right shape. Consequence to hold at build time: living-practitioner substrate here ships at FULL specificity - real toponyms (Lake Toba, Samosir, the Barisan mountains, Sumatera Utara), the seven tribe names, and the real technique, species and script names (Aquilaria malaccensis, Manggistan, baja, dammar, pagar, porhalaan, pormesa, hadatuhon, surat na sampulu sia) all carry. Related check: Naga Padoha is NOT defanged - the region page rows it as boss BE_0005_S3, bound and re-settled rather than slain, which follows the tradition's own cosmology instead of softening a monster.
[NOTE] docs/proposals/SIGNIFICANCE_LENSES.md - the C-group lens codes — LENS-DOC DEFECT CONFIRMED AND THE ORIGINAL FINDING CORRECTED AS OVERSTATED; the finalizer read both passages. §2C (lines 51-55) lists five C-lenses in the order religion, mythology, philosophy, esoteric/occult, sacred landscape, which by position gives C3 = philosophy. §3 states explicitly that philosophy is C4 (line 104, 'Philosophical - lens C4') and that occult is C5 (line 102, 'Occult - lens C5'), corroborated by §5.3's 'C5 divination'. Those cannot both hold with the listing order - but the ORDER is what is wrong, not the code space: taking §3 as authoritative (it names codes explicitly where the list only implies them) resolves the group completely as C1 religion, C2 mythology, C3 sacred landscape, C4 philosophy, C5 esoteric/occult. So C3 is recoverable rather than unusable; the original audit's 'unresolvable' is too strong, though its conservative workaround cost this record nothing since no row here is a sacred-landscape row. Every C5 used in this record is a divination/occult assignment, consistent with §3, and every C1 is lived religion and every C2 mythology - so all C-codes here are semantically as well as syntactically right. The doc still needs one explicit code table, and until it has one every C-code should be checked against §3 rather than listing order. All 25 lens assignments were re-verified against the apply script's roster (A1-A4, B1-B6, C1-C5, D1-D4, E1-E2, F1-F4, G1-G8) and every one is legal; no lens correction was needed, so corrected is empty.
[NOTE] record metadata (source_year) and the RAG chunking of this source — BOTH ITEMS CONFIRMED BY THE FINALIZER. (1) source_year is null in the record and is recoverable from the source's own first page: the DOI printed on print p.164 is https://doi.org/10.15804/aoto202209, i.e. Art of the Orient, 2022 issue - a free metadata fill. It is NOT written by this payload because source_year is not an audit-owned key under exr_audit_apply.py; it needs a separate metadata pass. (2) The RAG chunking splices footnote text into body text at page boundaries, most consequentially at print p.166/167, where footnote 6's continuation runs straight into the body sentence and yields 'The profession was hereditary / in Angola and Toba, and three in Mandiling' - that tail belongs to Marsden 1811's tribe list, not to the datu profession, and the true sentence resumes at the top of p.167 with 'and dominated by males.' I verified this splice in the PDF's own text order. This record did not fall into the trap, but any regeneration or downstream extraction over these chunks can, and it would produce a confidently wrong claim about a real people's institutions. (3) This source's chunk metadata stores page_start/page_end in PDF-index space (1-16), not print space, so anything citing from chunk metadata rather than running heads produces citations off by 163 - the bali failure class one step upstream. Worth a chunker note; substrate defects, not defects of this audit's target.
mint_flags (4) — these are RULINGS-PENDING, never invented into the page
S01: seed_kind trade_expression has no ratified quest_type - proposed_quest_type left NULL, category-mint ruling required
S02: seed_kind minigame_class has no ratified quest_type - proposed_quest_type left NULL, category-mint ruling required
S06: seed_kind trade_expression has no ratified quest_type - proposed_quest_type left NULL, category-mint ruling required
S08: seed_kind festival_or_daily_immersion has no ratified quest_type - proposed_quest_type left NULL, category-mint ruling required
Totals
CONSUMABLE: 17
HELD: 3
CUT: 3
UNKNOWN: 0
Generated by harness/site/structure_site.py — the URL path is the repo path. review root