Nisamina / LILA · Evidence command

Live Encyclopedia Build Tracker

Current construction, linguistic authority, enrichment, and release gates—refreshed from the working evidence ledger every 15 seconds.

LIVE

Evidence updated

Last checked

Active engineering operation

V91 public release candidate built and verified

The adíaha/adíahani source boundary is corrected in the canonical atlas; the 1,471-record fail-closed public projection is exported into every runtime atlas and awaits Hostinger upload and live verification.

verified
Atlas headV91620,692 records · f8659cc65135
Evidence registryV28022,020 citation forms · 199 certified roots
Formal launchWITHHELDRelease follows evidence, rights, and community authority.

Verified source subset

Regular-root census

Partial confirmed census; joint conservative bound accounts for all 76 validated families and five possible identity overlaps. Operative additional allowance: 0.

Confirmed source-regular roots60
Roots with validated derived forms35
Maximum under the 80% ceiling4813 additional families allowed within this subset

0 additional distinct regular-root families can fit under the global 80% ceiling conservatively. This allowance accounts for all currently validated families and possible root overlaps. Each new family still needs its applicable source-backed operation; the allowance is recomputed as families are completed.

The total regular-root population is still being resolved. These subset counts are not overall project coverage.

Evidence lane

Recovery

Dictionary families represented

100.00%

12 / 12

Includes all three Hadel volumes and Stöckle.

Parsed rows admitted as forward evidence

88.84%

30,678 / 34,532

Registry V97/V98: 30,678 forward rows (+19 Hadel II V5 dual-OCR rows).

Rejected Hadel candidates with rendered page evidence

100.00%

103 / 103

All 103 rows have a PDF hash, exact physical page, rendered-page hash, OCR evidence, and fail-closed review fields.

Hadel page rows independently transcribed and reviewed

18.45%

19 / 103

Director plan (b): dual-OCR disagreement filter (tesseract + Apple Vision, strict exact) released 19 of 103 held rows into Hadel II V5; 84 stay held; human transcription remains the route for them.

People's V11 parser holds assigned evidence roles

100.00%

118 / 118

The V11 hold ledger is fully classified; no held row has automatic lexical or public authority.

People's false headword fragments attached to source entries

100.00%

7 / 7

All seven POS/punctuation fragments attach to exact source parents and are retired from the V11 hold ledger.

People's actionable V10 holds resolved from exact PDF evidence

75.00%

6 / 8

Six exact printed rows recovered. dawadawa and igarigu remain held because no English gloss is printed.

GLD rows with exact structured field separation

72.19%

1,194 / 1,654

V16 exact declared-heading member plus full lexical-paragraph match: seven paired-morphology bilingual splits; 460 holds remain. Independent raw XML audit PASS.

Unresolved GLD V15 rows assigned to repair lanes

100.00%

460 / 460

V16 holds clustered on Atlas V22: priority 1 = 0, priority 2 = 229, priority 3 = 231.

Original priority-1 GLD rows cleared

100.00%

95 / 95

All six priority-one source/sense conflicts resolved by Director rulings P1.1–P1.6 (ledger sha edd053d9…).

Citation-form registry target

89.01%

22,020 / 24,740

V278 exact citation keys.

Original identity residue resolved

98.16%

9,426 / 9,603

177 exact identity cases remain in the current residue.

Evidence lane

Authority

Declared roots with exact dictionary identity

75.57%

4,529 / 5,993

V100 / V22 manifest census; exact identity alone does not certify a root.

Certified lexical roots against historical inventory

3.32%

199 / 5,993

V278 preserves 199 source-backed root/sense certificates; 60 confirmed source-regular roots.

Rule labels with resolved locator

12.28%

28 / 228

15 thesis + 9 section-ruling + 4 skill Director-ruling locators; 200 pending canonicalization.

Historical full-rule-contract census

0.00%

0 / 231

Historical registry baseline, not current operation-specific evidence. Validated source-bound operations are counted separately.

Current stress-only root families classified

100.00%

6 / 6

All six stress-only identity rulings APPLIED in atlas V20 (2 merges, 5 renames, 74 root re-points); both surfaces retained via the alias ledger.

Compatible stress routes with Director-ready evidence

95.74%

45 / 47

45 of 47 Director docket items ruled and applied; open: R.1 (root aba family) and H.1 (4 Hadel stress-only engine-agreement rows).

Live parser and governance contract wired into skill

100.00%

4 / 4

Entrypoint and references enforce the V86/GLD V14 heads, the dual-law-source candidate lineage (V87/V88), and the executable registry lineage preflight gate.

Historical collaboration handoff

100.00%

1 / 1

Archived collaboration milestone; not a current agent or account-usage status.

Evidence lane

Teaching

Entries containing English and Spanish

98.41%

610,817 / 620,692

V90 effective V12 bilingual-or-explicitly-held gate passed; exact report pinned in current artifact evidence.

Derived entries with morphology structure

62.49%

369,843 / 591,856

Presence does not prove operation, order, seam, or sense.

Contract-complete teaching entries

0.24%

1,497 / 620,692

V91/V280 full scan: 1,497 contract-complete entries; 619,195 unsupported records remain fail-closed and nonpublic.

Candidate root families with a validated operation

2.59%

76 / 2,935

Current V21 queue: 1423 validated derived entries across 76 families; 35 of 60 confirmed regular roots covered; conservative additional capacity 0.

Root-copy families with permitted route

100.00%

45 / 45

The two no-route false-root families were retired losslessly; all remaining families have an exact or governed stress-variant route.

Evidence lane

Release

Enriched public candidates after scoped holds

0.24%

1,471 / 620,692

V90 effective V12 holds; 336 scoped example omissions; restricted scan PASS. Candidates are not deployed.

Byte-identical shard round trips

100.00%

2,560 / 2,560

V43: all 2,560 roundtrips PASS; three existing forms repaired, 620,682 entries unchanged.

Blocked-rule records explicitly held

100.00%

548,573 / 548,573

RULE_PUBLICATION_COVERAGE_AUDIT_V100_SCOPE_V2 PASS: all 548,573 blocked-rule records explicitly held on unchanged Atlas V22.

Executable controls

Gate board

  • pass
    Bilingual or explicitly held

    V90 effective V12 hold gate PASS; exact current report retained in publication verification.

  • pass
    Named attribution or explicitly held

    V90 effective V12 hold gate PASS; exact current report retained in publication verification.

  • pass
    Registry lineage and enrichment preservation

    V278: 199 certificates; current queue 1423 derived entries across 76 families. Full gate 1497 verified.

  • fail
    Entry-specific academic enrichment

    V90 full gate: 1497 verified; 619195 pending.

  • held
    Regular-root coverage

    60 confirmed regular roots, 35 exactly covered; conservative additional capacity 0. Global regular-root denominator remains unresolved.

  • pass
    Public candidate content scan

    V90: 1471 candidates, 336 scoped example omissions, zero restricted findings.

  • pass
    Application production build

    Next.js 15.5.18 production build passed; progress page verified in the local production server. Website deployment remains outstanding.

  • held
    Encyclopedia release

    The complete root-coverage target and remaining enrichment are unfinished; the updated encyclopedia has not been deployed.

Verified movement

Recent activity

  1. V90/V278 evidence and progress reconciled

    1497 verified entries; 1471 public candidates; current queue 1423/76; confirmed regular census 60, covered 35, capacity 0.

  2. V89/V276 evidence and progress reconciled

    1490 verified entries; 1464 public candidates; current queue 1416/75; confirmed regular census 60, covered 34, capacity 1.

  3. V88/V274 evidence and progress reconciled

    1140 verified entries; 1114 public candidates; queue 1066/74; confirmed regular census 60, covered 33, capacity 2.

  4. V87/V272 evidence and progress reconciled

    882 verified entries; 856 public candidates; queue 808/67;confirmed regular census 60, covered 26, capacity 9.

  5. V86/V268 evidence and progress reconciled

    812 verified entries; 786 public candidates; queue 738/57; confirmed regular census 57 with 16 covered and conservative capacity 19.

  6. Six-family conservative allowance completed

    Fifty-six forms across six source-regular roots passed all gates. Current totals: 728 enriched entries, 654 validated derived forms across 57 families, and 702 public candidates.

  7. Second additional regular-root family completed

    Fourteen élecha fruit-object UMU forms passed all gates. Verified total 700; public candidates 674; four conservative family slots remain.

  8. First additional regular-root family completed

    Fourteen áhulucha UMU forms passed exact source, morphology, bilingual, attribution, round-trip, hold and public-content gates. Verified total 686; public candidates 660.

  9. Current milestone integrated: 672 enriched entries and 34-root census

    173 certificates, 646 publication candidates. Six additional regular-root families fit within the independently checked conservative 80% bound. First 14-form family integration underway.

  10. Seven root certificates integrated; stale gate board corrected

    159 certificates and 20 source-regular selected senses. All 665 verified entries retained. Production build passed; old two-entry gate-board result replaced with current evidence.

  11. 49 lexical entries integrated; first regular-root subset measured

    665 verified entries and 639 publication candidates. Suazo subset: 13 confirmed source-regular roots, 10 covered, ceiling 10; global census continues.

  12. Seven negative-series entries and satisfied holds integrated

    V78/V245:56 enriched;15public candidates pass restricted-content scan.38source-validated derived entries span17roots.

Percentages are independent evidence gates. They are never averaged into a false overall completion score.