Date: 7 October 2026. Engine owner: moira.varga; VED-003. Outcome: the explicit classical-derived full-position policy is implemented and locally validated. Its source chain and generalisation are disclosed; direct classical D60 degree prescription is not claimed.

• moira/varga.py: adds D60Method.CLASSICAL_DERIVED_LINEAR, the reviewed source/derivation chain and VargaPoint.d60_degree_attribution. Shares the existing forward/proportional source-profile coordinate calculation. Source-profile strength normalizes inputs before other divisions and D1 relationships.
• moira/_facade_vedic.py: source-profile Shodashvarga composition uses the same canonical circular input before evaluating its other divisions. Existing named/chart delegation carries the new enum without new exports.
• moira_server/models/varga.py: all seven named/batch/chart full-position selectors advertise the new policy; strict source-profile numeric preflight applies to both full source profiles. Responses type the degree-attribution category.
• moira_server/serializers/varga.py: copies the canonical degree-attribution property. All eight placement responses use this serializer.
• moira_server/services/varga.py: source-profile composition shares circular normalization; strength and vargottama flags receive consistent inputs.
• Unit and server D60 tests: new source-scope/rational checks, both-profile external-witness checks, receipt/applicability/OpenAPI assertions and tiny negative composition regressions. Existing Chara admission checks retain their behavior and acknowledge the expanded D60 selector.
• Standards, API/REST references, the work register, Home and changelog: reconcile current admission and historical claims. The new source adjudication records the wider search and source identities. The earlier negative admission decision is explicitly superseded for the derived category.

The governing degree object is proportional progress in a selected D60 subdivision. BPHS Santhanam governs half-degree partitions, destination signs and deity ordering. Saravali, Jataka Parijata and Raman's Prasna Marga notes support separately scoped arithmetic/correspondence. The extension to D60 degrees belongs to Moira. The stricter direct-classical-attribution criterion is retained in the source standard.

No Chara arithmetic, deity spelling, ephemeris/frame/ayanamsa default, other division sign doctrine, harmonic default or modern PVR source locator is changed by this reconciliation. Existing dirty Chara/D60 work is preserved. No comparator implementation becomes a runtime dependency. No oracle, snapshot or baseline tolerance is rewritten. Website work, release and publication are separate from this local execution.

All 720 half-degree subdivisions across 12 natal signs are checked at their starts, representable right/left neighbours, interiors and excluded ends, including natal-sign and zodiac wrap. The independent oracle uses exact Fraction arithmetic over the actual binary input: source sign index must agree exactly, degree must equal the rationally derived float, and mapped longitude must remain inside the selected half-open sign. Even-sign deity reversal is checked separately.

The source-published Capricorn 13 degrees 25 minutes -> Pisces sign is retained as a sign witness. Its derived 25-degree coordinate is not relabelled as a published coordinate. The test accounts exactly for conversion of that decimal input to binary64 rather than introducing a wider oracle tolerance.

Raman's notes provide two primary commentarial D9 full-position examples: Cancer 24 degrees 25 minutes -> Aquarius 9 degrees 45 minutes, and Taurus 25 degrees 11 minutes -> Leo 16 degrees 39 minutes. Exact rational arithmetic and the existing Navamsa implementation agree within 1e-12 degrees after binary input conversion. The inconsistent Gulika line is recorded as an erratum and excluded. These D9 examples are corroboration of the stated degree construction, not published D60 coordinates.

The existing immutable 4,552-case d60_cross_engine_2026-10-07.json is now exercised against both named full source profiles. Its recorded profile identity remains the original modern profile; this does not rewrite the evidence's provenance. The new policy shares its declared numeric mapping, and must meet those same independent expected values and thresholds.

Fresh evaluation of classical_derived_linear also checked all 24,432 inputs against the previously recorded independent-function outputs:

Rational disagreements: 0. Secondary-comparator sign disagreements: 0. Degree disagreements exceeding the unchanged predeclared budget: 0. Absolute degree tolerance remains 8*ulp(2160.0) = 3.637978807091713e-12 degrees, with no sign tolerance. Maximum degree residual is 1.1368683772161603e-13 against PyJHora full positions and 2.2737367544323206e-13 against Maitreya's instrumented intermediate. Maitreya's returned product is a sign; its intermediate is not misrepresented as a published full-position API.

Pinned comparator identities, extraction/build scope and the original corpus generation are owned by the broader validation receipt and scripts/d60_cross_engine_sources.json. The new run verifies those source checksums, corpus identity/order/count, exact rational invariants and the new policy, recording hashes of consumed input/output files. It consumes the prior comparator outputs, rather than claiming a new full-application comparator run. Private receipts are C:\dev\outputs\vedic-d60-library-audit-2026-10-07\validate_classical_derived.py and classical_derived_corpus_validation.json.

These are conditioned D1-to-D60 checks: the planetary inputs originate in Moira DE441 and are not independent ephemeris truth. Software agreement corroborates the admitted coordinate mapping, not historical authorship or predictive validity.

Every Python execution uses C:\dev\moira\.venv\Scripts\python.exe (Python 3.14.3). Imports resolve to this checkout and its CPython 3.14 native extension. Varga mapping is Python arithmetic; the native extension is not claimed to supply a separate D60 implementation.

Tests set MOIRA_TEST_MODE=1, MOIRA_NO_DOWNLOAD=1, and MOIRA_STRICT_KNOWN_ISSUES=1, with external-network tests excluded. The known issue register is empty. The project resource resolver selects C:\Users\nilad\.moira\kernels\de441.bsp; no download is enabled.

.\.venv\Scripts\python.exe -m pytest `
  tests\unit\test_d60_classical_derived.py `
  tests\unit\test_d60_full_position.py `
  tests\unit\test_d60_cross_engine.py `
  tests\unit\test_d60_published_evidence.py `
  tests\unit\test_vedic_chara_d60_admission.py `
  tests\server\test_d60_full_position.py `
  tests\server\test_d60_cross_engine.py `
  tests\server\test_vedic_chara_d60_admission.py `
  -q -m "not external_network" `
  --junitxml=C:\dev\outputs\vedic-d60-library-audit-2026-10-07\classical_derived_tests.xml

.\.venv\Scripts\python.exe C:\dev\outputs\vedic-d60-library-audit-2026-10-07\validate_classical_derived.py

The main focused run passed 1,528 tests, with zero failures, errors or skips. Its resource receipt records 44 real-reader tests, all run with DE441, one content probe and zero resource failures/skips. This includes both full profiles across all seven REST shapes, sixteen external-witness date/frame chart batches per profile, two strength groups, input rejection before resource work, and preservation of modern/harmonic receipts. Two additional REST strength left-wrap regression cases were subsequently added. The supplementary run passed 132 tests (110 full-position REST tests and 22 public API adversarial checks), with zero failures/errors/skips and nine successful DE441 resource receipts. Across both final runs, 1,552 distinct test cases pass; overlapping cases are counted once.

.\.venv\Scripts\python.exe -m pytest `
  tests\server\test_d60_full_position.py `
  tests\unit\test_api_surface_adversarial_audit.py `
  -q -m "not external_network" `
  --junitxml=C:\dev\outputs\vedic-d60-library-audit-2026-10-07\classical_derived_supplementary.xml

.\.venv\Scripts\python.exe -m ruff check `
  moira\varga.py moira\_facade_vedic.py `
  moira_server\models\varga.py moira_server\services\varga.py `
  moira_server\serializers\varga.py `
  tests\unit\test_d60_classical_derived.py tests\unit\test_d60_cross_engine.py `
  tests\server\test_d60_full_position.py tests\server\test_d60_cross_engine.py `
  tests\server\test_vedic_chara_d60_admission.py

.\.venv\Scripts\python.exe scripts\check_doc_consistency.py
.\.venv\Scripts\python.exe scripts\sync_rest_api_reference.py --check
.\.venv\Scripts\python.exe C:\dev\outputs\vedic-d60-library-audit-2026-10-07\verify_policy_docs.py
git diff --check

All these checks pass. The API example executes and returns Cancer 7.5 degrees with the classical-derived receipt. All nine touched canonical wiki pages' local links and fences pass, and 357 canonical pages render through the actual wiki generator in memory, including the new untracked pages. No mirror file is written or published. The tracked-only publishing script would omit untracked pages before staging. Private receipts retain XML test results and the documentation-check JSON. Live SHA-256 checks confirm all five book/facsimile identities in the source adjudication.

Publication review additionally runs Ruff across the full pending Chara/D60 package. The 20 changed implementation/test/validator files outside the two curation modules pass. moira/__init__.py has 60 diagnostics and moira/facade.py has five; a stdin-based comparison against git show HEAD:<path> confirms exactly the same rule/message multiset in the baseline and working copy, with zero newly introduced diagnostics. Line-number references are normalised for comparison. Existing curation lint debt is not represented as a successful whole-package lint run or repaired by this publication. The doc and REST inventory checks, the executable API example and all nine touched pages' local-link checks also pass after the closure update.

The first new tests exposed a pre-existing raw-input composition fault: -1e-300 % 360 rounds to 360 in another division and can produce an invalid sign index. Both opted-in full source profiles now normalize the shared input before composing the other divisions and strength relationships. The harmonic compatibility path is unchanged; this is not a general rewrite of legacy Varga normalization. Direct D60, composed facade/REST results and strength are covered for the repaired source-profile boundary.

The exact-input example test initially used an undersized decimal/binary comparison budget, and a facade test omitted its reader slot. Both test assumptions were corrected. No source coordinate, frozen fixture or numerical baseline was changed to obtain a pass.

The implementation uses the declared subdivision/sign/progress objects and the existing source-profile arithmetic; it contains no comparator branch, external slot ordering, copied helper, or output-driven adjustment. The representable left-limit guard follows the selected half-open domain. Provenance properties distinguish equal coordinates from equal authority. Rational/source-scope checks provide the derivational evidence; secondary software is corroboration rather than the historical admission authority.

The finite derived-profile implementation and validation are complete. Direct classical D60 degree prescription, exact primary-publication planetary pairs and predictive validity remain separately unclaimed. Their absence does not leave the admitted product's engineering validation unfinished.