Date: 6 October 2026. Owner: Python engine in C:/dev/moira. Scope: the bounded seven-classical-planet Phaladeepika snapshot core, with a typed policy surface and a separately admitted research catalogue. No website, REST, native port, node implementation or dated forecast is implied.

Governing sequence: Moira Subsystem Constitutional Process. Current standard: Gochara backend standard.

The earlier file called “Phase 1” was an initial core delivery checkpoint, not a claim of constitutional Phase 1 closure. Its early eight-name exports were provisional local work. This ledger maps the real dependency sequence; package curation follows the invariant and architecture freeze.

This status applies to the frozen snapshot domain. Source-attested concepts in the catalogue are not automatically complete runtime domains. Expansion requires its own source admission and a revision of the governing standard.

All runs used Python 3.14.3 from the repository .venv, MOIRA_TEST_MODE=1, strict known-issue expiry and external-network exclusion. tests/KNOWN_ISSUES.yml is empty. Native/import smoke reports engine 6.9.9 and moira/_moira_native.cp314-win_amd64.pyd; numerical Gochar doctrine remains Python-governed.

The pre-edit core baseline passed 173 tests. Policy/classification/inspection gates passed 188, relation hardening 189, integration/aggregate/network 204, and full-subsystem hardening 218. The final 45-test constitutional slice passed separately before writing the architecture freeze. Ruff and scoped module/class/machine-contract docstring checks passed for both engine modules.

Final curation and surrounding command:

$env:MOIRA_TEST_MODE='1'
$env:MOIRA_STRICT_KNOWN_ISSUES='1'
.\.venv\Scripts\python.exe -m pytest tests/unit/test_gochara.py tests/unit/test_gochara_constitutional.py tests/unit/test_api_surface_adversarial_audit.py tests/unit/test_ashtakavarga.py tests/unit/test_muhurta.py -m 'not external_network' -o addopts='--import-mode=importlib' -q

Outcome: 520 passed in 28.90 seconds, no failures or skips. This selected run includes 239 Gochar cases after public-surface expansion, the complete API-surface audit, surrounding Ashtakavarga and Muhurta tests. It is not a full repository run. The initial core's earlier broader run exposed one pre-existing Shadbala reader-context failure, reproduced with unchanged HEAD source; see its checkpoint receipt. That unchanged Vedic facade suite was not selected again for this final run. No test suppression, known-issue addition or unrelated Shadbala/reader change was made.

.\.venv\Scripts\ruff.exe check moira/gochara.py moira/gochara_policy.py moira/vedic.py tests/unit/test_gochara.py tests/unit/test_gochara_constitutional.py tests/unit/test_api_surface_adversarial_audit.py --no-fix --output-format concise
git diff --check

Both pass. Ruff JSON comparison of current root/facade source against git show HEAD:<path>, with the same stdin filename and UTF-8 bytes, gives 59 root findings before/after and five facade findings before/after, with no introduced findings. These are existing audit debt, not a claim of repository-wide clean lint.

Scoped check_module_docstrings, check_class_docstrings and check_machine_contracts each give zero violations for both new engine modules. The initial checkpoint's Python example and both backend-standard Python examples execute successfully in the project runtime. All thirteen local/downloaded source hashes still match the research inventory; the books repository is clean. No astronomical authority corpus, forecast dates or predictive accuracy study was exercised.

The post-correctness lineage audit inspected helper structure, branch handling, assembly and proof framing in both modules. The policy catalogue owns explicit evidence/status records; omission/completeness branches follow named input scope; relation and graph assembly preserve typed source witnesses and their declared direction. The fixed planet order and sign arithmetic come from the named domain, with source tables kept private. No external-engine shape, ephemeris dependency, array-slot compatibility scheme, hidden score or software parity oracle was introduced. This is a scoped technique inspection, not a repository-wide sovereignty certification.

The user authorized committing and pushing this completed work to engine main on 6 October 2026; release and deployment remain separate actions. Earlier Gochar work and the separate REST-fix branch/commits are preserved. Astronomy, native code, dependencies, existing BAV/Muhurta/Sade Sati conventions, source books, test policy, REST, website/Workspace and generated moira.wiki are unchanged. Validation used Python 3.14.3 only; other supported Python versions were not run. Caller-supplied sidereal consistency and BAV birth/reduction provenance remain explicit assumptions.

The original 30 catalogue entries covered sixteen topics. Executable scope controls do not pretend to implement alternative textual schools. The source profile, natal-Moon reference and conservative classical participant universe remain inseparable. Attestation, disagreement, unknown provenance and product scope are visible status categories.

Future source profiles must carry a complete favorable/indication/relation corpus. Nodal baseline, blocker participation, Ketu narrative, exemptions and mean/true astronomy must be considered separately. Counter-Vedha needs explicit relief examples and overlap handling. Strength, dasha and dignity context must retain distinct testimony. Forecasts require a real sidereal ingress solver partitioned on subject and blocker events, including retrograde re-entry. No invented score closes any of these admissions.

The subsequent date-derived standard admits complete reader-bound natal/transit inputs and optional raw natal BAV composition. It adds eleven curated names and two engine-scope catalogue records (32 total, still sixteen topics), with two dated REST routes. The original 28-name core and eleven-name policy owner remain unchanged. The new validation receipt records its separate verification. No new textual school, BAV threshold, ingress solver or forecast window follows from this composition admission.

The separately authorized seven-stage package adds the whole-sign window standard and validation receipt. Its governing object is a complete seven-participant partition under the same Phaladeepika/Sastri doctrine, with all twelve date-policy ayanamsas, optional same-birth raw BAV, interval exclusion evidence and explicit uncertainty bands. Twenty curated names, two calculation routes and resource-free capabilities are additive; the 32-entry doctrine catalogue admits its existing forecast.dated_events record for this product only.

Common interval, clock/frame, astronomy, isolation, budget and reader-loan owners move to private _event_search modules with marriage import adapters. The focused review repaired angular-lift loss, request-meter bypass through a prewarmed owner's cache, and endpoint existence proofs that could drop the full interval's uncertainty. It also strengthened nested result invariants and exact engine/HTTP JSON accounting. No rule table or scientific tolerance was relaxed. Review, numerical-model qualifications, source cases, native-record parity, external evidence and actual synchronous performance remain distinct in the validation receipt. VED-018/019 are separate source admissions.

This checkpoint concerns local implementation and verification. Source commit, push, version release, deployment and website adoption require their separate publication steps.