VED-009 implementation and validation receipt

Date: 9 October 2026. Baseline: engine c33dca1, generated wiki 37681cb, version 6.9.9. State: implemented and validated; included in the containing authorized engine/wiki source-publication package. This receipt covers the selected six-detector/three-exception package. Release, deployment and website adoption remain separate.

The source packet and admission standard own the finite scope, exact profiles, inspected editions and explicit exclusions.

1. Evidence and implemented surfaces

Evidence class	What was verified
Textual/source arithmetic	All 27 MC/KP offsets, KP dual Mula, both clocks, the Rohini worked example, both seven-row Yamaghanta tables, both Gandanta width profiles and three scoped exceptions. Independent fixture transcription from identified pages; production constants are not imported as expected answers.
Pure boundary and admission	Rational starts/widths, every relevant Gandanta parent, half-open boundaries, short-Lagna clipping, previous-star carryover, nullable prerequisites, opt-in/source exclusions and hostile input rejection.
Analytical day	Independent linear sky and analytic altitude, off-midpoint cell states, sunrise weekday ownership, complete parent context, merged root bands, a derived uncertainty band whose midpoint precedes sunrise, bounded invalid/nonmonotonic failures and reader cleanup on exceptions.
Real astronomy	Six DE441 date/frame cases, independent solar and phase residuals on both sides of roots, dense/off-midpoint states, DST, polar absence/partial Lagna and active/explicit reader parity. These use the existing astronomical substrate; they are not external astrological oracle comparisons.
Public/REST	Fifteen exact curated export identities and three facade methods; strict nested request/response schemas, canonical error envelopes, full lossless engine/HTTP projection and both configured/discovered serving readers.
Compatibility/release	Existing named Muhurta and Panchanga Shuddhi tests; all pytest file groups selected by the Release Hardening workflow, exact public API snapshots, class docstrings, generated inventories and publication manifest.

The source fixture is tests/fixtures/muhurta_dosha_sources.json. The full source matrix exercises three tables, two clocks and three parent durations; Gandanta tests visit all 27/30/12 parents for both profiles and six edge probes. Every weekday/star combination is checked, together with the boundaries of the necessary-work half-window. Cancellation retains source restrictions, parent identity and prerequisite evidence, including source exclusions.

The six real date cases cover Delhi 9/10 October 2026, New York 8 March and 1 November 2026, equator/UTC 2 January 2000 and latitude 65.99 degrees on 22 September 2026. Each day is probed at 48 interior grid points and two off-midpoint positions in every cell. Solar altitude and full parent phase roots are independently re-evaluated at their lower/upper brackets. Root tolerance remains 0.1 seconds; no threshold was relaxed to pass.

Tromso summer/winter solar absence returns unavailable; its equinox day retains non-Lagna results with explicit partial status. Configured and discovered FastAPI startups use their own reader and match the canonical engine result exactly. Readers remain caller-owned and active contexts restore on success, coverage failure, missing resource and unexpected exceptions.

2. Execution receipts

Runtime: C:/dev/moira/.venv/Scripts/python.exe, Python 3.14.3. MOIRA_NO_DOWNLOAD=1, MOIRA_TEST_MODE=1, MOIRA_STRICT_KNOWN_ISSUES=1; no known-issue waiver. Tests deny network access except marked loopback. The discovered planetary kernel is C:/Users/nilad/.moira/kernels/de441.bsp, content identity DE-0441LE-0441.

Private machine-readable receipts are under C:/dev/outputs/ved009-research-2026-10-09/. The feature and compatibility acceptance contains 387 distinct passing tests, zero failures/errors/skips:

• 153 new cases: 77 source/pure boundary cases, 21 analytical/lifecycle/ public cases, 43 REST cases and 12 DE441 integration cases.
• 234 existing regression cases: 57 named-Muhurta, 104 special-Muhurta and 73 Panchanga Shuddhi cases.
• The final complete 150-case feature run took 108.30 s and recorded 15 successful DE441 resource uses, no resource skips/failures. Three additional uncertain-versus-missing-prerequisite cases were then added; the complete 141-case unit/analytic/REST slice passed in 22.92 s after that state-label refinement. The final 234-case compatibility run passed in 2.57 s.

Owning XMLs: final-dosha.xml, final-contracts.xml, final-regression.xml; feature-acceptance.json deduplicates their test IDs. Earlier exploratory runs are not added to these totals. In particular, a test-only uninitialized facade fixture was corrected before the passing final runs; the earlier failed fixture run is not the acceptance receipt.

$env:MOIRA_NO_DOWNLOAD = "1"
$env:MOIRA_TEST_MODE = "1"
$env:MOIRA_STRICT_KNOWN_ISSUES = "1"
.\.venv\Scripts\python.exe -m pytest tests/unit/test_muhurta_dosha.py tests/unit/test_muhurta_dosha_day_contract.py tests/server/test_server_muhurta_dosha.py tests/integration/test_muhurta_dosha_ephemeris.py -q
.\.venv\Scripts\python.exe -m pytest tests/unit/test_named_muhurta.py tests/unit/test_special_muhurta.py tests/unit/test_panchanga_shuddhi.py -q

The private run_release_tests.py reads the current .github/workflows/release-hardening.yml and selects all 55 distinct pytest files from its four release groups, including route discoverability. It runs them in one process with the same strict environment and network exclusion; the selected filenames are saved in release-test-selection.json. This is local Python 3.14 execution, not a new hosted Python 3.10 CI result. Python 3.10 grammar admission is checked separately for changed Python files.

Sixteen changed Python files pass Python 3.10 grammar checks. The standard's Python example executes successfully. Full new-file Ruff and the complete CI source check (F401,F821,F822,F823) pass. No test policy, threshold or exact API assertion was weakened.

Final Release Hardening result

All 55 selected files completed in 718.91 s: 1,612 passed, one skipped, zero failures/errors (release-gates.xml). The existing test_no_aspect_voc_last_aspect_is_none skipped because its 30-day scan found no qualifying no-aspect window; RULE-06 was not exercised. This is not a VED-009 skip or a missing-kernel skip. All 216 DE441 resource uses and the one supplemental small-body resource check ran successfully.

The deduplicated combined result is 1,999 passed, one existing skip, zero failures/errors across 2,000 distinct test IDs, recorded in acceptance-summary.json. Feature plus release runs used DE441 successfully 231 times. Repeated feature/contract runs are not added to either total.

All six release-facing artifact checks pass: documentation consistency, release identity, Hellenistic generated inventories, REST reference, generated Git wiki and website publication manifest. The REST inventory has 495 paths/operations. Hellenistic generated content is unchanged. Canonical and generated pages are synchronized, including the final receipt. Whitespace checks pass. These are local execution results; subsequent source publication does not convert them into a hosted CI receipt.

3. Publication and limits

The REST reference, generated wiki and website_docs/publication.json must be refreshed in that order after source/doc edits. The publication manifest retains the existing immutable v6.9.9 release identity and document allowlist; refreshing hashes is not a new release. New API snapshot entries are explicit admissions; exact equality guards remain enabled. Generated Hellenistic inventory content must not acquire unrelated Vedic route-count changes.

Completion means the finite named computation and transport package is implemented and verified. It does not establish predictive efficacy, universal source agreement, all twenty-one doshas, all remedies or complete purpose elections. The source record explicitly assigns strength/aspect/Navamsa exceptions to VED-010 and activity elections to VED-011. Other Tyajya and Gandanta variants remain separately identified research extensions; classical availability is not denied. No native substrate, dependency, version, legacy score/search default, website or Urania code changed.