Moira

Astronomy-First Ephemeris and Astrology Computation Engine

Moira is a sovereign, astronomy-first computation engine built for transparent astrological derivation, reproducible chart reduction, and an inspectable calculation chain from physical ephemerides to astrological outputs. Grounded in JPL DE441/DE440 state vectors, IAU 2000A/2006 precession-nutation models, ERFA/SOFA-aligned astrometry, and Gaia DR3-linked stellar catalogs, Moira pairs a zero-dependency Python orchestration and policy layer with a required C++17 numerical extension (_moira_native, compiled via pybind11).

Where conventional astrology libraries return terminal numbers from opaque defaults, Moira operates as a Light Box: computation stages are named and individually toggleable, doctrinal and source variants are explicit parameters rather than silent assumptions, and validation is maintained as version-bound, replayable evidence.

---

Table of Contents

• Why Moira & The Light Box Doctrine
• Requirements and Installation
• Quick Start
• The Reduction Pipeline
  • Worked Example: Mars at 2000-01-01 12:00 UT1
  • Pipeline Controls
• What Moira Computes
  • Positions and Celestial Catalogs
  • Chart Architecture and Traditional Western
  • Predictive and Time-Lord Techniques
  • Vedic / Jyotish Suite
  • Phenomena, Visibility, and Locational Astronomy
• REST API Server ()

---

Why Moira & The Light Box Doctrine

Astronomy is the foundation. Astrology is the purpose. Validation is the proof. Visibility is the doctrine.

Pillar	What Moira Guarantees
Sovereign Astronomical Substrate	Direct native C++17 SPK/DAF reading of published JPL DE441/DE440/DE430 Chebyshev records and Type-13 Hermite small-body shards. No jplephem, Swiss Ephemeris, or third-party runtime wrapper.
Light Box Inspectability	Every major reduction stage—light-time iteration, gravitational deflection, relativistic aberration, IAU 2006 frame bias, P03 precession, IAU 2000A nutation, WGS-84 topocentric parallax, and refraction—is exposed as a typed breakdown or ComputationTruth receipt.
Explicit Doctrinal Policy	Competing classical and modern traditions (house polar fallbacks, sect horizon conventions, Egyptian vs. Ptolemaic bounds, BPHS vs. Saravali vs. Jaimini profile lineages, Addey harmonic orbs) are represented by immutable policy objects and source-cited receipts rather than silent defaults.
Zero-Dependency Base Runtime	Base moira-astro requires only the Python standard library plus its compiled _moira_native extension. Optional FastAPI transport (moira-astro[server]) and lunar-graze tooling (moira-astro[lunar-graze]) live behind strict feature boundaries.
Audit-Ready Conformance	Across 1,252 canonical public calculations in 150 families, 100% of P0 production-user calculations (172 / 172) and 210 total canonical surfaces are governed by 64 exact-source replay contracts under LIGHT_BOX_CURRENT_STATUS.generated.md.

Who It Is For: Developers, ephemeris engineers, historians of astronomy/astrology, and researchers who require a programmable, audit-grade engine for reproducible pipelines and rigorous comparison against primary texts and physical oracles.

What It Is Not: Moira is not a consumer UI application, not a thin wrapper over opaque compiled stacks, and not convenience-first text generation without mathematical traceability.

---

Requirements and Installation

• Python: >=3.10 (3.10–3.15 supported; Python 3.14 is the canonical CI baseline and 3.15 has its own compatibility gate).
• Native extension: Installed automatically from published platform wheels, or compiled from source via a C++17 compiler, cmake >= 3.24, and pybind11 >= 2.12.
• Planetary kernel: A JPL DE-series .bsp kernel (de440.bsp, de441.bsp, or de430.bsp — downloaded separately; see Kernel Setup).

# Standard engine install (uses a published wheel when available)
pip install moira-astro

# With the FastAPI REST server (FastAPI, Starlette, Uvicorn, Pydantic)
pip install "moira-astro[server]"

# With Lunar Graze support (spiceypy, laspy, requests)
pip install "moira-astro[lunar-graze]"

[!NOTE] This README documents the 6.11.1 patch-release baseline. The PyPI badge identifies the latest packaged release; features merged after that release require a source checkout until the next version is published.

---

Quick Start

Moira initializes cleanly even before a planetary kernel is configured. Kernel-dependent calls (such as m.chart()) raise an explicit MissingEphemerisKernelError with remediation instructions until a kernel is present in kernels/ or ~/.moira/kernels/.

from datetime import datetime, timezone
from moira import HouseSystem, Moira
from moira.facade import planet_reduction_breakdown_at
from moira.houses import house_of
from moira.patterns import find_all_patterns

m = Moira()
dt = datetime(2000, 1, 1, 12, 0, tzinfo=timezone.utc)

# 1. Planetary chart reduction
chart = m.chart(dt)
print(f"Sun:  {chart.planets['Sun'].longitude:.6f} deg ({chart.planets['Sun'].sign})")
print(f"Moon: {chart.planets['Moon'].longitude:.6f} deg ({chart.planets['Moon'].sign})")

# 2. House cusps and placement (Placidus, London)
houses = m.houses(
    dt,
    latitude=51.5074,
    longitude=-0.1278,
    system=HouseSystem.PLACIDUS,
)
sun_house = house_of(chart.planets["Sun"].longitude, houses)
print(f"ASC: {houses.asc:.4f} deg | MC: {houses.mc:.4f} deg | Sun House: {sun_house}")

# 3. Aspect pattern detection
for pattern in find_all_patterns(chart.longitudes()):
    print(f"{pattern.name}: {', '.join(pattern.bodies)}")

# 4. Light Box inspectability: per-stage astrometric breakdown at JD_UT1 2451545.0
breakdown = planet_reduction_breakdown_at("Mars", 2451545.0)
print(f"Mars apparent longitude: {breakdown.final_longitude:.6f} deg")
print(f"Light-time delay (days): {breakdown.light_time_days:.6f}")

---

The Reduction Pipeline

flowchart TD
    A["JPL Planetary Kernel<br/>Chebyshev state vectors"] --> B["SSB Barycentric Position<br/>km · ICRF"]
    C["Sovereign Star Registry<br/>1,809 named stars"] --> D["Stellar Astrometric Position<br/>proper motion · parallax"]
    B --> E["1 · Light-Time Iteration<br/>body at t − τ  (τ = d/c)"]
    E --> F["2 · Gravitational Deflection<br/>Sun · Jupiter · Saturn"]
    F --> G["3 · Annual Aberration<br/>relativistic · IAU SOFA"]
    G --> H["4 · IAU 2006 Frame Bias<br/>ICRF → Mean Equator J2000"]
    D --> H
    H --> I["5 · IAU 2006 Precession<br/>P03 polynomial series"]
    I --> J["6 · IAU 2000A Nutation<br/>1358 luni-solar + 1056 planetary terms"]
    J --> K["True Equinox and Equator of Date"]
    K --> L["7 · Topocentric Parallax<br/>WGS-84 · optional"]
    K --> N["Ecliptic Projection<br/>True obliquity of date"]
    L --> R["8 · Diurnal Aberration<br/>observer rotation · optional"]
    R --> M["Atmospheric Refraction<br/>sky altitude only · optional"]
    N --> O["Zodiacal Longitude · Latitude · Distance"]
    K --> P["Sidereal Frame · Ayanamsa<br/>optional"]
    K --> Q["House Cusps · 22 Systems<br/>requires lat/lon"]

Worked Example: Mars at 2000-01-01 12:00 UT1

This snapshot is reproduced by planet_reduction_breakdown_at("Mars", 2451545.0, reader) with the admitted DE441 kernel. The input is JD_UT1 2451545.0; it is about 63.829 seconds after the exact J2000.0 TT epoch and is therefore not labelled as exact J2000.0.

Time: JD_UT1 2451545.000000 → JD_TT 2451545.000739 ($\Delta T = +63.829\text{ s}$)

Stage	Operation	Engine receipt	Longitude contribution
0	Geometric geocentric	SSB-to-Mars minus SSB-to-Earth; $\lambda = 327.975456^\circ$; distance = 276,697,408.4 km	reference
1	Light-time iteration	$\tau = 0.010683\text{ days} = 15.383\text{ min}$	$-15.762371^{\prime\prime}$
2	Gravitational deflection	Sun, Jupiter, and Saturn	$-0.003057^{\prime\prime}$
3	Annual aberration	Earth barycentric velocity; relativistic	$-14.069158^{\prime\prime}$
4	IAU 2006 frame bias	ICRF to dynamical mean J2000	$+0.000300^{\prime\prime}$
5	IAU 2006 precession	P03; mean equator/ecliptic of date	$+0.006571^{\prime\prime}$
6	IAU 2000A nutation	$\Delta\psi = -13.932004^{\prime\prime}$; $\Delta\varepsilon = -5.769413^{\prime\prime}$	$-13.932004^{\prime\prime}$
7–8	Topocentric parallax and diurnal aberration	disabled for this geocentric example	$0$

Final position: Aquarius 27° 57′ 48″ · $\lambda = 327.963300^\circ$ · $\beta = -1.067779^\circ$ · distance = $1.849688\text{ AU}$ · speed = $+0.775674^\circ/\text{day}$

Total visible reduction from the geometric J2000-ecliptic reference to the final true-of-date ecliptic longitude: $-43.759719\text{ arcseconds}$.

The stage contributions above come from the typed reduction receipt. They should not be reconstructed by subtracting unrelated coordinate frames or by treating apparent=False as "no frame transformation."

Pipeline Controls

Selected correction policies can be toggled via planet_at(). The table below shows the measurable effect of disabling each exposed policy in this JD_UT1 = 2451545.0 Mars snapshot.

Parameter	Default	Effect on snapshot longitude	Function
apparent=True	True	Full pipeline active	planet_at()
apparent=False	—	Omits light-time, deflection, and aberration while retaining the declared ecliptic-of-date frame and nutation policy. $\Delta = \mathbf{+29.835\text{ arcsec}}$	planet_at()
aberration=False	—	Aberration stage skipped. $\Delta = \mathbf{+14.069\text{ arcsec}}$	planet_at()
grav_deflection=False	—	Deflection stage skipped. $\Delta = \mathbf{+0.003\text{ arcsec}}$	planet_at()
nutation=False	—	Nutation skipped; mean equinox used. $\Delta = \mathbf{+13.932\text{ arcsec}}$	planet_at()
observer_lat/lon	None	When supplied together, adds WGS-84 topocentric parallax and observer-rotation diurnal aberration; the effect is body-, location-, and epoch-dependent.	planet_at()
refraction=True	True	Applies the declared pressure, temperature, and humidity model to geometric sky altitude.	sky_position_at()
delta_t_policy	None	Controls UT → TT conversion branch (IERS tables, polynomial, hybrid physical).	both

---

What Moira Computes

Positions and Celestial Catalogs

• Planets and luminaries — Geocentric and topocentric reduction with iterative light-time, annual aberration, multi-body relativistic deflection (Sun, Jupiter, and Saturn, excluding the target body), IAU 2006 frame bias, and WGS-84 topocentric parallax.
• Received-light cone (light_cone) — Explicit comparison of apparent observed ecliptic positions (at photon emission epoch $t - \tau$) against instantaneous geometric positions (at $t$), exposing one-way light travel time, emission Julian Date, and angular displacement.
• Fixed stars — Sovereign registry of 1,809 named stars with 3D proper motion, annual parallax, epoch propagation, and Stellar Quality classification. Audited anchor residual against SOFA/ERFA: 0.00048 arcseconds (J1000–J3000).
• Variable and multiple star systems — Phase and magnitude engine for eclipsing binaries and intrinsic variables (including a dedicated Algol API); Keplerian orbital mechanics for visually resolvable binary pairs (Sirius AB, Alpha Centauri AB) across VISUAL, WIDE, SPECTROSCOPIC, and OPTICAL system classes.
• Deep-sky coordinate anchors — A bundled, release-bound selection of 60 galaxies, nebulae, star clusters, compact objects, stellar remnants, and confirmed exoplanet host stars. Extended objects are represented by sourced SIMBAD ICRS/J2000 catalog centers rather than point-mass claims; host-star positions delegate to the sovereign star registry.
• Typed cosmic references and galactic frames — Full equatorial-to-galactic transforms, the legacy five-point compatibility table, and 12 fixed ICRS/J2000 directions kept in three explicit classes: physical objects, coordinate landmarks, and declared proxies (keeping the formal supergalactic origin distinct from the legacy Virgo/M87 astrological SGC, and exposing selected cluster anchors for broad structures such as the Great Attractor and Shapley Concentration).
• Position-capable asteroid ephemeris — Two layers:
  1. Wheel catalog (moira-asteroids-wheel release 2026.08.14.1): 25 named bodies (classical four, named Centaurs including Chiron, named TNOs, and selected others) bundled as one Type-13 shard so Chiron and the locked roster compute immediately after pip install plus a planetary kernel.
  2. External archive (release 2026.10.10.1): 11,223 selected asteroids in release 2026.10.10.1 (449 Type-13 SPK shards from JPL Horizons, a 10-day / 7-node base with certified adaptive apsidal sampling for 59 audited bodies, manifest-discovered under any kernel search root). The wheel also carries the matching 11,223-body canonical identity registry (names only, no positional vectors). Separate from the positional shards, Moira bundles the Proper25/PDS asteroid-family membership catalog (342 families, 200,726 unique numbered asteroids, 221,095 membership rows).

Chart Architecture and Traditional Western

• House systems — 22 systems including Placidus, Koch, Regiomontanus, Campanus, Morinus, Porphyry, Whole Sign, Equal, Alcabitius, APC, Krusinski-Pisa-Goelzer, Pullen Sinusoidal Delta/Ratio, and Sunshine. Includes branch-aware high-latitude doctrine where admitted, explicit polar fallback policy receipts, 3D house-boundary geometry curves, and house_of placement lookups.
• Aspects and motion witness — 46 standard zodiacal aspects plus 28 optional exact-fraction targets in H13, H32, H36, and H40 sets, a shared alias registry, and applying/separating/stationary signed motion witnesses. Select additions with optional_sets=("H13", "H32", "H36", "H40"); existing tier defaults are unchanged. Quindecile remains 24° and Tao remains 165°. Includes declination parallels and contra-parallels, antiscia and contra-antiscia, exact partile (is_partile) and orbed platic (is_platic) status markers, and geometric aspect strength. See the complete aspect catalog for every angle, exact fraction, alias, default orb, optional set, and harmonic preset.
• Aspect patterns and Jones chart shapes — 37 named multi-body configurations, including source-qualified Huber figures and ten mathematical constructions, plus opt-in exact harmonic triangles, rectangles and polygons. Normal searches cover 35 standard-aspect names; enabling H13 adds the two H13 Spectrum layouts. Exact layouts preserve rational-turn geometry, vertex assignments and orientation; aliases select canonical named detectors. H13/H32/H36/H40 targets remain optional. See the pattern catalog and migration rules. All 7 Marc Edmund Jones temperament chart shapes remain available with computation truth receipts.
• Planet-first stelliums — Explicit Strict (four-planet) and Broad (three-planet) core counts, independent sign/house/total-span evidence, and associated bodies that attach without manufacturing a cluster. See STELLIUM_ANALYSIS_STANDARD.md.
• Midpoints — Full midpoint matrix, midpoint trees, 90°/45°/22.5° dial projections, and planetary pictures.
• Traditional dignities and almutens — Domicile, exaltation, triplicity (Dorothean/Ptolemaic diurnal/nocturnal), Egyptian/Ptolemaic/Chaldean terms, face, sect, hayz, Reception, and Almuten Figuris.

Predictive and Time-Lord Techniques

• Progressions — Secondary, tertiary, minor, solar arc (longitude and right ascension), Naibod, and ascendant arc, with direct and converse variants and explicit rate-authority receipts.
• Primary directions — 8 admitted geometry methods across in-mundo and in-zodiaco policies (Placidus mundane, Ptolemy semi-arc, Placidian classic, meridian, Morinus, Regiomontanus, Campanus, and topocentric), exposed through 25 named presets, direct/converse doctrines, speculum computation, fixed-star targets, and Ptolemy/Naibod/Cardan/solar time keys.
• Returns, transits, and cycles — Solar and lunar returns; arbitrary planetary and small-body longitude returns; station-aware transit window search; and outer-planet synodic cycle tracking.
• Time lords and vitality — Annual profections with exact civil-anniversary-bounded monthly intervals; Zodiacal Releasing (Vettius Valens L1–L4 with loosing-of-the-bond); admitted Decennials L1/L2; Firdaria (diurnal and nocturnal sequences including the Bonatti variant); Lord of the Orb and Lord of the Turn; and Hyleg/Alcocoden vitality calculations.

Vedic / Jyotish Suite

Admitted Vedic surfaces name their governing classical source, modern lineage, or source-scoped witness (including Brihat Parashara Hora Shastra [BPHS], Brihat Jataka, Saravali, Phaladeepika, Uttara Kalamrita, Jataka Parijata, Jaimini Upadesa Sutras, and the named 1879 Agastya- and 2024 Bogamuni-attributed Pancha Pakshi witnesses). Where sources disagree or a modern composition is required, Moira exposes the policy or qualification instead of presenting one silent universal rule.

• Sidereal foundation and Panchanga — 12 admitted ayanamsha systems (including Lahiri, True Chitrapaksha, Raman, Krishnamurti, Fagan-Bradley, Pushya Paksha, and star-anchored True ayanamshas); 27-nakshatra and 28-Abhijit division with pada and ruler truth; and classical five-limb Panchanga (Tithi, Vara, Nakshatra, Yoga, Karana) plus daily sunrise-owned Panchanga (DAILY_PANCHANGA_STANDARD.md).
• Vimshottari, Kalachakra, and Alternate Dashas —
  • Vimshottari Dasha with exact nakshatra balance and multi-level sequence trees.
  • Kalachakra Dasha (bphs_santhanam_longitude_row_cycle_v1, exposed in Python and via GET /v1/dasha/kalachakra/profiles and POST /v1/dasha/kalachakra/sequence) with exact Savya/Apasavya BPHS/Santhanam source years, Deha/Jeeva signs, Gati jump classification, and bounded Mahadasha/Antardasha trees (KALACHAKRA_DASHA_STANDARD.md).
  • Alternate Conditional Dashas (Ashtottari, Shodashottari, Dwadashottari, Panchottari, Shatabdika, Chaturashiti Sama, Dwisaptati Sama, Shashtihayani, Shat-Trimshat Sama, and Yogini) with explicit applicability gates (ALTERNATE_DASHAS_BACKEND_STANDARD.md).
• Jaimini system and source-defined Chara/Argala profiles — Rasi drishti, arudha padas A1–A12 (Rath/JHora exception default, Raman variant as policy), karakamsa (Rath D9 vs. K.N. Rao D1), plus 6.10.0 source-defined Chara Dasha profiles and bounded Argala (argala) assessments across three Chara and three Argala source selectors (JAIMINI_PROFILE_ASSESSMENT_STANDARD.md).
• Yogas and source-defined Yoga extensions — 61 classical yogas across six families (Pancha Mahapurusha, Chandra, Surya, all 32 Nabhasa, Raja core, Dhana core) returned as proof objects with formation conditions and cancellation () clauses, plus source-defined Yoga Extensions covering four named profiles and 21 admitted rules with explicit frame/branch receipts ().

• Madras 1879 Fixed-Clock Schedule & Directed Relationship Profile: An explicitly named Agastya-attributed Madras 1879 aksara/query-or-name-initial fixed-clock schedule and directed relationship profile, with exact rational timing, source locators, declared omissions, and no implicit or universal default.
• Standalone Astronomical Paksha: Evaluates apparent geocentric Moon-minus-Sun longitude in the true ecliptic of date and maps [0, 180) Shukla/waxing to the profile's source-attested Purva label and [180, 360) Krishna/waning to Amara. It accepts no location and never selects a schedule, materializes a clock, supplies paksha to another operation, or infers natal identity.
• Local-Solar Context, Fixed-Clock, and Solar-Proportional Policies: An explicit modern local-solar context policy derives topocentric day/night and local-mean-solar weekday for an aware instant and location while requiring the caller to supply Purva or Amara. A separate modern fixed-clock policy anchors the selected schedule at sunrise or sunset, treats one nazhigai as exactly 1,440 SI seconds on reader-bound TT, and publishes half-open UT1 cell intervals plus their unclipped relation to the astronomical half end (returning unmaterialized_solar_half_tail when a long solar half outlasts the fixed span). A distinct modern solar-proportional policy maps nominal offset fractions across the complete governing day or night half on reader-bound TT (25 half-open TT/UT1 cells with exact boundary closure; status is always selected). The 1879 witness is not credited with that proportional rule.
• Stage 2I First-Samam EAT Seed Lookup: Exposes the selected 1879 generator's first-samam EAT seed from explicit profile Paksha, day/night half, and weekday without materializing the schedule or promoting that seed into a whole-day eating bird, Padu, authority, Adhikara/Bharana, condition, score, or forecast. All 28 cells are bound to the 1879 leaves and corroborated by the 1934 Uromarisi publication.
• Bogamuni 2024 Nakshatra & Padu Profiles: Preserves the 54-cell Purva/Amara-by-nakshatra bird table (where the Amara verse governs and the malformed adjacent commentary duplicating Shravana and omitting Revati is retained as rejected conflict evidence). The fixed modern bogamuni_2024_apparent_lahiri_natal_moon_identity_v1 composition evaluates apparent geocentric Sun and Moon positions on one reader-bound TT epoch, applies Lahiri true ayanamsa, assigns the Moon to one of 27 equal half-open nakshatras, and returns the selected source-table bird with every intermediate visible. A separate bogamuni_chennai_2024_padu_bird_mapping profile preserves the 14-cell Purva/Amara-by-weekday Padu-bird lookup without day/night axis or schedule/forecast routing.

Phenomena, Visibility, and Locational Astronomy

• Eclipses and occultations — NASA-canon contact solver for solar and lunar eclipses, Saros series classification with heptagonal vertex labelling, local eclipse circumstances, and lunar occultations of stars and planets.
• Visibility and heliacal phenomena — Heliacal and acronychal rising and setting, planetary elongation extremes, Yallop lunar-crescent classification, and opt-in physical point-source visibility (moira-physical-heliacal-visibility 1.2.0 data pack; see PHYSICAL_HELIACAL_VISIBILITY_CAPABILITY_MATRIX.generated.md).
• Stations, solar conditions, and void-of-course Moon — Retrograde/direct stationary-point search with velocity/acceleration motion truth; cazimi, combustion, and under-the-beams solar condition events; and void-of-course Moon windows with exact aspect-domain and ingress receipts.
• Locational astrology and parans — Astrocartography (ACG) horizon/meridian curves and crossing lattices, Local Space azimuth/altitude bearings, Gauquelin 36-sector plus-zone analysis, and paranatellonta (paran) field analysis with contour extraction.
• Harmonics and harmograms — Addey/Hamblin harmonic chart projection, inverse-harmonic orb conjunctions ($O_H = O_1 / H$), cluster pattern scoring, vibrational fingerprints, and continuous H1–H5 harmogram wave traces.
• Synastry, relationship, and Natal Discovery — Inter-chart synastry aspects and house overlays, midpoint composite charts, spherical-midpoint Davison charts, and bounded Natal Discovery packets (traditional dispositor chains, Jones 10-planet shape evidence, and source-verified natal lunar context).

---

REST API Server (moira_server)

Moira includes an optional FastAPI transport layer (moira_server) that exposes 531 OpenAPI operations (527 versioned /v1 routes plus 4 operational endpoints) with strict Pydantic request/response validation and embedded provenance receipts.

pip install "moira-astro[server]"
uvicorn --factory moira_server:create_app

For latency-sensitive deployments, opt in to a bounded per-worker startup warmup before accepting computational traffic:

$env:MOIRA_SERVER_PREWARM = "1"
uvicorn --factory moira_server:create_app

• Readiness vs. Liveness: With MOIRA_SERVER_PREWARM=1, /ready returns HTTP 503 until kernel-backed warmup succeeds, while /health remains immediately available for liveness checks.
• Transport Boundary: All astronomical and astrological doctrine resides in moira; moira_server provides typed serialization, input validation, bounded LRU caching, and OpenAPI discovery (/openapi.json, /docs, /redoc).
• Complete Route Inventory: See REST_API_REFERENCE.md for the generated inventory of all 531 operations.

---

Kernel Setup and Small-Body Archives

Planetary Kernels (JPL SPK)

Moira requires a JPL DE-series SPK planetary kernel for planetary calculations. All kernel reading is performed by Moira's native C++17 SPK/DAF reader (_moira_native) with zero jplephem fallback. Ordinary ephemeris calculations are strictly local and never make outbound network requests.

Kernel	File	Size	Date range	Notes	Direct JPL Download
DE441	de441.bsp	~3.1 GB	~13,200 BCE – ~17,200 CE	Full historical/deep-time target; maximum date coverage	Download DE441
DE440	de440.bsp	~114 MiB	1550 BCE – 2650 CE	Recommended modern-range kernel for most users	Download DE440
DE430	de430.bsp	~114 MiB	1550 BCE – 2650 CE	Widely deployed predecessor to DE440	Download DE430

Place your .bsp file in kernels/ (relative to the repository root) or ~/.moira/kernels/, or configure a custom path in Python:

from moira import Moira
from moira.spk_reader import set_kernel_path

set_kernel_path("/path/to/de440.bsp")
m = Moira()
print(m.is_kernel_available(), m.available_kernels)

Kernel Manager GUI, CLI, and Custom DAF Writer

# CLI: list kernel status or download missing planetary kernels
moira-download-kernels --list
moira-download-kernels --yes

# Tkinter GUI: browse, download with progress bar, and activate kernels
moira-kernel-manager

# Tkinter GUI: build custom Type-13 SPK kernels from JPL Horizons
moira-daf-writer

Small-Body Catalogs (Asteroids and Comets)

• Bundled Wheel Roster (moira-asteroids-wheel 2026.08.14.1): Ships inside the wheel under moira/kernels/asteroids_wheel/ with 25 named bodies (including Chiron, Ceres, Pallas, Juno, Vesta, Pholus, Nessus, Chariklo, Eris, Sedna, Haumea, Makemake, Quaoar, Orcus, Hygiea, Psyche, Eros, Lilith, Sappho, Iris, Astraea, Hebe, Flora, Metis, and Eunomia) ready for immediate calculation once a planetary kernel is present.
• Full External Catalogs (Release 2026.10.10.1):
  • Asteroids: 11,223 selected asteroids across 449 Type-13 SPK shards in release 2026.10.10.1 (with certified adaptive apsidal sampling for 59 audited high-eccentricity bodies).
  • Comets: 497 comets (1P/Halley through 516P) across 20 Type-13 SPK shards in release 2026.10.10.1 (with certified adaptive apsidal sampling for all 497 bodies).
  • Download shards and manifests from the Moira Small-Body Ephemerides Archive and place the complete release contents (manifest.json, SHA256SUMS, .bsp shards, and .metadata.json files) directly inside kernels/asteroids/ or kernels/comets/ (or ~/.moira/kernels/asteroids/ and ~/.moira/kernels/comets/).

Verify an extracted small-body catalog release or archive before use:

.\.venv\Scripts\python.exe -m moira.small_body_catalog_release verify C:\path\to\catalog-release
.\.venv\Scripts\python.exe -m moira.small_body_catalog_release verify-archive C:\path\to\catalog-release.zip

---

Data Inventory

Layer	Source	Bundled	Note
IAU 2000A/2006 nutation and precession tables	IAU	Yes	2,414 terms (1358 luni-solar + 1056 planetary terms); native C++ (_moira_native)
DE-series planetary kernel	JPL	No	de430 (~114 MiB), de440 (~114 MiB), or de441 (~3.1 GB); download separately
Named star registry	Sovereign (star_registry.csv + JSON provenance)	Yes	1,809 stars; license-independent
Deep-sky coordinate-anchor catalog	Moira selection; SIMBAD coordinates/identities; NASA Exoplanet Archive host confirmation	Yes	60 non-Solar-System anchors; per-artifact SHA-256 and source receipt included
Canonical small-body identity registries	Moira release catalogs with JPL identity lineage	Yes	11,223 asteroid and 497 comet identities; names and disambiguation only, with no positional vectors implied
Wheel asteroid ephemeris	JPL Horizons (Moira Type-13 packaging)	Yes	Catalog moira-asteroids-wheel 2026.08.14.1; 25 named bodies including Chiron; one Type-13 shard
Full position-capable asteroid ephemeris	JPL Horizons	No	11,223 selected asteroids; 449 Type-13 shards in release 2026.10.10.1; adaptive apsidal sampling for 59 audited bodies; public archive, manifest-discovered
Asteroid-family membership catalog	Proper25 + retained NASA PDS exclusions	Yes	342 families; 200,726 unique numbered asteroids; 221,095 many-to-many membership rows; no positional ephemeris implied

---

Native C++ Performance

Moira's required numerical extension (_moira_native) is implemented in C++17 with pybind11. Published wheels install the compiled binary directly; source builds compile it via CMake during installation.

• Native-Owned Numerical Substrates: IAU 2000A nutation, P03 precession, SPK/DAF segment reading and Chebyshev/Hermite evaluation, apparent planetary reduction via NativePlanetaryEvaluator, coordinate frame rotations, light-time iteration primitives, harmogram wave synthesis, and selected event-search kernels.
• Python-Owned Semantics & Policy: Public API contracts, doctrinal policy selection, source-conflict disclosure, and compositional orchestration remain in Python so numerical speed never obscures astrological or astronomical provenance.

---

Validation Evidence and Light Box Conformance

Moira is validated across a three-layer corpus—Astronomy (IAU ERFA/SOFA, JPL Horizons, NASA, IERS), Astrology (named primary texts, source-owned fixtures, doctrine-grounded invariants, and bounded secondary comparisons), and Experimental/Sovereign domains—governed by three gates:

1. Gate of Source — Inputs and reference tables are tied to an independent primary authority.
2. Gate of Flow — The computational derivation is explicit and inspectable.
3. Gate of Validation — Outputs satisfy a product-appropriate external authority, source-owned golden, independent invariant, or replayable release contract.

Light Box Conformance Program (GAP-007)

Beyond narrative validation ledgers, Moira enforces machine-checked Light Box Conformance across the entire public calculation denominator (moira.facade.__all__ + FastAPI OpenAPI):

• Complete Public Denominator: 1,252 canonical calculation surfaces (1,255 total including 4 aliases) classified across 150 reviewed families.
• 100% P0 Production-User Coverage: All 172 canonical P0 surfaces used by production consumers—plus 38 additional P1/P2 exemplar surfaces (210 total)—are structurally admitted and bound to 64 replayable evidence contracts (tests/evidence/lightbox/contracts.json).
• Replayable Release Gate: Exact-release receipts (scripts/replay_lightbox_receipts.py) and the release gate (scripts/check_lightbox_release_gate.py) verify contract coverage and inventory fingerprints while keeping remaining P1/P2 unresolved residuals explicitly visible rather than hidden behind universal claims.

Validation / Conformance Ledger	Scope and Verification Authority
LIGHT_BOX_CURRENT_STATUS.generated.md	Machine-generated exact-release conformance status (1,252 canonical calculations, 172/172 P0 partial/receipted, 64 passed replay contracts).
LIGHT_BOX_ASSURANCE_MATRIX.generated.md	Per-surface, per-evidence-class matrix (authority, doctrine_source, invariant, regression, native_parity, release).
LIGHT_BOX_CONFORMANCE_PROGRAM.md	Seven-stage Light Box conformance architecture and release-gate protocol.
RELEASE_VALIDATION_6.10.0.md	Integrated 6.10 release validation receipt covering Kalachakra Dasha, Chara/Argala profiles, Yoga extensions, and 2026.10.10.1 small-body catalogs.
VALIDATION_ASTRONOMY.md	Product-specific IAU ERFA/SOFA, JPL Horizons, NASA, IERS, and invariant evidence with named tolerances and qualifications.
PLANETARY_REDUCTION_PIPELINE.md	Major-body route identity, reader-bound time and context ownership, reduction stages, and target/time/frame-matched Horizons contracts.

---

AI and LLM Visibility

Moira is structured for high-integrity discovery and collaboration with AI coding and research agents:

• Machine-Readable Indexes: llms.txt provides a concise root summary and llms-full.txt provides the comprehensive documentation index.
• Agent Operational Constitution: AGENTS.md defines the Urania collaboration persona, protected zones, runtime laws, and verification gates.
• Canonical Wiki & Standards: Full architectural standards and validation ledgers live in wiki/.

---

Project Documentation and Artifact Governance

For release preparation or an explicitly requested CI audit, run the checked-in release verification steps from the project environment:

.\.venv\Scripts\python.exe scripts/check_ci.py

This reads release-hardening.yml and release-acceptance.yml directly, including the complete test_lightbox_*.py suite and generated-artifact checks. It is not required for routine engine commits or pushes. Install the project's dev and server extras beforehand and initialize the wiki submodule. The dev extra supplies IANA timezone data on Windows, matching the resource provided by Ubuntu CI; the runner checks its availability before the long gates. Use --list to inspect the plan. The command stops on the first failure, writes website candidate artifacts only to a temporary directory, and performs no installation, upload, release, or deployment. Jobs for a different Python version (including the Python 3.10 compatibility gate) are explicitly reported as CI-only. Kernel-backed cases use the harness's normal resource admission. When planetary data is installed, the selected small-body cases also need their complete catalogues; DE441 alone does not provide comet trajectories such as Halley's.

The canonical documentation tree lives in wiki/. The flat moira.wiki/ Git wiki mirror is synchronized from wiki/ and should not be edited by hand in isolation. Website-publishable documents are governed by website_docs/publication_sources.json.

Routine engine development uses focused tests for the changed behavior, relevant scientific/compatibility evidence, and lint/docstring checks on touched files. Commit and push directly to main when requested. Pull requests are used only when the maintainer asks for one; CI does not impose a PR workflow.

Update the source documentation needed to explain the change. Global generated metadata and the Git wiki mirror may wait until release or explicit documentation publication; they do not block routine commits or pushes. Deferred artifacts retain their previous evidence identity and must not be presented as current.

For release preparation, use the project .venv:

.\.venv\Scripts\python.exe scripts\maintain_generated.py --refresh

This refreshes mechanical outputs in dependency order and checks the result. Scientific receipt replay, reviewed admissions, product-use audits and release baselines retain their separate evidence requirements; they are never silently rewritten by this command. See the conformance program.

Release Hardening and Release Acceptance run on manual request and as required prerequisites of tag-triggered PyPI builds. Routine main pushes and optional PRs do not trigger these broad suites. Release work also requires the relevant local kernel-backed validation; hosted checks do not substitute for those resources. Wiki and website publication remain explicit, separate actions.

Document	Contents
CHANGELOG.md	Versioned release history and current unreleased boundary.
RELEASE_NOTES_6.10.1.md	Published 6.10.1 release notes, Light Box conformance summary, and Natal Discovery scope.
COMPATIBILITY_NOTES_6.10.1.md	6.10.1 public contract compatibility notes and migration boundaries.
API_REFERENCE.md	Canonical Python engine and facade reference (moira and moira.facade).
REST_API_REFERENCE.md	Generated FastAPI OpenAPI route inventory (531 operations) and transport contracts.
01_LIGHT_BOX_DOCTRINE.md	Transparency and visible derivation as foundational design constraints.
BEYOND_SWISS_EPHEMERIS.md	Capabilities enabled by sovereign catalogs, explicit policy, and modern Python/C++17.

---

License

MIT (c) 2026 TheDaniel166. See PROVENANCE.md for license and Swiss-lineage provenance clarity.