TIER2_VEDIC_WORK_TRACKER - TheDaniel166/moira GitHub Wiki
Tier 2 Vedic Competitive Edge Work – Paused State
OUTDATED — historical record only (6 October 2026). This May checkpoint is not the current Vedic backlog. Several absence and implementation claims differ from current code. Use the Vedic remaining work register. Earlier source leads and pauses below are preserved for provenance; these historical resumption instructions do not authorize new implementation.
Last active: 2026-05-29 (second resumption push)
Status: Paused by user request (new focus declared). Rich state captured for clean future resumption.
Context:
Per MOIRA_COMPETITIVE_ANALYSIS_2026-05-25.md, after Phase 8 the highest-leverage remaining work for competitive Vedic capability is in Tier 2 (practitioner workflow gaps). The user chose to prioritize closing these before moving into Phase 9 (Vedic transport surfaces).
Current Overall State
- Phase 8 is formally closed (see
MOIRA_SERVER_PHASE8_LEDGER.mdv1.9). - Decision was made to focus on Tier 2 Vedic core work instead of immediately starting Phase 9.
- Primary vehicle:
moira/muhurta.py(Muhurta / Electional doctrine layer). - General infrastructure (
moira/electional.py+moira/panchanga.py) is already strong. - Work is research-driven only (BPHS, Muhurta Chintamani, Brihat Samhita, etc.) — no reliance on internal model memory.
Muhurta Module – Present State (as of pause)
What Has Been Implemented
Baseline (pre-2026-05-29 resumption):
- Core vessels:
MuhurtaPolicy,MuhurtaClassification,MuhurtaScore - Dagdha Yogas (BPHS Ch. 85), Vishti Karana
- Named Muhurtas (Abhijit, Brahma)
- Starter
ACTIVITY_MUHURTA_GUIDANCE muhurta_scorer+find_best_muhurta_windowsintegration
2026-05-29 Resumption Push (before second tabling):
- Panchanga Shuddhi Depth (High): Full Tara Bala (9 Taras + 3 Paryaya cycle nuance + proper 0-26 NakshatraPosition indexing). Basic Panchaka Rahita (mod-9 formula).
- Dosha Detection + Parihara (High):
detect_muhurta_doshas()fully implemented and wired — Vishanadi/Tyajya, Yamaghanta, expanded Gandanta (Nakshatra + Tithi). Heavy scoring penalties. Parihara notes. Time-data support for full detection in electional scans. - Lagna & Planetary Strength (High): New
evaluate_muhurta_lagna_strength(chart, activity=None). Jupiter in Kendra from Muhurta Lagna, benefics influencing Lagna, basic shuddhi factors. Foundation wired into scorer path. Activity-specific Lagna notes. - Activity-Specific Rules: Incremental classical granularity (Kaulava for love/choosing bride, Gara for construction, etc.) + enhanced notes.
- All work strictly source-driven with explicit citations (Muhurta Chintamani, Kalaprakasika, Ernst Wilhelm Classical Muhurta Ch.10, 15, 17–18, B.V. Raman, K.K. Joshi, etc.).
- Rich
MuhurtaClassificationnow carries tara, panchaka_rahita, doshas, dosha_severity, lagna strength data. - Full integration through the electional scanner.
Exposed via moira/__init__.py and main Moira facade.
What's Missing / Needs Expansion (Prioritized)
| Area | Current Maturity | What's Missing | Priority for Competitive Edge | Notes / Sources |
|---|---|---|---|---|
| Panchanga Shuddhi Depth | Good progress (as of second pause) | Full Tara Bala (9 Taras + cycles + proper indexing). Basic Panchaka Rahita. Chandra Bala skeleton + more granular Yoga/Karana still needed. | High (partially addressed) | Muhurta Chintamani, Kalaprakashika, Ernst Wilhelm Classical Muhurta Ch.10 |
| Activity-Specific Rules | Good starter + incremental | Classical Karana/activity mappings added (Kaulava for love/choosing bride, Gara for construction, etc.). Enhanced marriage notes. Still needs deeper purpose-specific variants (love vs arranged, Vastu subtypes, travel/business specifics). | High (solid incremental) | Ernst Wilhelm Classical Muhurta (Ch.22 + Karana chapter), Muhurta Chintamani |
| Named & Special Muhurtas | Basic (2 done) | Godhuli, Vijaya, Amrita, Ravi Yoga, Sarvarthasiddhi, etc. + exact windows | Medium | Muhurta Chintamani + traditional lists |
| Dosha Detection + Parihara | Strong progress (as of second pause) | detect_muhurta_doshas() fully wired (Vishanadi/Tyajya, Yamaghanta, Gandanta). Scoring penalties, parihara notes, time-data support in scanner. More doshas + complete neutralization logic still open. |
High (major wiring complete) | Ernst Wilhelm Classical Muhurta Ch.17–18 + Vishanadi section, Muhurta Chintamani, Kalaprakasika |
| Lagna & Planetary Strength | Foundation implemented (as of second pause) | evaluate_muhurta_lagna_strength(chart, activity=None) delivered. Jupiter in Kendra, benefics on Lagna, basic shuddhi. Lightly wired. Deeper integration, Navamsha shuddhi, full chart strength rules still needed. |
High (foundation complete) | Ernst Wilhelm Classical Muhurta Ch.15 "Lagna", Muhurta Chintamani |
| Integration Helper Quality | Good | True high-performance ranked search that computes full Panchanga + Muhurta score at scan time, activity-aware scoring, better handling of sidereal vs tropical | Medium | Currently works but can be tighter |
| Validation & Testing | None | Tests that match behavior of Jhora / Kala / Parashara's Light on real dates | High | Essential before claiming competitiveness |
| Documentation | Code-level only | Proper wiki/Muhurta standard document, usage examples, limitations | Medium | — |
Other Tier 2 Vedic Items (from Competitive Analysis)
These are also flagged as remaining gaps but have received zero work yet:
| Item | Current Engine State | Priority | Notes |
|---|---|---|---|
| Jaimini Chara Dasha | Karakas exist (good). Full time-lord calculation system does not exist. | High | Often requested by serious Jaimini practitioners |
| Natal Yoga Catalog | Some building blocks exist. No systematic named yoga catalog (Raja, Dhana, Nabhasa, etc.) | Medium | Big for interpretive depth |
| Varshaphal / Tajika Refinement | Already quite strong (P8-11/12/13). Remaining work is long-tail yogas and polish | Medium | Lower urgency than Muhurta |
| Research & Filtering Tools | Almost nonexistent | Medium | Would benefit all Vedic work |
| KP Astrology | Nothing | Low (per analysis) | Very specialized, high effort |
Why This Is Being Tabled (Second Time)
- User declared a new focus.
- Work is in an excellent pausable state with rich, source-cited implementation and clear next steps.
- All doctrine added strictly from classical sources only (Muhurta Chintamani, Kalaprakasika, Ernst Wilhelm Classical Muhurta, etc.).
This is the second deliberate pause. Previous pause was 2026-05-29 initial; this resumption delivered major high-priority advances before tabling again.
How to Resume
- Read this file (
TIER2_VEDIC_WORK_TRACKER.md) — it now contains the complete record of two resumption pushes. - Read the latest central memory (see below) and
.remember/remember.md. - Start with the highest remaining priority in the updated "What's Missing" table.
- Maintain the strict "research from source material only" rule with explicit citations.
Key new surfaces from latest push (ready for use):
detect_muhurta_doshas()evaluate_muhurta_lagna_strength(chart, activity=None)- Enhanced
MuhurtaClassification(tara, doshas, dosha_severity, panchaka fields, etc.) - Full wiring through
muhurta_scorerandfind_best_muhurta_windows
Last updated: 2026-05-29 (second pause — rich state captured after major high-priority advances)