A Vedic astrology marriage compatibility engine that generates astronomically accurate birth charts and AI-ready structured data for two individuals. Built with Swiss Ephemeris for precision and designed so AI (ChatGPT, etc.) can provide deep astrological analysis from the output.

• Lahiri Ayanamsa + Whole Sign house system (standard Vedic setup)
• Sidereal planetary positions via Swiss Ephemeris
• 9 planets (Sun through Ketu) with sign, house, degree, nakshatra, pada, retrograde flag, and speed
• 12-house chart with lords and occupants
• Planet dignity classification (exalted / own / friendly / neutral / enemy / debilitated)

• Ashtakoota Guna Milan (8-point matching system, scored out of 36)
  • Varna, Vasya, Tara, Yoni, Maitri, Gana, Bhakoot, Nadi
• Dosha detection: Manglik (with cancellation logic), Nadi, Bhakoot
• House scoring with full AI-explainable breakdown (lord, dignity, occupants, aspects)
• Vimshottari Dasha timeline with current period and antardasha
• Derived timing windows for marriage and career

• Planetary Aspects (Drishti) — full aspect matrix with benefic/malefic classification
• Navamsa (D9) — divisional chart for marriage depth analysis
• Transit Snapshot — current Jupiter & Saturn positions relative to natal lagna
• House Strength Breakdown — lord, dignity, occupants, benefics/malefics, aspecting planets

• 6 ready-to-paste ChatGPT prompts covering:
  • 💍 Marriage Compatibility
  • 💰 Wealth & Finance
  • 💪 Health & Longevity
  • 👶 Children & Family
  • 💼 Career & Growth
  • ⚡ Overall Verdict (all-in-one)
• Each prompt includes a system prompt (expert Vedic astrologer role with tier-based verdict rules) and focused data extraction (only relevant chart data, not raw JSON dumps)
• Click-to-preview and one-click clipboard copy

• Secure Authentication: Google OAuth integration with HTTP-only JWT cookies
• Place autocomplete via geocoding (Nominatim / Google Places)
• MongoDB profile persistence (optional — saves entered profiles linked to user accounts)
• Strict validation on all chart output (no partial data ever returned)

kundali/
├── backend/
│   ├── main.py                  # FastAPI app, endpoints, CORS
│   └── services/
│       ├── ephemeris.py          # Swiss Ephemeris wrapper, chart computation
│       ├── chart_builder.py      # Chart enrichment, house scoring, doshas
│       ├── guna_milan.py         # Ashtakoota scoring (8 gunas)
│       ├── dasha.py              # Vimshottari Dasha computation
│       ├── aspects.py            # Planetary aspects (Drishti) matrix
│       ├── navamsa.py            # D9 divisional chart
│       ├── geocoding.py          # Place resolution + timezone
│       └── validation.py         # Strict chart completeness checks
├── frontend/
│   ├── index.html               # Single-page app
│   ├── app.js                   # UI logic + AI prompt generators
│   └── style.css                # Design system
├── requirements.txt
└── .env                         # MONGODB_URI, GOOGLE_PLACES_KEY (optional)

• Python 3.9+
• Swiss Ephemeris (pyswisseph)

cd kundali
pip install -r requirements.txt

Create a .env file:

# Optional — enables profile persistence
MONGODB_URI=mongodb+srv://user:pass@cluster.mongodb.net/kundali

# Optional — enables Google Places autocomplete (falls back to Nominatim)
GOOGLE_PLACES_KEY=your_key_here

# Optional — enables Google OAuth login
GOOGLE_CLIENT_ID=your_google_oauth_client_id
JWT_SECRET=your_secure_jwt_secret

python3 -m uvicorn backend.main:app --reload --host 0.0.0.0 --port 8000

Open http://localhost:8000.

{
  "boy": {
    "name": "Mayank",
    "dob": "1996-06-04",
    "time": "18:34",
    "time_accuracy": "exact",
    "place": "Rudrapur"
  },
  "girl": {
    "name": "Tanya",
    "dob": "1999-10-31",
    "time": "00:49",
    "place": "Saharanpur"
  }
}

Place can also be an object with pre-resolved coordinates:

"place": {
  "label": "New Delhi, India",
  "lat": 28.6139,
  "lon": 77.209,
  "timezone": "Asia/Kolkata"
}

{
  "boy": { <chart> },
  "girl": { <chart> },
  "guna_milan": { score, max_score, verdict, breakdown }
}

Each <chart> contains:

The app generates 6 specialized ChatGPT prompts from the chart data. Each prompt:

1. System prompt — Sets ChatGPT as an expert Vedic astrologer with rules for using only provided data, considering both D1/D9, factoring aspects, and giving clear tier ratings
2. Focused data — Extracts only the relevant subset (e.g., marriage prompt only sends 7th house, Venus, D9, doshas — not the entire JSON)
3. Analysis questions — Specific questions for the topic area
4. Verdict requirement — Forces a clear 🟢/🟡/🟠/🔴/⛔ rating

• Swiss Ephemeris provides sub-arcsecond precision for planetary positions
• Whole Sign houses — each house = one full sign, starting from the ascendant sign
• Lahiri Ayanamsa — the most widely used ayanamsa in Indian Vedic astrology
• Vimshottari Dasha — 120-year cycle based on Moon's nakshatra at birth
• Navamsa (D9) — each sign divided into 9 parts (3°20' each), mapped by element cycle
• Drishti aspects — all planets aspect 7th; Mars +4th/8th; Jupiter +5th/9th; Saturn +3rd/10th

While this system uses the high-precision Swiss Ephemeris for astronomical calculations, astrological accuracy for historical birth dates depends heavily on Timezone Resolution.

Historical timekeeping in India was complex. Before 1955, many cities observed local municipal times rather than the Indian Standard Time (IST):

• Bombay Time (UTC +4:51): Widely used in Mumbai until 1955.
• Calcutta Time (UTC +5:54): Widely used in Kolkata until 1948.
• War Time (UTC +6:30): Observed across India during WWII (Oct 1941 – Oct 1945).

For international births (USA, Europe, etc.), the system uses historical IANA databases to resolve Daylight Saving Time (DST). While highly accurate, political and local changes can occasionally lead to drift in very old records.

Recommendation: For any birth before 1960, we recommend cross-verifying the "UTC Offset" against official birth records to ensure the chart's Lagna (Ascendant) is correct.

Internal use only.