Give your AI agent accurate Vedic astronomy

Two-minute install

Add this to your Claude Desktop config

Paste the snippet below into your claude_desktop_config.json (macOS: ~/Library/Application Support/Claude/, Windows: %APPDATA%/Claude/). Restart Claude. Ask about panchang or a birth chart — the four tools appear automatically.

1. 1.Locate or create claude_desktop_config.json
2. 2.Paste the mcpServers block below
3. 3.Restart Claude Desktop — the tools appear under the MCP menu

{
  "mcpServers": {
    "dekhopanchang": {
      "command": "npx",
      "args": ["-y", "@dekhopanchang/mcp"]
    }
  }
}

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

MCP Tools

What your agent can do

Four tools cover the core computational surface. Every response includes a citation block naming the classical sources (BPHS, Surya Siddhanta, Muhurta Chintamani) and the ephemeris (Swiss Ephemeris DE441).

• get_panchang

  Five limbs of the daily almanac (tithi, nakshatra, yoga, karana, vara) plus sunrise / sunset, Rahu Kaal, Yamaganda, Gulika, masa, samvatsara for a given date and location.

• get_kundali

  Vedic birth chart from birth date, time, and coordinates. Ascendant, planetary positions with signs / nakshatras / houses / dignities, house cusps, Vimshottari mahadashas, and detected named yogas.

• get_muhurat

  Auspicious dates in a given month + location for a specified activity (marriage, griha_pravesh, mundan, vehicle, travel, property, business, education). Each date graded excellent / good / acceptable.

• get_matching

  36-point Ashta Kuta (Guna Milan) compatibility score between two people from their Moon-sign facts. Includes per-kuta breakdown and Nadi-dosha status.

Design constraints

Deliberate choices we made — and their reasons.

• Local execution

  The server runs on the user's machine. dekhopanchang.com is never contacted at query time. No per-call cost, no rate limits, no data leaving your box.

• No LLM inference

  Pure astronomical computation. If your agent wants natural-language interpretation, feed the JSON response to your model of choice — but the numbers are ground truth.

• Swiss Ephemeris DE441

  Sub-arcsecond planetary positions. Meeus fallback if the native binary fails to compile (Sun ~0.01°, Moon ~0.5°, outer planets 1-3°).

• Lahiri sidereal ayanamsha by default

  Indian government standard. Raman and KP are opt-in via the ayanamsha parameter on get_kundali.

• Citations in every response

  Every payload names its sources: BPHS, Surya Siddhanta, Muhurta Chintamani, Swiss Ephemeris DE441. Attribution-friendly for LLM-generated summaries.

• Deterministic

  Same inputs, same outputs, always. No hidden state, no A/B, no per-user calibration.

Requirements

Node.js 18.17 or newer. On first run the sweph dependency compiles a native Swiss Ephemeris binary (needs a C++ toolchain: Xcode CLT on macOS, build-essential on Linux, MSVC Build Tools on Windows). If the compile fails, the server still runs via the Meeus fallback with slightly reduced accuracy.