Pillar B — Build guides

Published 2026-06-10 · Updated 2026-07-15 · By Avinash Chandan, Founder, Navamsha

Connect Claude or ChatGPT to Real Vedic Astrology Data

Connecting Claude or ChatGPT to real Vedic astrology data means giving the model structured chart JSON from a deterministic calculation engine — not letting it invent planetary positions. Navamsha hosts an astrology MCP server at navamsha.fastmcp.app; you can also wire the same data with OpenAPI-defined tools, an API key, and the agent guide at /llms.txt.

Hosted astrology MCP at navamsha.fastmcp.app

Also available: OpenAPI + API key + agent guide

Point your agent harness at three sources: (1) the live OpenAPI schema for exact paths and parameters, (2) an API key in the X-API-Key header, and (3) the llms.txt agent guide at www.navamsha.in/llms.txt for conventions, free tier limits, and response metadata rules. Prefer MCP when your client supports Model Context Protocol natively.

Define tools from OpenAPI

Most agent frameworks let you register HTTP tools. Map GET /api/v1/kundali/basic and GET /api/v1/dasha/vimshottari as explicit tools the model must call before answering chart questions. Constrain the system prompt: never guess positions; always fetch first.

const navamshaTools = [
  {
    name: "get_kundali_basic",
    description: "Fetch sidereal birth chart with lagna and planets. Requires date, time, latitude, longitude.",
    parameters: {
      type: "object",
      properties: {
        date: { type: "string", description: "YYYY-MM-DD" },
        time: { type: "string", description: "HH:MM 24h" },
        latitude: { type: "number" },
        longitude: { type: "number" },
      },
      required: ["date", "time", "latitude", "longitude"],
    },
  },
];

async function getKundaliBasic(args: {
  date: string; time: string; latitude: number; longitude: number;
}) {
  const q = new URLSearchParams({
    date: args.date,
    time: args.time,
    latitude: String(args.latitude),
    longitude: String(args.longitude),
  });
  const res = await fetch(`https://api.navamsha.in/api/v1/kundali/basic?${q}`, {
    headers: { "X-API-Key": process.env.NAVAMSHA_API_KEY! },
  });
  return res.json(); // includes data + meta
}

curl -G "https://api.navamsha.in/api/v1/kundali/basic" \
  -H "X-API-Key: $NAVAMSHA_API_KEY" \
  --data-urlencode "date=1990-01-15" \
  --data-urlencode "time=06:30" \
  --data-urlencode "latitude=28.6139" \
  --data-urlencode "longitude=77.2090"

import httpx, os

def get_kundali_basic(date: str, time: str, lat: float, lng: float) -> dict:
    r = httpx.get(
        "https://api.navamsha.in/api/v1/kundali/basic",
        headers={"X-API-Key": os.environ["NAVAMSHA_API_KEY"]},
        params={"date": date, "time": time, "latitude": lat, "longitude": lng},
    )
    r.raise_for_status()
    return r.json()  # pass data + meta to the LLM context

Claude vs ChatGPT wiring

• Claude (Anthropic): connect remote MCP at https://navamsha.fastmcp.app/mcp with X-API-Key, or register HTTP tools from OpenAPI.
• ChatGPT (OpenAI): use Assistants API function calling, Custom GPT Actions, or MCP connectors where supported — paste OpenAPI or point at the Navamsha astrology MCP.
• Cursor / Codex / Copilot: add the remote MCP server or docs/navamsha-api-agents.md (llms.txt) so codegen includes headers and env conventions.

MCP vs REST

The hosted astrology MCP replaces boilerplate tool definitions with discoverable curated tools and standardized transport. REST OpenAPI remains the full endpoint catalog. Underlying chart JSON and meta blocks stay the same either way.