The astrology API
you can verify.

Thirty-nine endpoints, one contract

Everything below is served from one versioned /v1 API, documented in OpenAPI with request and response schemas, and mirrored by an MCP server for AI agent integrations.

Don't take our word for it

Our accuracy isn't a self-reported percentage. The engine's positions are checked against reference values from authorities outside our own code: NASA/JPL, astro.com and the Tamil panchangam. We took each value from its source once and stored it with where it came from, and the running engine is measured against them. These are real checks from that report, 57/57 passing right now.

For AI agents

Built for agents, not just servers

The same 39 endpoints are exposed as an MCP server, so Claude, Cursor, or your own agent can discover and call them with typed input and output. One key, the same schemas as REST, structured data back (not prose to parse).

// POST /v1/mcp: Streamable HTTP, one API key (same as REST).
// An agent discovers the whole catalog, 39 tools, typed in AND out:
{
  "tools": [
    { "name": "chart",               "title": "Sidereal chart" },
    { "name": "dashas_vimshottari",  "title": "Vimshottari dasha" },
    { "name": "match_compatibility", "title": "Compatibility summary" },
    { "name": "panchang",            "title": "Panchang for a date + place" },
    { "name": "interpret",           "title": "Grounded interpretation (safety-passed)" },
    { "name": "safety_check",        "title": "Astrology-content safety check" }
    // … 33 more
  ]
}

What you can build

From a matrimony match to a panchang feed to an AI assistant, the calculations are the same, the safety is the same, the contract is the same.

How Lumo compares

The trust axis is the whole point: most astrology APIs ask you to take accuracy and safety on faith. Ours are checkable.

Built to integrate

The developer experience surface, in full. During the beta we issue keys and SDKs directly, and the contract and the public proofs are live today.

Questions developers ask

How do I authenticate?

An x-api-key header on every request, the same key for REST and the MCP server. Keys are issued during the private beta. Self-serve keys and rotation arrive with general availability.

What are the rate limits?

The free sandbox is 1,000 calls a UTC month, 10 a minute. Paid plans raise the monthly cap, and the limit per minute scales with your plan. You get clear headers, never a silent cutoff.

What happens to the birth data I send?

It is used to compute the response. No profile building, no resale. The calculation endpoints keep nothing but your usage counts. /v1/interpret is the one exception: every answer the safety guard changed, and about one in twenty of the rest, is kept with its question and chart facts so our team can review it for safety.

How accurate is it, really?

We verify calculation, not prediction. The engine's positions for a set of reference charts and dates are checked against values from authorities outside our own code, and the report is public and updates on every release. See the live accuracy report →

Is there an SLA?

Not yet. We are in private beta, and we would rather say so than promise uptime we have not measured. Growth plan partners get an uptime commitment as we leave beta, and the engine status is on /v1/health today.

Why should I trust the safety claim?

Because you can test it. POST your own astrology text to /v1/safety/check and watch the guard's fixed rules pass it, soften it, or refuse it. On /v1/interpret, every reading is also built only from what the engine computed, and a sample goes to our team for review. See the safety guard →

What does it cost?

Flat, published pricing in ₹ and $. Sandbox free, Starter ₹1,499/$19, Growth ₹6,499/$79. No “contact sales to learn the price.”