Connect Any AI Agent to Verified Astronomical Data with the Open Ephemeris MCP Server

Ask any large language model where Venus is right now. You will get a confident answer. You will probably get a wrong one.

This is not a flaw in the model. Language models predict tokens — they do not track planetary motion. Astronomy is deterministic, meaning it can be computed precisely, not learned from text. Asking a model to recall a planetary position is like asking a historian to report today's weather. The training data is just not the right tool for the job.

The Open Ephemeris MCP server exists to close that gap. It gives any MCP-compatible AI agent — Claude, ChatGPT, Gemini CLI, Cursor, Windsurf, and more — direct, tool-call access to real ephemeris data calculated from JPL DE440 files, the same source NASA uses for deep-space mission planning. Your agent asks. The calculation engine answers. The position is precise.

What MCP Is, and Why It Changes This

MCP — Model Context Protocol — is an open standard for connecting AI agents to external tools and data sources. When a model is configured with an MCP server, it gains access to callable tools that run outside the model itself. The model decides when to call a tool, what arguments to pass, and how to use the result in its response.

Think of it as giving the model a set of instruments it can pick up mid-conversation. A calculator for math. A search engine for current events. And now: a calculation engine for the night sky.

The Open Ephemeris MCP server exposes over 100 tools across the full range of astrological traditions — natal charts, transits, synastry, astrocartography, Human Design, BaZi, eclipse windows, lunar phases, planetary hours, and more. Each tool returns structured, typed data computed from verified ephemeris files.

What the Model Gets

When a conversation triggers an astronomical query, the model calls a tool. The tool call might look like this:

{
  "tool": "get_planet_positions",
  "arguments": {
    "datetime": "2026-04-23T14:00:00Z",
    "location": { "lat": 51.5074, "lon": -0.1278 },
    "planets": ["sun", "moon", "venus", "mars"]
  }
}

The server responds with calculated positions — longitude, latitude, speed, retrograde status, sign, degree — which means your agent can say "Venus is at 28°14' Pisces, direct, applying to a conjunction with Neptune" and be correct.

No hallucination. No estimated recall from training data. A calculation, run at the moment the question is asked.

Two Minutes to Connect

The server is live at https://mcp.openephemeris.com/mcp — a Streamable HTTP endpoint compliant with the MCP 2025-11-05 specification. You can point any compatible client at it immediately with an API key from your dashboard.

Claude Desktop:

{
  "mcpServers": {
    "openephemeris": {
      "url": "https://mcp.openephemeris.com/mcp",
      "headers": { "X-API-Key": "YOUR_API_KEY" }
    }
  }
}

Cursor / Windsurf / VS Code (Copilot):

{
  "mcpServers": {
    "openephemeris": {
      "command": "npx",
      "args": ["-y", "@openephemeris/mcp-server"],
      "env": { "OPENEPHEMERIS_API_KEY": "YOUR_API_KEY" }
    }
  }
}

Gemini CLI (~/.gemini/settings.json):

{
  "mcpServers": {
    "openephemeris": {
      "httpUrl": "https://mcp.openephemeris.com/mcp",
      "headers": { "X-API-Key": "YOUR_API_KEY" }
    }
  }
}

Restart the client. The tools appear. That is the full setup.

What You Can Build

The tools are not just a novelty for chatting about horoscopes. They are a building block for production products.

A natal chart reading agent that calculates the full chart from a birth date and location, then interprets each placement with a configured voice and context — no astrology math in your application code, just data in, interpretation out.

A transit alert system that queries upcoming planetary aspects against a stored birth chart and notifies users when a major event — a Saturn return, a Jupiter conjunction to their natal sun — is within a configurable orb.

An astrocartography tool that maps planetary lines across a world map for any birth data, letting a user ask their AI assistant "where in Europe would Jupiter be on my MC line?" and receive a precise answer with coordinates.

A Human Design profile generator that computes gates, channels, and centers from birth data and feeds them to a reporting agent — which means you can build a personalized HD reading pipeline without implementing the Rave Mandala yourself.

These are not hypothetical. The endpoints exist and are live. The free tier is enough to explore all of them.

The Calculation Engine

This is worth saying explicitly: the data does not come from a language model's memory. Each API call triggers a deterministic calculation against JPL DE440 ephemeris files — which means positions accurate to arc-second precision across a date range from 1550 CE to 2650 CE.

The planetary engine is independent. It runs in Go, handles coordinate transforms, applies the requested ayanamsa for sidereal traditions, and returns results in sub-100ms at production load. The MCP layer is a clean interface on top of that engine. The model does not do the math — it calls the tool, and the tool does the math.

Connecting

The server is available now. A free tier gives full tool access at a rate suitable for exploration and development. Paid tiers lift rate limits for production applications.

The Smithery registry listing includes one-click configuration snippets for Claude and other clients. The full documentation is at openephemeris.com/docs.

If you build something with it — tell us. We are genuinely curious what comes out of putting real astronomical data in the hands of agents that are very good at interpretation.