Falcı Hatun MCP Server
Provides real astrology and tarot computations using actual ephemeris and a 78-card deck, returning structured data such as natal charts, synastry, transits, and tarot draws without relying on an LLM for astrological facts.
README
Falcı Hatun — Astrology & Tarot Engine
An OKX.AI A2MCP service. It returns facts, not prose — because the agent calling it is already a language model.
Most "AI astrology" products bolt an LLM onto a vague prompt and let it improvise the sky. This one does the opposite. Falcı Hatun computes the chart with a real ephemeris and hands the agent structured truth: every planet's longitude, sign, house and retrograde state; every aspect with its orb; every transit bisected to the exact moment it perfects; the drawn cards and their meanings.
The agent writes the reading. That split is the whole design — and it means this service needs no LLM of its own: no API key, no vendor bill, no rate limit, and an answer that is byte-identical every time you ask.
Agent → "birth data" → Falcı Hatun → the actual sky → Agent writes the reading
Tools
| Tool | What it returns |
|---|---|
natal_chart |
Sun, Moon, Ascendant, Midheaven, 10 planets (sign, degree, house, retrograde), 12 house cusps, all major aspects with orbs. |
synastry |
Real cross-aspects between two charts, scored across four dimensions using classical significators, plus house overlays. |
transits |
When transiting planets form exact aspects to a person's natal placements — a dated timeline, searched from the ephemeris. |
sky_ahead |
Real New/Full Moons, real retrograde stations, real sign ingresses. Every date solved for. |
tarot_draw |
A full 78-card Turkish Rider-Waite deck, shuffled and dealt, upright/reversed, with meanings. Seedable for reproducibility. |
reference |
Signs, planets, aspect orbs, spreads, and every recognised birthplace. |
Chart engine: astronomy-engine (JPL-grade positions) + luxon (IANA timezone rules, so historical DST is handled rather than assumed). House system: Equal — stated in the response, because a chart that hides its house system cannot be verified.
Bilingual by construction
This grew out of a Turkish app, and OKX.AI is a global marketplace. A tool that answers "Kılıç Dokuzlusu: kaygı, endişe, uykusuzluk" to an English-speaking agent has not really answered.
So nothing is replaced — everything is doubled. Every sign, planet, aspect, spread position and card meaning ships in both languages on the same response:
{ "name": "Boğa", "en": "Taurus", "element": "toprak", "elementEn": "earth" }
{ "name": "Kılıç Dokuzlusu", "nameEn": "Nine of Swords",
"meaning": "kaygı, endişe, uykusuzluk, kabuslar",
"meaningEn": "anxiety, worry, sleeplessness, nightmares" }
The agent replies in whichever language the person used — the MCP instructions tell it so. All 78 cards, both orientations, are covered; a test asserts there are no gaps. Turkish stays first-class: it is the original, and it is the market this was built for.
What was fake, and what is real now
This engine grew out of Falcı Hatun, a shipped Turkish astrology app. Porting it surfaced two features that were not doing what they claimed.
Compatibility was a hash of the partner's name.
const seed = hash(partner.name + partner.date);
emotional: 40 + (seed % 60),
communication: 40 + ((seed >> 3) % 60),
attraction: 40 + ((seed >> 6) % 60),
Change one letter of the name and "emotional compatibility" moved. Neither chart was ever consulted. synastry now computes actual cross-aspects — Venus to Mars, Moon to Saturn, the whole grid — weighted by classical significators and orb exactness. There is a test that asserts renaming a person does not move the score, and that changing their birth date does.
The transit timeline was six hardcoded events.
{ days: 7, title: 'Merkür Retrosu Başlıyor' }
{ days: 14, title: 'Boğa Burcunda Dolunay' }
Fixed offsets from today, identical for every user. Mercury was not necessarily stationing; the Moon was not necessarily full; nobody's chart was involved. sky_ahead and transits now search the ephemeris and bisect to the exact moment — and a test checks that consecutive new moons come out one synodic month apart (~29.53 days), which a hardcoded list could never satisfy.
It tells you what it doesn't know
A reading that looks authoritative and is quietly wrong is worse than no reading.
Unknown birthplace → error, not a guess. The original silently fell back to { lat: 39.0, lon: 35.0 } — the geographic centre of Turkey — for any city it didn't recognise. The ascendant and every house cusp depend on latitude and longitude, so an unrecognised birthplace produced a confident, wrong chart. Now it raises UnknownPlaceError and suggests near-matches. Explicit latitude/longitude work anywhere on Earth.
No birth time → the chart says so. Without a time, the ascendant, the midheaven and every house placement rotate a full circle every 24 hours. Falcı Hatun still returns the chart, but the caveats field states plainly which parts are unusable, and the MCP instructions tell the calling agent not to speak about them. Transits to the Ascendant are omitted entirely rather than invented.
A dimension with no planetary contacts scores null, not a number. If two people have no Moon–Venus contact, the emotional score is absent — not filled in with something plausible.
Run it
npm install
npm test # 60 assertions
npm run dev # http://localhost:8788/mcp
Deploy
vercel --prod
Your MCP endpoint is https://<your-domain>/api/mcp — that HTTPS URL is what you register with OKX.AI.
Call it
curl -X POST https://<your-domain>/api/mcp \
-H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{
"name":"natal_chart",
"arguments":{"date":"1990-05-15","time":"09:30","place":"İzmir"}}}'
{
"sun": { "name": "Boğa", "degree": 24.17, "element": "toprak" },
"moon": { "name": "Oğlak", "degree": 12.4 },
"rising": { "name": "Yengeç" },
"planets": [
{ "name": "Merkür", "sign": { "name": "Boğa" }, "house": 10, "retrograde": true },
{ "name": "Plüton", "sign": { "name": "Akrep" }, "house": 5, "retrograde": true }
],
"aspects": [
{ "from": "Güneş", "to": "Ay", "aspectTr": "üçgen", "orb": 0.16, "exactness": 0.977 }
],
"houseSystem": "equal",
"caveats": []
}
Pricing
Free.
Relationship to the Falcı Hatun app
This is a separate, standalone service. The mobile app is untouched — its Cloudflare Worker, its prompts and its reading flows are not used, not called, and not modified. What was carried over is the data layer (the 78-card Turkish deck, the zodiac table, the city coordinates) and the astrology mathematics, rebuilt for Node and corrected where it was wrong.
MIT.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
E2B
Using MCP to run code via e2b.