קידוש החודש

Chat with this whole project

Point any AI (Claude, ChatGPT, Perplexity, Cursor…) at this site and have a real conversation with the Rambam's calculations, the class transcripts, and every concept in the system. The AI can even build you working artifacts from our live calculation engine.

The URL

One URL. Two doors. The AI picks the right one automatically.

https://shluchimexchange.ai/kh

Door 1

MCP (recommended)

For Claude Desktop, claude.ai connectors, ChatGPT developer mode, Cursor, etc. Add this as an MCP server:

https://shluchimexchange.ai/kh/mcp

The AI gets 8 tools: search, fetch, calculate, list_templates, get_template, list_source, get_source, stats.

Door 2

Web browsing

For consumer ChatGPT, Claude with web search, Perplexity, Gemini, or anything that can fetch URLs. Just tell the AI:

Browse shluchimexchange.ai/kh
and read /kh/llms.txt first — it is
a map of the site written for you.
Then answer my question.

No install, no setup. Works in every browsing-capable AI today.

What the AI can do

Answer questions about the Rambam's astronomy, the galgalim, emtzoi vs. amiti, maslul, nekudah hanichaches, moon visibility, molad timing — citing the exact KH chapter.
Run calculations for any Gregorian date and walk you through all 21 pipeline steps.
Pull from the class transcripts — Rabbi Zajac's lectures on KH 12-16 are indexed and searchable.
Build you an artifact. Ask for "a standalone page that shows me the next 30 days of moon visibility" or "a Node script that dumps the maslul table" and the AI can generate it using our live engine.
Fork and remix. The engine is served live at /kh/engine/pipeline.js so any generated artifact just imports it — no build step, no copy-paste.

Try it right now

With Claude

Open Claude Desktop or claude.ai.
In Settings → Connectors (or via Desktop config), add an MCP server with URL https://shluchimexchange.ai/kh/mcp.
Start a new chat and ask: "When is the next time the new moon is visible at sunset in Jerusalem? Use the Rambam's calculations."

With ChatGPT (browsing mode)

Open ChatGPT with web search enabled.
Paste: "Browse shluchimexchange.ai/kh and read /kh/llms.txt. Then explain the difference between emtzoi and amiti for the sun, with a specific example for today's date using /kh/api/calculate."
The model will fetch the site map, then the concept page, then call the calculate endpoint, then answer.

Build yourself an artifact

With either Claude or ChatGPT connected, try:

Build me a one-page calculator showing
the moon's phase and visibility for every
day of Nisan 5786. Use our live engine.

The AI will fetch /kh/templates/standalone-calculator.html, modify it for your request, and hand it to you as a working artifact that already imports our engine.

What it knows (content index)

Call stats via MCP or just visit /kh/mcp in a browser for the live count. Current corpus:

6 docs — calculation spec, building guide, etc.
2 class transcripts — Rabbi Zajac, via Chabad.org
21 calculation steps — every step in the Rambam pipeline, with Hebrew names, KH refs, formulas, and teaching notes
9 galgalim — celestial spheres with Rabbi Losh's pedagogical framing
9 Rambam chapters — KH 11-19, linked to Sefaria
4 concept pages — sun/moon galgalim, emtzoi vs. amiti, nekudah hanichaches
4 source categories — provenance (Rambam / Interpolated / Deduced / Losh)

Under the hood

Full site map for agents: /kh/llms.txt
MCP descriptor: /kh/mcp / /kh/.well-known/mcp
HTTP API: /kh/api/search, /kh/api/calculate, /kh/api/source
Live engine modules: /kh/engine/
Templates: /kh/templates/
Building guide: /kh/docs/BUILDING_WITH_THE_ENGINE.md
Source: github.com/rayistern/kidushhachodesh (MIT)