A free project of TorahApp.org

TorahLM

Grounded Torah, Talmud & halacha for AI.

❦

A curated library of primary Jewish texts, Halachipedia and shiurim, served as clean Markdown with every text addressable by a simple URL.

Why it matters: an AI makes up far fewer facts when it can read the real sources instead of answering from memory. Giving an assistant a large, primary-source library to quote from is what the AI world calls grounding, and that's what TorahLM provides.

AI still makes mistakes. Even grounded in primary sources, an assistant can be wrong, so treat it as a way to find and open sources, never as a posek for practical or halachic rulings. Always verify against the original and ask a rabbi.

Watch the demo

TorahLM.org Demo: Grounded Torah, Talmud & Halacha for AI

Ready to use · nothing to set up

AI notebooks, already grounded in the sources

Open a NotebookLM pre-loaded with a major curated selection of the corpus: Tanakh, Talmud, practical halacha, and more. Ask in plain language, and every answer comes back with citations you can open and check yourself.

A NotebookLM is Google's free grounded-AI notebook (recently renamed Gemini Notebook). It answers only from the sources loaded into it, not the open web.

Open the core NotebookLM →

Or build your own, picking exactly the books and collections you want.

In your own assistant · free, no login

Bring TorahLM into your AI

Connect one server and your assistant can read any text, search the whole corpus, pull up related commentaries, and look up Halachipedia articles and shiurim. It puts the Torah library, the practical halacha, and hundreds of thousands of classes in one Orthodox-curated place it can quote and link.

https://torahlm.org/mcp

Add it as a custom MCP connector. If a client asks for authentication, choose None / Open. Works in ChatGPT, Claude, Grok, Perplexity, and most agent tools (consumer Gemini not yet). Expand your assistant for exact steps:

ChatGPT
  1. Settings → Connectors → Advanced, then turn on Developer mode.
  2. Settings → Connectors → Create, then name it “TorahLM”, URL https://torahlm.org/mcp, no authentication, then create. ChatGPT lists the tools it finds; enable it per-chat from the + menu.

OpenAI: Developer mode & MCP →

Claude
  1. Web / mobile: Settings → Connectors → Add custom connector, then name “TorahLM”, URL https://torahlm.org/mcp, then Add. (Team/Enterprise: an Owner adds it under Organization settings → Connectors.)
  2. Claude Code (CLI): claude mcp add --transport http torahlm https://torahlm.org/mcp

Claude: Custom connectors (remote MCP) →

Gemini

The consumer Gemini app (including Gems) can't add MCP servers or follow links from custom instructions yet. For a grounded Google-side assistant, use the ready-made NotebookLM above.

Developers: use Antigravity (below), Google's agent-first IDE, which does support MCP servers.

Grok
  1. Settings → Connectors → add a custom connector with URL https://torahlm.org/mcp, authentication None. (paid plan)

Open Grok →

Perplexity
  1. Settings → Connectors → Add custom connector, then set the MCP server URL to https://torahlm.org/mcp, authentication Open / none. (paid plan)

Perplexity: custom MCP connectors →

Codex
  1. Codex CLI: codex mcp add torahlm --url https://torahlm.org/mcp (no token needed — the server is open).
  2. Start a session and run /mcp to confirm it connected and see its tools.

The same entry is picked up by the Codex VS Code extension and the Codex app, since they share ~/.codex/config.toml.

Codex: Model Context Protocol →

Antigravity
  1. Open the MCP manager: … in the agent side panel → MCP Servers → Manage MCP Servers → View raw config. In Antigravity CLI, type /mcp to open the interactive MCP manager.
  2. Edit the config — global ~/.gemini/config/mcp_config.json, or per-project .agents/mcp_config.json: 
    { "mcpServers": { "torahlm": { "serverUrl": "https://torahlm.org/mcp" } } }
  3. Save the file. Use the MCP manager to reload the configuration and check the connection status.

Antigravity: MCP →

OpenClaw
  1. openclaw mcp set torahlm '{"url":"https://torahlm.org/mcp","transport":"streamable-http"}'
  2. Check it: openclaw mcp doctor torahlm --probe

This writes an mcp.servers entry to ~/.openclaw/openclaw.json; restart the gateway if it was already running. OpenClaw can also just fetch SKILL.md on its own — see the developer section below.

OpenClaw: MCP →

Hermes
  1. hermes mcp add torahlm --url https://torahlm.org/mcp
  2. Test it: hermes mcp test torahlm

Equivalent entry in ~/.hermes/config.yaml:

mcp_servers:
  torahlm:
    url: "https://torahlm.org/mcp"

Hermes Agent: MCP →

Developers: API, SKILL.md & downloads

Tools that can fetch the web on their own (coding agents like Claude Code or Cursor, and autonomous agents like OpenClaw) don't strictly need the connector: the whole API is self-documenting at torahlm.org/SKILL.md, written to be read by an AI. Install it as a skill, or add a standing instruction:

When I ask about anything Jewish, Torah, or halachic, first read https://torahlm.org/SKILL.md and use the TorahLM.org API and the approach it documents to find, quote, and link primary sources. Cite what you use; never invent citations.

This route only works in tools that can actually open URLs. Chat apps like ChatGPT and Gemini won't fetch links from custom instructions; use the MCP connector above there.

Build directly on the corpus

  • Text API: read, search, related links, dictionary, Halachipedia, and shiurim over simple URLs, at torahlm.org/SKILL.md
  • MCP server: the same corpus as an agent tool server, torahlm.org/mcp
  • Download: per-book Markdown + combined NotebookLM sources on GitHub, at torahlm.org/downloads
  • Custom notebook builder: assemble a source list for your own NotebookLM, at torahlm.org/builder

Or explore the library yourself

Clean, addressable pages you (or an AI) can open, quote, and link. Start with the two companion collections, or jump into a text category below:

Halachipedia Practical halacha articles, sourced back to the codes שיעורים Shiurim Hundreds of thousands of classes from YUTorah, OU and more
Sefaria text library
תנ״ך Tanakh Torah, Nevi'im and Ketuvim, with the classic commentaries משנה Mishnah All six orders, with Bartenura and Kehati תלמוד Talmud Bavli and Yerushalmi, with Rashi and Tosafot מדרש Midrash Aggadic and halachic midrashim הלכה Halacha Shulchan Arukh, Mishneh Torah, Mishnah Berurah and more מוסר Musar Mesilat Yesharim, Duties of the Heart and other works

Browse the full library →

A plain keyword search over the library — texts, Halachipedia articles and shiurim titles. It finds where something is said, not what it means: type a word, a phrase, or a book name and you get back matching sources to open.

It is not a chatbot and there is no AI here — no questions answered, no summaries, no opinions. It's the same lookup the MCP connector and API hand to your assistant, which is what lets the assistant find real sources instead of recalling them from memory.

About the content: sources, curation & licensing

Orthodox focus: the texts & content are curated for an Orthodox audience, under rabbinic guidance.

  • Texts: an Orthodox (and, for technical reasons, partial) subset of the Sefaria library, selected under the halachic guidance of Rabbi Gil Student. TorahApp carries a haskama (rabbinic approbation) from the Rabbinical Alliance of America (RAA).
  • Halachipedia: practical halacha articles from Halachipedia.com.
  • Shiurim: episode listings from YUTorah, OUTorah/AllTorah, TorahAnytime, Chabad.org, Torahcasts, and other podcast feeds.
  • Dicta Library: books from the Dicta Library, in their own notebook and repo.
  • License & usage: see the download repo's LICENSE and the upstream projects for text licensing details.

Please consider contributing directly to those creating the content. For us, just share TorahLM.org & TorahApp.org with friends and family.