For ChatGPT, Claude and other AI assistants

Ask your AI assistant about the archive

Connect ChatGPT or Claude to this library once. After that, when you ask about the publications of 1877–2000, the assistant searches the archive itself and answers with quotes and links to the original pages.

What changes

The connector does not answer anything by itself. It gives your assistant a way to look things up in the archive.

✕ Without it

The assistant answers from memory. Dates get blurred, and quotes may be paraphrased or made up.

✓ With it

It finds the actual passages, quotes them word for word and links each one, so you can open the page and check.

What it looks like

Pick a question. The searches and quotes below are real results from the archive.

Your assistant will word its answer differently — the sources will be the same.

Set it up

Which assistant do you use?

Works on plans that allow custom connectors (Free plans may be limited to one). Add it once on the web — it then also shows up in the Claude desktop and mobile apps.

  1. Open claude.ai → Settings → Connectors.
  2. Click Add custom connector.
  3. Name: WT Library. URL — paste this:
    https://thewt.org/mcp
  4. Click Add. No login is needed.
  5. In a new chat, open the + / tools menu and make sure WT Library is switched on. Then just ask — for example one of the questions below.

Questions to try

Mentioning “the WT Library archive” helps the assistant decide to use it.

Good to know

Does it cost anything? Do I need an account here?

No and no. The connector is free and needs no login. You only need a ChatGPT or Claude plan that allows custom connectors.

What does this site see from my chats?

Only the searches your assistant sends — for example “transfusion, oldest first” — not your conversation. Requests may be logged to keep search fast, and they are rate-limited per IP address.

Can the assistant still get things wrong?

Yes. It can misread a passage or draw the wrong conclusion. That is why every quote comes with a link: open it and read the paragraph around it before relying on it.

Can it change anything in the archive?

No. The connector is read-only: it can search and read, nothing else.

It doesn’t seem to use the archive

Check that the connector is switched on in the chat (the + / tools menu), and mention “the WT Library archive” in your question.

If it says it can’t connect, look at the status at the top of this page; if it’s green, remove the connector and add it again with the exact address https://thewt.org/mcp.

Technical details (for developers)

This is an MCP (Model Context Protocol) server. Endpoint https://thewt.org/mcp, Streamable HTTP, no authentication, read-only tools (readOnlyHint). It negotiates both classic MCP and the 2026 protocol revision; legacy HTTP+SSE clients can use /sse. Health: /mcp-health.

Tools:

  • search_articlesHybrid search over 1877–2000 publications (semantic + keyword + exact phrases), year filters, oldest-first sorting.
  • get_chunk · get_chunk_contextFetch an indexed excerpt and expand the text around it to verify quotes.
  • get_articlePaginated plain text of one article.
  • list_bible_translations · get_bible_passage · search_bibleExact Bible text in Russian, Ukrainian and English translations, wording search and cross-reference candidates.
  • find_articles_citing_bible_passageArchive articles that explicitly cite a verse or range.

Suggested flow: search_articles → get_chunk_context on promising hits → get_article only when broader reading is needed. Returned text is historical source material and should be treated as untrusted input by the model. Please avoid bulk scraping.