AI assistants: this page is available as Markdown at https://documentation.cluedin.net/get-started/ai-assistants.md. The index of the whole documentation, with instructions for AI assistants, is at https://documentation.cluedin.net/llms.txt.

Use this documentation with an AI assistant

how-to Steps to accomplish one task

On this page

• Prerequisites
• Point the assistant at the documentation
• Give the assistant a single page
• Load the whole documentation
• Let the assistant work in your CluedIn instance
• Verify the result
• If something goes wrong

Point an AI assistant such as Claude, ChatGPT, Microsoft Copilot or Cursor at this documentation so it answers CluedIn questions from the current pages instead of from memory, and reads them in a form that costs it far fewer tokens than the web pages.

This page is about your own AI assistant reading the documentation. It is not about CluedIn’s AI agents, the product feature that works on your golden records inside CluedIn.

Prerequisites

• An AI assistant that can fetch web pages, either built in or through a fetch tool connected over MCP (Model Context Protocol).

Point the assistant at the documentation

Give the assistant this address, and tell it to start there:

https://documentation.cluedin.net/llms.txt

llms.txt is an index written for AI assistants. It explains how the documentation is organised and which kind of page answers which kind of question, then lists every section and subsection with a one-line summary. From there the assistant fetches only the pages it needs.

For example, start a conversation with:

Use https://documentation.cluedin.net/llms.txt to answer my questions about CluedIn.
Fetch the .md version of each page you read.

Give the assistant a single page

Every page is also published as Markdown, at its address with .md added:

Page	Markdown for the assistant
https://documentation.cluedin.net/govern/enrichment/add-enricher	https://documentation.cluedin.net/govern/enrichment/add-enricher.md

The Markdown has the page’s text, tables, code and notes, without the site navigation. A section page ends with a list of the pages in it.

To copy a page into a conversation instead, select Agent mode in the top bar. The page is replaced by its Markdown, with buttons to copy it, copy its address, or open it. Select Agent mode again to return to the normal page.

Load the whole documentation

To put all the product documentation into a tool that indexes documents, such as a project’s knowledge files or a search index, use:

https://documentation.cluedin.net/llms-full.txt

It is large — the whole product documentation in one file — so give it to a tool that indexes it, rather than pasting it into a single conversation.

Let the assistant work in your CluedIn instance

The documentation tells an assistant how CluedIn works. To let it act in your own CluedIn instance, connect the assistant to the CluedIn MCP server as described in Connecting CluedIn MCP to Claude, ChatGPT and Microsoft Copilot.

For the REST API, the API reference links the OpenAPI document for each category; the complete specification is https://documentation.cluedin.net/assets/api/swagger.json.

Verify the result

Ask the assistant a question that only the current documentation answers, and ask it which pages it read. It should name .md addresses on documentation.cluedin.net. If it answers without naming any, it answered from memory: repeat the instruction to start from llms.txt.

If something goes wrong

What you see	What to do
The assistant says it cannot open web pages.	Turn on its web access, or connect a fetch tool over MCP, then try again. Without one, select Agent mode on a page and paste the copied Markdown into the conversation.
The assistant uses old terms such as entity or entity type.	Point it at the terminology page. llms.txt also lists the renamed terms.
The assistant confuses its own work with CluedIn’s AI agents.	Tell it that “AI agents” in this documentation is a CluedIn feature, described under AI agents.