> ## Documentation Index
> Fetch the complete documentation index at: https://docs.nemu.cc/llms.txt
> Use this file to discover all available pages before exploring further.

# Indexed sites

> Point the assistant at documentation it should search before the open web

An index is a website you have crawled so the assistant can search it directly.
It is for the documentation you rely on: your own, or a dependency's.

The assistant searches your indexes **before** it searches the web, so a
question about something you have indexed is answered from that source rather
than from whatever a search engine returns.

## Creating one

In the console, open **Indexes** and add a website.

| Field     | Notes                                          |
| --------- | ---------------------------------------------- |
| URL       | Where to start                                 |
| Model     | Which of your models writes the page summaries |
| Title     | Optional label                                 |
| Max pages | 1 to 50, 20 by default                         |

The crawler fetches the starting page and the pages it links to directly. It
does not follow links further than that, so point it at a documentation index
rather than a home page.

Each page is summarised by the model you chose, which is what makes it
searchable by meaning rather than only by keyword.

## Status

An index shows as Pending, Indexing, Indexed or Failed. Only Indexed ones are
searched.

## What it costs

Indexing runs on the model you selected, using your own provider key, so it
appears in your gateway usage like any other request. It does not draw from
credits.

Fifty pages of summaries is a small number of requests.

<Note>
  The Indexes and Prompts dialogs in the chat sidebar do not currently list
  anything. Use the Manage button in them, or go to the console directly.
</Note>