# Use the docs with an AI assistant

> Copy a page as Markdown, give an assistant a link to read, or connect it to the documentation server. None of what you wrote in the app is involved.

Every page of these docs is available as plain Markdown, so an assistant can
read it the way you do. There are three ways to hand it one.

## Copy a page

Under each title, **Copy page** puts that page's Markdown on your clipboard, and
**View as Markdown** opens the same text at its own address. **Ask an
assistant** opens Claude or ChatGPT with a message that names the page's
address and nothing else.

## Read it from a link

A page's Markdown lives at its address with `.md` on the end, so
`/en/docs/format/overview` is also `/en/docs/format/overview.md`. The docs home
is `/en/docs/index.md`. A client that sends `Accept: text/markdown` to the
normal address gets the Markdown from it.

For the whole documentation at once there are three files:

- `/llms.txt` lists the English pages, with a link to the Markdown of each, and
  points to the other languages.
- `/en/llms.txt` and `/id/llms.txt` list one language each.
- `/en/llms-full.txt` and `/id/llms-full.txt` hold every page of that language
  in a single file.

## Connect an assistant

Assistants that support the Model Context Protocol can read the docs directly
from `/api/mcp`. It has three tools: `list_pages`, `get_page` and `search`. Each
takes an optional `lang` (`en` or `id`, English by default). In Claude Code:

```bash
claude mcp add --transport http rejarsume-docs https://shopsalone.com/api/mcp
```

The command carries the address of the site you are reading, so it works as
it is on a copy you host yourself.

## What is shared

Only the documentation, which is public. Your resumes stay in your browser, and
nothing in this section can reach them: the server has no copy to give.