> ## Documentation Index
> Fetch the complete documentation index at: https://context7-enes-docs7-docs.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# AI & Agents

> llms.txt, Markdown exports, page actions, and agent discovery, generated for every site

Every Docs7 site is readable by AI tools with no setup. Everything below is generated automatically; repository files with the same names always win.

## Ask AI

Hosted Docs7 sites include **Ask AI** in the navigation bar. Readers can ask questions about the current documentation without leaving the site.

One visitor IP can send up to 10 Ask AI requests per minute to one site. One deployed site can receive up to 100 Ask AI requests per minute across all readers. See [Limits](/docs7/limits#interactive-requests).

## llms.txt and llms-full.txt

* `/llms.txt`: the site name, description, links to local API specs, and an index of every page with its description.
* `/llms-full.txt`: the entire site as one Markdown document, streamed in chunks so it works at any size.
* Both are also served under `/.well-known/`, and a `Link` header on every page advertises them to crawlers.

## Raw Markdown for every page

Each page has a Markdown sibling: `/guides/deploy` is also available as `/guides/deploy.md` (the home page as `/index.md`). API pages export a structured Markdown rendering of the endpoint.

Use `<Visibility for="agents">` in MDX for content that should appear only in these exports, and `for="humans"` for web-only content.

## Agent instructions

`markdown.instructions` in `docs.json` appends an "Agent Instructions" section to every Markdown export and to `llms.txt`. Guidance like "prefer the v2 API in examples" travels with your content into any AI tool that reads it.

## Page actions

Every page shows an action menu next to the title (defaults: **Copy page**, **View as Markdown**, **Open in ChatGPT**, **Open in Claude**). Configure it site-wide or per page:

```json theme={null}
{
  "contextual": {
    "options": ["copy", "view", "chatgpt", "claude", "perplexity", "cursor"],
    "display": "header"
  }
}
```

Built-in options: `copy`, `view`, `chatgpt`, `claude`, `perplexity`, `grok`, `cursor`, `vscode`. Custom entries take `{ "title", "description", "icon", "href" }`, where `href` may use `$page` (the absolute URL of the page's Markdown export) and `$path` placeholders. Set `"options": []` to disable the menu.

## Agent discovery files

Each site also publishes:

* `/skill.md` and `/.well-known/agent-skills/…`: an agent skill describing how to use the docs.
* `/.well-known/agent-card.json`: an A2A agent card.

Ship your own `skill.md` in the repository to replace the generated one.

## Context7 indexing

A Docs7 site can be indexed into Context7 itself, so coding agents get your latest docs through the Context7 MCP. Indexing is on by default when you connect the repository (the **Add to Context7 index** checkbox); for an existing site, the same button lives in the Settings tile. Every production deployment refreshes the index automatically.

To see which AI systems actually read your docs, and what they read, check [Analytics](/docs7/analytics).