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

# llms.txt

> Configure the auto-generated /llms.txt site index served to AI agents.

# llms.txt

> Configure the auto-generated /llms.txt site index served to AI agents.

AgentXP automatically serves a `/llms.txt` file following the [llmstxt.org](https://llmstxt.org) specification. This file gives AI agents a structured, human-readable index of your site — what it does, what pages exist, and how they're organized — without the agent needing to crawl every URL.

## Configuration

```typescript middleware.ts theme={null}
import { withAgentXP } from '@reaganhsu/agentxp-next'

export default withAgentXP({
  llmsTxt: {
    siteName: 'My SaaS App',
    description: 'Project management tool for engineering teams',
    pages: [
      { path: '/', title: 'Home', section: 'Product' },
      { path: '/pricing', title: 'Pricing', section: 'Product' },
      { path: '/docs', title: 'Documentation', section: 'Documentation' },
      { path: '/blog', title: 'Blog', section: 'Resources' },
    ],
  },
})
```

## Options

### `siteName`

Type: `string`

The name of your site. AgentXP uses this as the top-level heading in the generated `/llms.txt` file.

```typescript theme={null}
llmsTxt: {
  siteName: 'Acme Corp',
}
```

### `description`

Type: `string`

A one-line description of what your site does. AgentXP renders this as a blockquote beneath the site name — it's the first thing an agent reads.

```typescript theme={null}
llmsTxt: {
  siteName: 'Acme Corp',
  description: 'B2B invoicing software for freelancers and small teams',
}
```

### `pages`

Type: `LlmsTxtPage[]`

An explicit list of pages to include in the index. If you omit this field, the `/llms.txt` file is served with just the site name and description, with no page entries. Provide a page list to give agents a structured view of your content.

Each entry in the array is an object with the following shape:

#### `pages[].path`

Type: `string` (required)

The URL path of the page, relative to your site root (e.g., `'/pricing'`).

#### `pages[].title`

Type: `string` (required)

The display title of the page as it should appear in the index.

#### `pages[].section`

Type: `string`

An optional grouping label. Pages sharing the same `section` value are rendered under a common heading. If omitted, AgentXP groups unsectioned pages under an `Other` heading when other sections are present.

#### `pages[].description`

Type: `string`

An optional one-line description of the page. When provided, it appears after the link in the generated file: `- [Title](url): description`.

Include your most important pages — pricing, core features, and documentation. These are the pages agents consult when helping users decide whether your product fits their needs or when answering how-to questions.

## Generated output

Given the configuration above, AgentXP serves the following at `/llms.txt`:

```markdown theme={null}
# My SaaS App

> Project management tool for engineering teams

## Product

- [Home](https://yoursite.com/)
- [Pricing](https://yoursite.com/pricing)

## Documentation

- [Documentation](https://yoursite.com/docs)

## Resources

- [Blog](https://yoursite.com/blog)
```

Pages with a `description` field render with a colon-separated summary:

```markdown theme={null}
- [Pricing](https://yoursite.com/pricing): Plans starting at $12/month
```

## Caching

AgentXP sets the following cache headers on `/llms.txt` responses:

```
Cache-Control: public, max-age=3600, s-maxage=86400
```

The file is cached at the edge for 24 hours and revalidated hourly by browsers. If you update your page list, redeploy your application to bust the edge cache.