# llms.txt

epresso emits an [llms.txt](https://llmstxt.org/) file describing the site's
pages for LLM crawlers, the same way it emits `sitemap.xml` and `robots.txt`.
It is generated by default; opt out under `[seo.llms]`:

```toml
[seo.llms]
enabled = true          # default
path = "/llms.txt"      # output path
title = ""              # defaults to the site name
description = ""        # defaults to the site description
```

The file has an H1 site title and (when set) a blockquote description, followed
by one bullet per HTML page grouped under a `## Section` heading named after the
page's top-level path segment:

```markdown
# My Site

> Docs, guides and examples.

- [Home](https://example.com/)

## Blog

- [Hello world](https://example.com/blog/hello/): Why we rebuilt our docs.

## Guides

- [Getting started](https://example.com/guides/getting-started/)
```

Each bullet's link text and description come from the page's `title` and
`description`/`summary` metadata (front matter, entry data, or `.ep` frontmatter).
Paths without metadata fall back to a title-cased URL segment. Pages already
provided by the user (e.g. `public/llms.txt`) are left untouched.