astro-llms-txt

Astro integration that generates llms.txt and llms-full.txt at build time by reading your rendered HTML output.

• llms.txt - a lightweight index of all pages with links and descriptions
• llms-full.txt - the full text content of every page

Install

npx astro add @alexcarol/astro-llms-txt

Note: astro add will auto-generate the import name alexcarolllmsTxt from the scoped package name. You may want to rename it to llmsTxt for readability.

Or install manually:

npm install @alexcarol/astro-llms-txt

Quick start

If you used astro add, your config is already set up. Otherwise, add the integration manually:

// astro.config.mjs
import { defineConfig } from 'astro/config';
import llmsTxt from '@alexcarol/astro-llms-txt';

export default defineConfig({
  site: 'https://example.com', // required
  integrations: [llmsTxt()],
});

That's it. With no options the integration will:

1. Use the homepage <h1> as the # heading (falls back to hostname)
2. Use <h1> text as page titles (falls back to <title> if no <h1> is present)
3. Exclude /404 pages
4. Extract content from <main>, descriptions from <meta name="description">

Options

All options are optional.

| Option | Type | Default | Description | |---|---|---|---| | name | string | homepage <h1>, then hostname | Site or brand name used as the # heading in generated files. | | titleSource | 'h1' | 'title' | 'h1' | Which HTML element to read page titles from. 'h1' uses the first <h1> (falls back to <title>). 'title' uses the raw <title> tag text. | | excludedPaths | string[] | ['404'] | Path segments to exclude. Matched without leading/trailing slashes (e.g. 'admin' excludes /admin/). |

Examples

// Zero-config
llmsTxt()

// Explicit name (recommended for personal/brand sites)
llmsTxt({ name: 'Alex Carol' })

// Use raw <title> tags instead of <h1>
llmsTxt({ name: 'My Site', titleSource: 'title' })

// Custom exclusions
llmsTxt({ name: 'My Site', excludedPaths: ['404', 'admin', 'drafts'] })

Title source

The titleSource option controls where page titles come from:

• 'h1' (default) - reads the first <h1> on the page. This avoids the common "Page Title - Site Name" suffix pattern in <title> tags, producing cleaner output. Falls back to <title> if no <h1> is found.
• 'title' - reads the raw <title> tag, including any site name suffix.

Most static sites use an <h1> that matches the clean page name, so the default works without configuration.

How it works

The integration hooks into two Astro lifecycle events:

1. astro:config:done - validates that site is set in your Astro config (throws if missing).

2. astro:build:done - reads each built index.html in parallel, extracts page data, infers the site name from the homepage <h1> if not provided, then writes both output files.

Pages are sorted with the homepage first, then alphabetically by path.

Output format

Follows the llms.txt specification:

llms.txt

# Site Name

> Homepage description from meta tag

## Pages

- [Home](https://example.com/): Homepage description
- [About](https://example.com/about/): About page description

llms-full.txt

# Site Name

> Homepage description from meta tag

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

Full text content of the homepage...

## [About](https://example.com/about/)

Full text content of the about page...

Requirements

• site must be set in your Astro config (the integration throws a clear error if missing)
• Pages should use <main> for primary content
• Pages should have <meta name="description"> for summaries
• For the default titleSource: 'h1', pages should have an <h1> element

License

MIT