astro-gen-markdown-pages

An Astro integration that generates .md companion files and an llms.txt index from your built HTML pages. The .md files are clean Markdown versions of your HTML content, suitable for consumption by LLMs or static site search tools.

Current version: 1.0.0 --npm

Installation

$ npm install astro-gen-markdown-pages

Setup

In astro.config.ts:

import genMarkdownPages from 'astro-gen-markdown-pages';

export default defineConfig({
  integrations: [
    genMarkdownPages({
      llmsTxtTitle: 'My Site Documentation',
      llmsTxtDescription: 'Complete docs for My Site.',
      trimTitleSuffix: ' | My Site',
    }),
  ],
});

Options

Option	Type	Default	Description
`llmsTxtTitle`	`string`	`'Documentation'`	Title in the generated `llms.txt`
`llmsTxtDescription`	`string`	`''`	Description in the generated `llms.txt`
`llmsTxtPath`	`string`	`'llms.txt'`	Output path for `llms.txt`
`trimTitleSuffix`	`string`	`''`	Strip this suffix from page titles in the index
`pageFilter`	`(path) => boolean`	`null`	Filter which pages get `.md` files
`indexFilter`	`(path) => boolean`	`() => true`	Filter which pages appear in `llms.txt`
`categorize`	`(path) => string`	first path segment	Group pages in `llms.txt` by category
`docsIndexUrl`	`string`	`''`	URL to add as the docs index reference in `llms.txt`

Output

After astro build:

Each HTML page gets a sibling .md file at the same path (e.g., /docs/intro.md)
A top-level llms.txt index lists all pages grouped by category

llms.txt format

# My Site Documentation

> Complete docs for My Site.

## Getting Started

- [Introduction](/docs/intro.md): The basics.
- [Installation](/docs/install.md): How to install.

## Reference

- [API Reference](/docs/api.md): Full API docs.

MarkdownOnly component

MarkdownOnly renders content that is invisible in the browser but included in the generated .md output. Use it to add LLM-facing context without cluttering the visual page.

import MarkdownOnly from 'astro-gen-markdown-pages/MarkdownOnly.astro';

<MarkdownOnly>
  This text appears only in the generated .md file, not in the browser.
</MarkdownOnly>

How it works

The integration hooks into astro:build:done. For each .html file in the output, it runs a Turndown-based HTML-to-Markdown conversion, then writes the .md file. The llms.txt is generated from the collected titles and paths.