HTML to Markdown converter

How to convert HTML to Markdown

1. Paste the HTML into the box, or select Or open an .html file to load a saved page (up to 5 MB). A fragment such as one article body works, and so does a whole page with its <head>.
2. Select Convert to Markdown. The summary counts the headings, links and tables found, so you can see at a glance if the structure came through.
3. Copy or download. Download gives you a .md file. For an opened file, it keeps the file name.

What each tag becomes

The last row matters for help centers. FAQ pages often hide each answer in a collapsed <details> block. Some converters drop the hidden text. This one keeps it, under a heading made from the question.

Example

This HTML from a help page:

<nav><a href="/">Home</a></nav>
<main>
  <h1>Shipping</h1>
  <p>We ship in <strong>2 days</strong>. See <a href="/rates">rates</a>.</p>
  <table>
    <tr><th>Zone</th><th>Price</th></tr>
    <tr><td>EU</td><td>$5</td></tr>
  </table>
</main>
<footer>© Acme</footer>

becomes this Markdown. The menu and the footer are gone:

# Shipping

We ship in **2 days**. See [rates](/rates).

| Zone | Price |
|---|---|
| EU | $5 |

What the converter leaves out

When you paste a whole page, most of the HTML is not content. The converter removes the parts that repeat on every page or do nothing in a text file:

• <script>, <style>, <noscript>, <template> and inline SVG.
• <nav>, <header>, <footer>, <aside> and elements with the navigation, search, banner or contentinfo role.
• Forms, buttons, inputs and iframes.
• Blocks whose class or label says cookie, consent, promo, newsletter or breadcrumb.

If the page has a <main> element, only that part is converted, and if not, the first <article>. When the content has no H1, the page <title> becomes the top heading. These are the rules Paperlink AI uses when it imports pages from a website, so the result also shows you what a support chatbot would read from that page.

One case needs care: some sites put the real article inside a <header> or an <aside>. If text you expected is missing, paste only the part you want.

Links, images and code

Links keep their address exactly as written, also relative ones such as /rates. If you will use the Markdown on another site, replace them with full addresses. Links to javascript: are not real destinations, so only their text is kept. Images with an address become Markdown images. Images stored inside the page as data (a data: address) would put thousands of characters into the file, so only their alt text is kept.

Code in <pre> keeps its line breaks and spaces. When the inner <code> has a class such as language-python or lang-sh, the fence gets that language, which most Markdown viewers use for syntax colors.

Limits

• Pages built with JavaScript. The page source of a single-page app is often an empty shell. Open the page, then copy the HTML of the content from the Elements panel of your browser developer tools.
• Layout tables. Old sites use tables for layout. Those come out as pipe tables too. Delete them from the Markdown, or paste only the cell you need.
• Merged cells. colspan and rowspan are not kept, because Markdown tables have no merged cells.
• Inline styles and colors are not part of Markdown and are dropped.

Why Markdown for AI and docs tools

HTML pages are mostly markup. Once the menus, scripts and class names are counted, the HTML of a short help article is often many times longer than its text. An AI model or a search index that reads the raw HTML spends most of its space on that markup, and the menu text gets mixed into every answer. Markdown keeps the headings, lists and tables in a few characters and leaves the rest out.

Use the Markdown to move pages into a docs repository, to give an AI tool clean context, or to check what a crawler sees on your site. To list your most useful pages for AI assistants, use the llms.txt generator. To turn Markdown back into HTML, use Markdown to HTML. If you want to answer customers from your help center, read how to turn a help center into an AI chatbot.