Inline styles
color, background, font-size — Markdown has no equivalent. They're dropped silently. If the styling carries semantic meaning, capture it as text or HTML before converting.
HTML → clean Markdown
The converter uses Turndown with the GFM plugin, so structural HTML round-trips faithfully — code blocks keep their language, tables keep their alignment, lists keep their depth. Inline styles and visual decoration don’t carry through, by design.
- Engine: Turndown + turndown-plugin-gfm for tables, strikethrough, and task lists.
- Output flavor: ATX headings (
# ... ######), fenced code blocks (```language),-for bullets, inline link style. - What round-trips: Headings, paragraphs, lists (any depth), code blocks (with language hint), inline code, blockquotes, tables (alignment honored), images, links, bold, italic, strikethrough, task lists, horizontal rules.
- What gets dropped: Inline styles, wrapper
<div>/<section>, iframes, forms,data-*attributes, CSS classes. - LaTeX detection: Custom Turndown rules detect
.math-inline,.katex, and.katex-displayspans, converting back to$ ... $/$$ ... $$. - Privacy: Conversion runs entirely in your browser. No upload.
The flow
- 1. Switch input mode. Click HTML Input at the top of the editor (or use the keyboard shortcut). The left panel now expects HTML.
- 2. Paste. Paste your HTML — full document or fragment, either works. The right panel renders the converted Markdown live.
- 3. Copy or export. Use the Copy Markdown button to grab the source, or click MD in the export bar to download a .md file.
What round-trips cleanly
Headings
<h1>–<h6> → # to ######
Paragraphs
<p> → blank-line separated text
Lists
<ul>/<ol> → -/1., nested cleanly
Code blocks
<pre><code class='language-x'> → ```x fences with the language preserved
Inline code
<code> → backticks
Links
<a href> → [text](url) — title attribute kept where present
Images
<img src alt> → 
Tables
<table> → GFM pipe tables, alignment honored
Bold / italic
<strong>/<em> → **bold** / *italic*
Blockquotes
<blockquote> → > prefix, nested where applicable
Horizontal rules
<hr> → ---
What doesn’t survive (and why that’s usually fine)
Wrapper divs / sections
Plain <div> and <section> get unwrapped — their children survive, the container doesn't. <article> and semantic landmarks behave the same way.
iframes / embeds
<iframe>, <video>, <object> have no Markdown equivalent. The converter strips them. If you need them in the output, leave the page as HTML or use a static HTML export.
Forms
<form>, <input>, <button> get dropped. Markdown isn't an interactive format.
Custom data attributes
data-* attributes don't survive. They're metadata for JS, not content.
Class-based theming
Tailwind classes, BEM, and CSS modules are all just strings to the converter — no styling carries through. Markdown styling is structural, not visual.
When this is the right tool
CMS migration
Moving from WordPress to a markdown-based static site? Paste each post's HTML, copy the resulting Markdown, drop into your repo.
Web clipping
Save a clean, portable version of an article. Strip ads, sidebars, and theme noise — keep the prose.
Documentation cleanup
Inherited an HTML doc set? Round-trip it to Markdown for cleaner diffs in version control.
AI input prep
Many LLMs handle Markdown better than raw HTML. Pre-convert before pasting into a long-context prompt.
Pro tip: trim before you convert
The cleaner the HTML in, the cleaner the Markdown out. If you’re grabbing from a live web page, open DevTools, find the article body in the Elements panel, right-click and Copy → Copy outerHTML. That gives you just the prose, none of the navigation, ads, or theme wrappers — and the resulting Markdown will be near-publishable.
Try it on a real page
Switch to HTML Input, paste, watch the Markdown render. Re-export as DOCX or PDF if you need the document form.
Open the converter