Free URL to Markdown Converter
Paste any public web page and get clean Markdown back — headings, lists, tables, code blocks and links intact, scripts and navigation stripped out. Useful for feeding a page to an LLM, moving content into a docs repo, or seeing what a crawler actually reads on your own site.
Why Markdown is the format LLMs prefer
Raw HTML is mostly not content. A typical marketing page is 80–95% tags, inline styles, analytics snippets and navigation, wrapped around a few hundred words that actually say something. When you paste that into a model, the majority of the context window goes to markup, and the model has to reconstruct the document's shape from it.
Markdown carries the same structure in a small fraction of the characters. A heading is ## rather than a <h2 class="…"> with six utility classes on it; a table is a table rather than forty nested cells. That is why nearly every documentation system, AI coding tool and retrieval pipeline standardises on it.
What this converter does
- Finds the real content — prefers
<main>, then a single<article>, and tells you which it used. - Removes the noise — scripts, styles, SVGs, iframes, form controls and anything marked
aria-hidden. - Resolves relative URLs — links and images become absolute, so they still work after the file moves.
- Keeps GitHub-flavoured structure — tables, strikethrough and task lists survive, and code blocks come back fenced.
- Adds YAML front matter — title, source URL, description and fetch time.
How to use this tool
- Paste a URL — any public page. Article and documentation URLs convert more cleanly than homepages, which are mostly navigation.
- Convert — the page is fetched server-side, so nothing depends on your browser or extensions.
- Check the counts — words, headings, links, images and tables. A word count far below what you can see on the page means the content is rendered in the browser, not served in the HTML.
- Copy or download — take the Markdown to the clipboard, or save it as a
.mdfile with front matter already in place.
Common issues
- Empty or nearly empty output: the page renders its content with JavaScript. This tool reads the served HTML, which is close to what a crawler sees first — so on your own site, treat it as a signal worth fixing.
- Navigation in the output: the page has no
<main>or<article>, so the whole body was converted. Deep-link to the article itself rather than a section index. - Blocked or 403: some sites refuse server-side requests, or sit behind a bot wall or a login. Only publicly reachable pages can be converted.
- Missing images: images loaded lazily via
data-srcattributes have no realsrcin the HTML, so there is nothing to link to.
Frequently asked questions
Why convert a web page to Markdown?
Markdown is the format large language models read most reliably. Feeding a model raw HTML spends most of the context window on tags, scripts and navigation that carry no meaning, and the model has to infer structure from markup noise. Markdown keeps the structure — headings, lists, tables, links — in a fraction of the tokens. It is also the native format for most documentation systems, static site generators and note apps, so a converted page drops straight into a repo or a knowledge base.
Which part of the page gets converted?
The tool looks for the page’s main content first: a <main> element, or a single <article> when there is no <main>. If it finds neither it converts the whole body. The result tells you which one it used, so when the output looks wrong you can see why. Scripts, styles, SVGs, iframes and form controls are always removed — they carry no readable text.
Are relative links rewritten?
Yes. Every relative href and image src is resolved against the page’s own URL, including after redirects, so links in the Markdown still work once the file lives somewhere else. Anchors, mailto: and other non-HTTP schemes are left exactly as the page wrote them.
What is the front matter at the top?
A short YAML block recording the page title, the URL it came from, its meta description and the time it was fetched. It is what makes a saved file traceable back to its source months later, and most static site generators and note apps read it natively. Delete it if you do not want it.
Why did I get an empty or partial result?
The tool reads the HTML the server returns; it does not run JavaScript. A page that renders its content client-side — many single-page apps and some documentation sites — ships an almost-empty shell, so there is nothing to convert. That is also what search engine crawlers and most AI crawlers see first, which makes an empty result a genuine finding about the page rather than only a limitation of this tool.
Does this help my site get cited by AI assistants?
Indirectly, and it is worth being precise. Converting someone else’s page changes nothing about your visibility. What is useful is running the tool on your own pages: if the Markdown comes back thin, badly ordered, or missing the answer you thought you published, that is roughly what an AI crawler extracts from it too. Clean structure — real headings, text in the HTML rather than assembled in the browser, tables as tables — is what makes a page quotable.
Enter your website to track its AI visibility across ChatGPT, Gemini, Claude, and Perplexity — and turn chat-bot mentions into traffic.
- Setup in 60 seconds
- Cancel anytime
Related tools
Count the words a page actually serves, without the markup.
Google Crawler SimulatorSee a page the way a crawler fetches it, before JavaScript runs.
Page Links ExtractorPull every link off a page with its anchor text.
Thin Content CheckerFind pages too light to satisfy the query they target.