Turns an HTML file into clean Markdown — headings, lists, tables and links survive; scripts, CSS, navigation and tracking markup do not. Open the converter. The file is processed in your browser.
Save an article from a browser and the file you get is mostly furniture: a navigation bar, a cookie
notice, a newsletter box, three analytics snippets, a stylesheet, a footer full of links to other articles.
The piece you wanted is somewhere in the middle. Converting to Markdown throws away everything that is not
content, because Markdown has no syntax for any of it — there is no way to express a <script>
or a CSS class, so they simply do not survive.
That is the real reason this conversion is useful before feeding something to a language model. You are
not reformatting the article so much as removing the several thousand tokens of markup wrapped around it.
What survives, and what does not
Kept
Dropped
Headings, in their levels
<script> and <style>
Paragraphs, lists, block quotes
Classes, ids, inline styles
Tables, as Markdown tables
Navigation, headers, footers
Links, with their targets
Images (the tag goes, the file was never here)
Bold, italic, inline and block code
Forms, iframes, embeds
Links keep their targets, which is worth checking before you paste the result somewhere: a page full of
relative links converts to a page full of relative links, and they will not resolve anywhere else.
What the structure costs
What the page holds
Plain text
Markdown
Difference
Prose only
53 tokens
52 tokens
1.9% cheaper
Headings and lists
57 tokens
60 tokens
+5.3%
A table
35 tokens
59 tokens
+68.6%
All three together
146 tokens
172 tokens
+17.8%
Those are token counts from our own converter, on a PDF we built for the purpose, measured with the same tokeniser the GPT models use. The pattern is the useful part, and it is not the one usually quoted:
On prose, Markdown is free. It came out 1.9% cheaper than the plain-text dump of the same page, because joining wrapped lines back into paragraphs saves about as much as the heading marks cost.
On headings and lists, near enough free — +5.3% for structure a model can navigate.
On tables it is expensive — +68.6%, all of it pipes and the separator row. And it is the one case where you have no choice, because plain text loses which number sits under which heading.
Where it is the wrong tool
It converts a FILE, not a URL. Fetching a page for you would mean a request from our
server to that site, which would break the promise that nothing here leaves your browser. Save the page
first, then convert it.
A single-page app saved from the browser may be nearly empty — if the content is
drawn by JavaScript after load, the saved HTML holds the shell and not the article. Use your browser's
Reader view or “Save as… Web page, complete” and check the file has words in it.
Images are referenced, never included.
Frequently Asked Questions
Open the converter, choose HTML as the input and Markdown as the output, and drop your .html file in. It converts in your browser and downloads as .md. Headings, lists, tables and links are kept; scripts, styling and navigation are dropped.
No, and that is deliberate. Fetching a page for you would mean a request from our server to that site, which would break the promise that nothing here leaves your browser. Save the page from your browser first, then convert the file.
Because most of a saved web page is furniture — navigation, cookie notices, stylesheets, analytics, footers. Markdown has no syntax for any of it, so it all disappears, and what is left is the article. You are mostly removing markup rather than reformatting prose.
No. Markdown references images by path rather than containing them, and a single .md download has nowhere to put the files. The image tags are dropped rather than turned into links that lead nowhere.
It was probably a single-page app whose content is drawn by JavaScript after the page loads, so the saved HTML holds only the shell. Try your browser's Reader view, or "Save as… Web page, complete", and check the saved file actually contains the words before converting.