Markdown to HTML with the settings shown

Only if you want it to look like anything in particular — the markup is semantic and carries no class names, so it inherits whatever styles surround it. That is the right default for a template or a CMS field, where your own CSS already covers h2, p and pre. Tick “Inline the stylesheet” when there is no surrounding CSS to inherit: it prepends a style block of about 2.5 KB scoped to a wrapper class, which is enough for headings, code blocks, tables and blockquotes to read properly on their own.

Because the default output is a fragment, which is what you paste into an existing page. Switch Output to “Whole HTML file” and you get a doctype, a charset meta, a viewport meta and a title taken from the first heading in the document — a file that opens correctly by double-clicking it. The two are genuinely different artefacts, which is why they are a choice rather than a guess.

GitHub renders footnotes and this converter reproduces them, but they are not part of the published GFM specification — GitHub added them in 2021, four years after the spec was written, so a strict GFM implementation is free not to have them. Here a [^1] reference becomes a superscript link to an ordered list at the end of the document, with a return arrow back to the reference, and switching to CommonMark turns the marker back into the four literal characters. If the HTML is going somewhere other than GitHub, footnotes are the construct most likely to be missing at the other end.

No — inline HTML in the source comes out escaped, so a <div> in your Markdown becomes &lt;div&gt; in the output and shows as five characters on the page. That is a deliberate trade: this converter is safe to point at Markdown you did not write, and the way it guarantees that is by never emitting a tag it did not generate itself. If your document depends on raw blocks, convert everything else here and paste those blocks back in by hand, or keep the file on a host that filters HTML rather than dropping it.

Turn them off when the destination generates its own, which most static-site generators and CMS platforms do — two id attributes on the same heading is invalid HTML and the anchors will fight. Leave them on when the HTML is the final artefact and you want #getting-started to work. The ids follow GitHub's rule: lowercase, punctuation dropped, spaces to hyphens, non-ASCII letters kept, and a numeric suffix on a repeat.

No, it is HTML5: void elements are written <br>, <hr> and <img …>, without the trailing slash. XHTML has not been the serialisation anyone parses since 2014 and the slash is ignored by every HTML parser, but if you are pasting into a JSX file or an XML pipeline that does care, a find-and-replace on <br> and <hr> covers the two that appear in Markdown output.

It is dropped and the link degrades to its own text — no anchor, no href. The same applies to vbscript: and to data: URLs that carry HTML rather than an image, because those three are how a Markdown link turns into script execution on the page that renders it. Ordinary schemes pass through untouched: http, https, mailto, tel, ftp, relative paths and #fragments, plus base64 data:image payloads on images, which READMEs use for inline badges.