Turn HTML into a PDF without uploading your markup

People render HTML to PDF to produce documents from a template they control: invoices, statements, certificates, delivery notes, reports. The markup is filled with real customer names and real amounts by the time it is worth converting, which makes pasting it into a hosted renderer a transfer of exactly the records you were being careful with.

This is the one page in this cluster with a real caveat, and it should be said plainly. Your markup is not uploaded. But a document that references an image, a stylesheet or a font by URL causes those to be fetched while it renders, and a fetch tells the host serving them that someone loaded your document. If the requirement is a document produced with no outbound requests at all, inline the images as data URIs and use locally available fonts.

Screen layout and page layout are different problems. A page has fixed edges and hard breaks, so long tables split, floated elements land in unexpected places, and backgrounds behave differently from how they look on screen. Produce the PDF and read it as a document, then adjust the markup rather than the expectations.

The network panel is unusually useful here, because you are not just confirming an absence. Render, look at the requests, and check that every one of them corresponds to a URL you can point to in your own markup. Anything you cannot account for is something to remove from the template.