Alright... in that case I’d first get all the page-contents stored in an array, and produce one PDF from that. With several hundred documents, you’ll gonna use up quite some server CPU there.
But server resources aside - there are still several questions you need to ask yourself: Do you have links in those pages that point to other modx documents? If yes, you need to replace those somehow to point to a certain anchor / page in your final PDF. Otherwise you’ll end up with links pointing to the online doc. Which defeats the whole purpose of having a separate PDF.
Also: What about the main index? A good manual/documentation always features a good index of some sort. Honestly, I don’t think a 600 page manual is worth much without an index. Finding a way to sync an online TOC with a print-document / PDF might be causing some major headache...