Give a long PDF the clickable bookmark sidebar it should have had, plus an optional contents page whose lines jump straight to the section. Headings are found for you, then you edit and nest them. The PDF never leaves your browser.
Your file is read and rewritten inside this tab, so proposals, SOPs and manuals stay on your machine. Your file itself is never sent anywhere. The page does make two other requests of its own: it loads two open source PDF libraries from a public CDN the first time you visit, and Castwright counts anonymous page visits and tool runs, which records the page count of the file you opened and how many bookmarks you exported, never the file or its text.
Detection compares the size of every line of text against the size the body text is set in. It works on PDFs that carry a real text layer. A scan or an image only PDF has nothing to measure, so use the buttons above to add bookmarks by page or by hand.
Top of the page needs no coordinates at all, so it is right on every document, including rotated pages and unusual page boxes. The heading line option scrolls further down the page and keeps the reader's current zoom. It only applies to bookmarks that carry a position, which means the ones found by detection and the ones imported from the file. Bookmarks you add by hand or with One per page always land at the top of their page.
The finished file also asks the reader's PDF app to open the bookmark panel, otherwise most people never notice the bookmarks are there.
A PDF bookmark is not a highlight and it is not a comment. It is an entry in the document outline, the panel that opens down the left side of Acrobat, Preview, Chrome, Edge and almost every reader on a phone. Click one and the file jumps to that section. A ninety page proposal without an outline is a scroll bar and a guess; the same file with an outline is a document someone can actually use. This page writes that outline into your file for real, as a proper /Outlines tree with a landing point per entry, so it survives being emailed, printed to PDF viewers, and opened years later.
Let it find the headings. Every line of text is measured against the size your body copy is set in. Lines that are noticeably larger are offered as bookmarks, and the different heading sizes become levels 1, 2 and 3 so the sidebar comes out nested instead of flat. Running headers and footers that repeat on every page are filtered out. Slide the sensitivity control if you get too many or too few.
Import what is already there. If the PDF came out of Word, InDesign or a merge tool it may already carry an outline. Import it, then rename, renest, delete and add, which is the editing part that the paid apps charge a subscription for. Importing is not optional housekeeping, it is the only way to keep what is already in the file: the copy you download carries the list on this page and nothing else, so an outline you never imported is replaced rather than added to.
Add them by hand. Type a page number and a title. This is the answer for scanned contracts and image only PDFs, where there is no text layer to measure. There is also a one bookmark per page button for reference material that gets flicked through rather than read.
Turn it on and a real page is generated and inserted at the front or the back, with your headings, dot leaders and page numbers. Every line is a live internal link, so the contents page works when the file is printed to screen and when it is opened on a phone. Page numbers can be shown as the position in the finished file, which is what someone counting pages will see, or as the original numbering if the document already prints its own folios.
It cannot add to an outline that is already in the file. The copy you download gets the list shown on this page as its complete outline, so if the PDF arrived with bookmarks of its own, use Import existing to bring them into the list first, otherwise the download replaces them. Detection needs a text layer. A scan is a picture of words, so nothing can be measured and the automatic list comes back empty; that is a limit of the file, not a bug, and the manual and per page routes exist for exactly that case. Password protected and edit locked files are refused at the door, because the library that rewrites the file cannot decrypt one, and a copy saved from it would not open at all; remove the protection first and then come back. If the PDF carries a digital signature, rewriting the file breaks that signature, which is true of any tool that edits a signed PDF. Right to left and vertical scripts are read approximately, so check the titles before you export. Very long documents are scanned to a page limit to keep the tab responsive, and you are told when that happens.
Editing a PDF outline is a paid feature almost everywhere. Acrobat Pro is about twenty four dollars a month and PDF Expert Pro is about eighty dollars a year, and both of them put bookmark editing behind the paywall. Free web converters will happily merge and compress a PDF but they nearly all skip the outline, which is why this is one of the few PDF jobs with almost no free option. Here it is free, there is no signup, and the file itself is never uploaded anywhere.