Drop in a PDF and mark it up: type text anywhere, highlight passages, draw boxes and lines, or scribble freehand. Pick a colour, work page by page, and undo anything you change your mind about. Highlights are translucent, so the text underneath stays readable rather than being covered by a solid block — which is what a highlight is for. Everything is drawn into the page content when you export, so it renders identically in every PDF viewer rather than depending on one that understands comment layers. The flip side is that it becomes part of the page and cannot be edited again afterwards, so keep your original. The document is never uploaded; the whole thing runs in your browser.
| Tool | How to use it | Drawn as |
|---|---|---|
| Text | Click where it should start | Helvetica at your size |
| Highlight | Drag over a passage | Translucent fill |
| Box | Drag around something | Outline only |
| Line | Drag from end to end | Straight stroke |
| Draw | Hold and move | Freehand stroke |
No. It is rendered, annotated and exported entirely in your browser. Documents people mark up are usually drafts and contracts, which is exactly the category not to hand to someone else's server.
No. This adds things on top rather than changing what is already there. Editing existing text means rewriting the document's content streams and re-flowing everything around the change, which is why the tools that do it properly are expensive desktop applications. For covering something up, draw a filled box over it — but note that only hides it visually; the original text is still in the file and can be extracted.
No. They are drawn into the page content, so the exported file has no separate layer to reopen. That is deliberate — it guarantees the marks appear in every viewer, including ones that ignore PDF comment objects. Keep your original if you might need to revise.
Because a solid rectangle over the text would hide what you are highlighting. It is drawn at about a third opacity, the same way a physical highlighter works.
Yes. Freehand drawing uses pointer events, so a finger or stylus works the same as a mouse.
You can type text onto it, which is enough for a printed form that has no real fields. A PDF with proper interactive form fields is better filled in your PDF viewer, which will store the values as form data rather than as drawn text.
Every annotation is stored as a percentage of the page box, so the same edit lands correctly whether the preview is 400 pixels wide on a phone or 900 on a desktop — storing pixels would tie the result to whatever size the page happened to be rendered at. Converting those percentages into the exported PDF is where the reliable mistakes live. Screen coordinates grow downward from the top left and PDF user space grows upward from the bottom left, so every vertical value has to be inverted; and pdf-lib anchors a rectangle at its lower-left corner, so the box height must come off the top edge as well. Miss that second part and every shape sits exactly one box-height too high, which looks close enough to correct in a thumbnail to survive a casual check. Text is a further special case, because drawText positions the baseline rather than the top of the glyphs, so the point is shifted down by the font size to put the text where it was clicked. A drag is normalised into a positive-sized box before anything else, so dragging up and to the left produces a shape rather than a negative width that renders as nothing. Highlights are drawn translucent and boxes as outlines, because an opaque rectangle over the text it is meant to mark hides the thing being highlighted. Everything is drawn into the page content rather than added as a PDF annotation object, which means it renders identically in every viewer and cannot be toggled off — and equally cannot be edited again later.