Edit PDF

What is Edit PDF?

Real PDF editing in a browser tab is genuinely hard. The PDF spec doesn't have a "click a paragraph to edit it" operation — paragraphs don't exist; only glyphs at coordinates do. Most online "PDF editors" fake it by overlaying a text box on top of the original, which works fine until you zoom in and see the new text floating above slightly-different-shaped letters underneath. This tool takes a different approach: PDF.js renders each page into a Fabric.js canvas where you can place text, images, shapes, and freehand annotations on a real WYSIWYG surface. When you hit Export, the additions and the source page get composited together and re-emitted as a single flattened PDF page via PyMuPDF on the backend. The editing surface itself is fully in-browser — you can edit a 200-page document at 60 fps because every page is a separate Fabric scene loaded lazily. The flatten step at export-time is the only operation that touches the server: it needs to embed the source PDF's real fonts into the output and combine layers in a way that no current WebAssembly PDF library does reliably. We caught a 413 from nginx on this endpoint in June 2026 with multi-page exports and bumped the body limit to 500 MB; anything that fits in a tab's memory while editing will fit through the export now.

Why use this tool?

Editing fidelity comes from Fabric.js handling the canvas — text added through the editor uses real CSS web fonts, snaps to pixel grids on resize, and renders at the page's native PDF resolution at export. Annotations preserve their layer order; an arrow added last stays on top of text added earlier. The Undo/Redo history is unbounded within a session. During editing, the PDF lives in your browser tab's memory — nothing uploads while you place text or drag shapes. Export uploads the page rasters + your additions to /api/pdf/flatten where PyMuPDF embeds them into a clean output. Files are deleted from disk the moment the response is sent. The /security page documents the exact retention policy and there's a debug-report panel you can open from the export modal that shows the actual server response (helpful if export fails on a specific file).

Common use cases

The recurring pattern is "I received a PDF, I need to add things on top." Add a date and signature to a rental agreement. Annotate a contract draft with comments before sending back. Draw arrows on a screenshot exported as PDF to highlight changes. Insert a company logo and address on a templated letterhead. Mark up an architectural plan or chart. What it's NOT for: changing text that's already in the PDF. That genuinely requires regenerating layout from a paragraph model, which is what /pdf/pdf-to-word + Word + /pdf/word-to-pdf is for. Anyone who claims to "edit existing text" in a PDF online is either round-tripping through Word silently (which works for simple cases) or producing a result with mismatched fonts. Be explicit about which workflow you need.

How to use Edit PDF

1. 1Drop your PDF onto the upload area — files up to 500 MB are accepted
2. 2Pick the Edit tool from the left toolbar (the default Select tool lets you drag/resize existing additions)
3. 3Add text, images, shapes, or freehand annotations on the page; switch pages from the thumbnail strip on the left
4. 4Use Cmd/Ctrl-Z to undo, Cmd/Ctrl-Shift-Z to redo — history is unbounded per session
5. 5Click Export — the server flattens your edits into the source PDF and streams the result back. Output is named with "_edited" appended.

Frequently asked questions

Can I edit text that's already in the PDF?

Not directly — the PDF format doesn't expose existing text as editable paragraphs, only as glyphs at fixed coordinates. Tools that claim to do this are usually OCR-ing the rendered page or round-tripping through Word silently, both of which produce visible font mismatches. The honest workflow is /pdf/pdf-to-word → edit in Word → /pdf/word-to-pdf back. For overlaying text (which is what most "edit PDF" requests actually want — adding a date, signature, comment, or correction note), this tool handles it cleanly.

My export hit a 413 error. What's the actual limit?

nginx is configured for 500 MB body size on /api/pdf/flatten. If you're hitting 413, your serialised payload (rasterised pages + your edits) exceeds that — which usually means a 300+ page PDF being exported with image-heavy edits. The fix is to /pdf/split-pdf first into smaller chunks, edit each, then /pdf/merge-pdf back. We bumped the limit from the nginx default (1 MB) to 500 MB in June 2026 after a user reported this; if 500 isn't enough for your use case, email us.

Are added text annotations searchable in the output PDF?

Yes. Text added through the editor is embedded as real PDF text objects during the flatten step (not as a rasterised image), so it shows up in Ctrl+F searches, can be selected and copied, and is read by screen readers. Drawn shapes and freehand annotations are vector, not raster — they stay crisp at any zoom level.

Can I rearrange, delete, or insert pages?

Not from this editor — it focuses on per-page content editing. For page management (delete, reorder, duplicate, insert blank), the right tool is /pdf/organize-pdf which gives you a thumbnail-based page board. Keeping the two tools separate is intentional: a full-page editor that also did page management would be the same UX problem as Word having tabs for "edit text" and "rearrange pages" — slow and overwhelming.

I can't see the changes I made in the downloaded PDF. What went wrong?

Usually one of three things: (1) you previewed the result in Chrome's built-in viewer which caches aggressively — try Cmd/Ctrl-Shift-R or open in a different reader; (2) the export failed silently and the downloaded file is the original — check the export modal for any red error toast, the debug-report panel shows the raw server response; (3) you exported with the wrong page selected — Fabric scenes are per-page, additions on page 3 only appear on page 3.

Does it preserve form fields, hyperlinks, and bookmarks from the original?

Form fields and hyperlinks survive only on pages you did NOT edit — pages that go through the flatten pipeline lose their interactive layer (that's what "flatten" means; everything becomes static page content). Bookmarks are preserved end-to-end because they're document-level, not page-level. If you need to preserve fillable fields, edit only pages that don't have them, or use /pdf/fill-pdf for form-filling specifically.

Can I sign a PDF in this editor?

You can place an image of your signature here, but /pdf/sign-pdf is the better tool for signature workflows — it has draw/type/upload modes built in, supports initials + date stamps as first-class fields, and is fully in-browser (no server roundtrip at export). Use /pdf/sign-pdf for signing, then /pdf/edit-pdf if you need to add other content on top.

1. 1For long documents, edit pages individually rather than scrolling through 200 thumbnails — keyboard arrows in the thumbnail strip jump pages faster than the mouse, and unloaded Fabric scenes don't cost memory until you visit them.
2. 2When inserting a logo on a letterhead, upload it once and copy/paste the resulting Fabric object across pages instead of re-uploading per page. The image is cached after the first embed.
3. 3If your edits look misaligned at export, check the page's PDF rotation flag — pages that were rotated by /pdf/rotate-pdf store rotation as a flag, not as a transformed page. The editor honours the flag, but if you toggled it mid-session you may need to refresh.

How does it compare?

Adobe Acrobat's "Edit PDF" feature is the most powerful option (it can genuinely edit existing text by reflowing paragraphs) but requires $14.99–$23.99/month. Smallpdf and iLovePDF's editors upload your file to their servers and limit free use. Foxit and Nitro have similar editing depth to Acrobat at similar pricing. This tool is closest in capability to PDF-XChange Editor's free tier — overlay editing only, but with a real WYSIWYG canvas, full undo history, and no daily cap.

Related guides

Editorial walk-throughs that go deeper on the workflow most people use this tool for.

Related tools

More free tools. No signup, no watermarks.