Skip to content
Home

No upload

Your PDFs never leave your device

0 bytes of your files leave this tab. Every tool runs in the browser window you’re looking at, and there is no server that could receive a PDF. The counter below watches every request this page makes — check it against your own Network tab.

Updated September 18, 2026

Bytes sent anywhere: 0 B

What happens when you add a PDF

  1. The file is read from your disk into this tab’s memory, the same way a PDF reader opens it. It isn’t copied anywhere.
  2. The work runs on your processor. pdf-lib rewrites the file’s structure, pdf.js renders and reads its pages, and qpdf and Tesseract run as WebAssembly in a background worker.
  3. The result is handed to your browser as a download. It is created in memory on your device and saved like any other download.

The meter above counts the bytes of your files that this page sends anywhere. It watches every network request the page and its workers make, so if a file ever did go out, it would show.

Check it yourself in under a minute

1. Watch the Network tab

  1. Open developer tools: F12 or Ctrl Shift I on Windows and Linux, ⌘ ⌥ I on a Mac. In Safari, turn on “Show features for web developers” in Settings → Advanced first.
  2. Choose the Network tab and clear it.
  3. Add a large PDF to any tool and run it.

You’ll see no request anywhere near the size of your file, and no POST or PUT requests at all. The only traffic is our own scripts and, for some tools, a one-time engine download.

2. Turn off your internet

Load a tool, then switch on airplane mode or unplug the network. Add a PDF and run it. It still works, because nothing needs a server. See using FixMyPDFs offline.

3. Try the PDF you’d never upload

That’s the real test, and the reason the site exists. Bank statements, contracts, passports, medical letters and tax returns are exactly the files that shouldn’t pass through a stranger’s server to be compressed or unlocked. If you’re weighing this site against the others, is it safe? answers the question directly, including what could still go wrong.

What the page does download

  • The page itself: HTML, styles and a small amount of script.
  • Engines, only when a tool needs one: pdf.js and pdf-lib for most tools, qpdf (2 MB) for passwords and repairs, Tesseract and its English model (11 MB) for OCR, and fonts for the PDFs we write. Each is downloaded once and kept by your browser.
  • One analytics script. Microsoft Clarity counts page visits and records how people move through a page. It is blocked from seeing the tool area, so it never has your document’s name, let alone the document. There are no ad networks and no third-party fonts. See the privacy policy.

Why it’s built this way

Most online PDF tools upload your file, process it on a server and send it back. That costs them bandwidth and storage, costs you waiting time, and leaves a copy of your document somewhere you can’t see. Browsers have been fast enough to do this work themselves for years, so there’s no longer a good reason to upload.

What we do and don’t store is set out in the privacy policy, and the architecture on how it works.

Questions

If nothing is uploaded, how does the PDF get fixed?

The page includes pdf.js (Firefox’s PDF engine), pdf-lib, and qpdf and Tesseract compiled to WebAssembly — code that runs inside the tab at close to native speed. The page hands your PDF to that code and writes the result straight back to your downloads.

Why does the network log show requests at all?

Loading a web page means fetching its code, and some tools fetch an engine the first time you use them: qpdf (2 MB) for passwords and repairs, the OCR engine and its English model (11 MB) for scans, fonts for PDFs we write. Those are downloads to your device. None of them carries your file, and their sizes don’t change with the file you add.

Can a website claim “no upload” and still upload?

It can, which is why the claim is worth testing rather than trusting. The Network tab and the airplane-mode test on this page work on any site, including this one.

What about the AI tools — do they send my PDF to a model?

No. Summaries and answers come from your browser’s built-in on-device model where it has one, and otherwise from a ranking model that runs in the page. OCR is Tesseract, running locally. None of them calls a cloud API.