How to check that robinpdf never uploads your files
Every browser-based PDF tool says your files stay on your device. The claim costs nothing to make, so it is worth nothing on its own. Here is how to check mine in under a minute — and the same three tests work on any other site, including the ones I compete with.
Test 1 — Turn off your Wi-Fi
Open any tool on this site and wait for the page to finish loading. Now disconnect from the internet: turn off Wi-Fi, or pull the cable. Then use the tool as normal — pick your files, run it, download the result.
It still works. A site that uploaded your document to a server could not possibly do that. This is the whole test, and it takes about thirty seconds.
Test 2 — Watch the network tab
Press F12 to open your browser's developer tools and switch to the Network tab. Leave it open while you process a file. You will see the page's own assets and the ads load, and you will not see your document go anywhere: no upload request, no POST carrying the file, nothing.
This is the test to run on any site that claims local processing. If you see your file in a request body, the claim was false.
Test 3 — Read the security policy
Browsers let a site declare, in a header called Content-Security-Policy, which outside servers the page may talk to. You can read mine in the Network tab: click the first request and look at its response headers. Besides Cloudflare's cookieless visit counter, it lists Google's ad and consent servers, because the site is paid for by ads.
Those servers get what any ad gets: the page you are on, your browser and your IP address. They do not get your file, and you do not have to take my word for that: Test 2 would show any request carrying it, and Test 1 works with no connection at all.
What I measured
robinpdf has an automated test that loads a tool in a real browser, cuts the network at the browser level, and then processes a file. Of the 16 tools it covers, 14 complete the job with the network down:
- Merge PDF
- Split PDF
- Rotate PDF
- Crop PDF
- Compress PDF
- PDF to JPG
- Grayscale PDF
- Pages Per Sheet (N-up)
- PDF to Text
- Add Page Numbers
- Edit PDF Metadata
- PDF to Word
- PDF to EPUB
- PDF to Excel
Where it does not hold
I would rather tell you this than have you find it.
- OCR and the password tools download a processing engine on first use — tesseract for text recognition, a WebAssembly build of qpdf for encryption. Together those are several megabytes, which is why they are not loaded for everyone who opens the page. That is program code coming down to your browser. It is never your file going up. PDF to Word is on the list above because a text PDF needs no OCR; a scanned one does, and then it needs that first download too.
- The ads run on the same page as the tool. Any script on a page can in principle read what is on that page, so what you can check is that nothing sends your file anywhere (Test 2). For a document that must not leave your computer under any circumstances, load the tool and then disconnect before you open the file: with no network, nothing can leave.
- Compressing a very small PDF does nothing at all: the tool checks whether it can actually make the file smaller and declines rather than handing you a worse copy. That is a refusal, not a failure.
Why build it this way
Because "we delete your file within a few hours" is a promise, and an architecture that never receives the file is a fact. For a holiday photo the difference does not matter. For a contract, a payslip, a medical report or a scan of your ID, it is the only thing that matters. If you handle documents you are not allowed to upload, you should not have to take anyone's word for it — including mine.