What you are agreeing to
Very little, which matches how the site is built. There's no account to create, nothing to pay, and no ongoing relationship: you load a page, your browser does the work, and you leave. Use the tools for lawful purposes on documents you have the right to process, and there's no further obligation either way.
What stays yours
- The decision that an output is fit to send
- A tool can remove pixels; only you can decide whether the result satisfies a court, a regulator, or a client. Open the file before you send it. If the stakes are high, check the output the way you'd check someone else's work: by looking at it.
- The right to process the document
- Employment contracts, medical records, another person's identification: whether you may run them through software at all is a question about you and the document, not this site.
- The file itself
- Everything stays in your browser. That also means there's no copy here to recover, no version history, and no undo once you close the tab. The tools never keep a duplicate, so keep your own.
Limits worth knowing before you trust a result
These are properties of what PDFs are and what a browser tab can do, not defects that might be fixed later if you wait. They matter most when someone else will rely on the file:
- OCR covers printed English, and only where you ask for it
- Text extraction reads the text layer that already exists in the file, and a scan that was never recognized has none. One tool can recognize it instead: the extractor's “OCR scanned pages” option, which pulls down roughly 7 MB of recognition engine the first time you use it. Handwriting, other languages, and poor scans come back wrong, so treat that output as a draft to check rather than a transcript to trust.
- Encryption isn't bypassed
- If a file opens with a password, the tools use the password you already have. There's no recovery service and nothing is cracked; that limit comes from the format, not a shortcut taken here.
- Some operations change what the file contains
- Redrawing pages as bitmaps is what makes redaction real and lets the compressor shrink a scan. It also removes the things that made those pages text: selectable text, fillable form fields, links, comments, and bookmarks. Where that matters, the tool tells you and offers the lossless output too. A black rectangle drawn over text is different: it changes none of that, which is why the redaction tool doesn't use it.
- Big files meet the same ceiling your browser has
- Everything happens in one tab, so thousands of pages or very high-resolution scans are limited by the memory that tab can get. Work is handed back to the interface regularly, but this is your machine doing the work, and it has limits.
This isn't a promise that the tools will never let you down. It's a list of the ways they can, so you find out here rather than at the other end.
No warranty
The tools are provided as they are, free, with no promise that they'll work for your particular file or be available when you want them. Third-party libraries are bundled as published by their authors: pdf-lib under Apache-2.0, pdf.js under Apache-2.0, JSZip under the MIT license or GPLv3 at your option, and MozJPEG (compiled to WebAssembly) under its bundled IJG/BSD-style license. Each carries its own license terms alongside the file that ships here.
Where liability can be excluded, it is. Where the law doesn't allow that, and different places draw the line differently, it isn't, and nothing above pretends otherwise.
Changes
Last updated 2026-09-17. These terms don't reserve the right to quietly rewrite this page and take something away from you; if the wording changes in substance, the date changes too.
Anything unclear, or anything you think belongs on this page and is missing:hello@pdfsanctum.com.