Frequently asked questions
Privacy, file-size limits, supported formats, and the answers to questions people open a private window to ask.
Is my PDF really not uploaded?
Really. Open your browser's DevTools (right-click → Inspect → Network tab), then drop a PDF. You'll only see requests for this site's own files (scripts, the WebAssembly decoders for scanned PDFs) — no third-party domains, ever. The page's Content-Security-Policy header (visible in the same DevTools panel under "Response Headers") forbids any cross-origin request, so the browser itself blocks any attempt to send your file off-site, no matter what the code tried. Step-by-step walkthrough: how to verify this tool.
Does this work on scanned PDFs?
Yes. Scanners and apps like Adobe Scan typically save pages using JBIG2 or JPEG2000 — image formats browsers cannot decode natively. This tool ships small WebAssembly decoders (extracted from Mozilla's pdf.js) that handle both formats locally. CCITT-Fax (older fax-format scans) is supported by the same WASM module. The decoders are downloaded lazily, only the first time a scanned PDF is dropped in.
Is there a file-size limit?
None imposed by us. The practical limit is your device's memory — the PDF is loaded fully into RAM while it's being processed. Most laptops handle PDFs up to 200–300 MB without trouble.
Why didn't my PDF get smaller?
Most of a PDF's size is usually images. If your PDF is mostly text (a Word export, a LaTeX document, a typeset book), there's very little to compress beyond a structural rewrite — typically under 5%. The tool reports this honestly when it's the case rather than pretending otherwise.
Is the compression lossy?
The three image-recompression presets (Strong, Balanced, Light) are lossy by design — that's how they save space. Use the Lossless preset if you need bit-for-bit image fidelity; it strips metadata and rewrites structure only, which typically yields 0–10% savings.
Why might a competitor still compress more?
Server-based tools like Smallpdf and Adobe Acrobat run a full PDF toolchain (often Ghostscript) with more aggressive techniques: JBIG2 re-encoding of bilevel images, font subsetting, vector simplification. They will sometimes beat us on edge cases — at the cost of uploading your file. We trade a little compression ratio for the privacy guarantee.
Do I need to install anything?
No. The tool is a single web page. Once you open it, every operation happens in the tab. You can save the page (Ctrl/⌘+S → "Save as web page, complete") and run it entirely offline if you want.
Does it work offline?
Yes. After the first page load, all the JavaScript and the optional WebAssembly decoders for scanned PDFs are cached by your browser. You can disconnect from the internet and the tool keeps working. For maximum confidentiality, save the page locally and run it without ever reconnecting.
Can I trust this tool with confidential documents?
Trust shouldn't rest on our promises — it should rest on what's verifiable. The deployed page is the source: View Source on the running page and you see exactly what runs. The site's Content-Security-Policy header (visible in DevTools) blocks any cross-origin network request. See how to verify this tool for the 30-second verification walkthrough and dependency hashes.