The current compression and merge tools read selected PDF content in browser workers. Document bytes are not uploaded to BytesPDF servers, but the browser, device, loaded code, and downstream recipient still belong to the security boundary.
Working blobs stay in tab memory and are not persisted in IndexedDB, Cache Storage, or BytesPDF server storage.
The site still downloads HTML, JavaScript, worker, and codec assets. Supabase can handle account, subscriber, and consent-audit data; Formspree can receive contact submissions; consented analytics and configured error reporting are separate from PDF content.
Read the complete data inventory →One selected PDF at a time.
The file is checked against the 150 MB desktop or 50 MB mobile input limit and read in the current tab.
The preset controls supported image re-encoding. Scans and photographs can change; born-digital vector text is not rasterized.
Browser workers inspect and rewrite supported content using local CPU and memory. Processing time varies by file and device.
The result is a separate file. If the generated version is not smaller, BytesPDF returns the source instead.
Whole-file ordering and rotation controls.
Desktop supports up to 20 files and 150 MB combined; mobile supports up to 10 files and 50 MB combined.
Arrange whole source files and apply the available source rotation controls.
The merge worker copies visible pages into a new document in the selected order.
Check page order, orientation, links, forms, bookmarks, annotations, attachments, and other structures you need.
Compression and merge reject files with a detected signature ByteRange rather than claiming a cryptographic signature will survive a rewrite.
Password-protected inputs are rejected. Unlocking must happen through an authorized workflow elsewhere.
BytesPDF does not certify PDF/A conformance. Keep the archival master and use an appropriate external validator.
BytesPDF does not create OCR. Existing OCR text may remain, but scan-image changes can affect visual alignment and later recognition.
Compression clears common descriptive and XMP fields. It is not a forensic sanitizer for every attachment, annotation, image value, or application-specific object.
The published cap is an input rule, not a promise that every file below it will fit a particular device's available memory.
A valid-looking PDF is not proof that every bookmark, form, annotation, attachment, tag, color profile, or recipient-specific requirement survived. Save the output as a separate copy and test the features your workflow uses.
Open the output-review checklist →No account is required for the core tools. Processing time and results depend on the selected PDFs and device.
