Clean PDF Metadata

What metadata is hiding in your PDF?

Every PDF document carries a small dictionary of "Info" metadata that most people never see. It's not the visible content of the document — it's data *about* the document. Common fields include:

• Author — usually the operating system username of whoever created it. ("Maria Silva" on a corporate laptop, or worse: "msilva" matching an email login.)
• Creator — the software that wrote the document originally ("Microsoft Word 2019", "LibreOffice 7.4", "Adobe Acrobat Pro DC").
• Producer — the PDF library that produced the final file. Tells you whether it was exported, converted, or printed-to-PDF.
• Creation date and Modification date — exact timestamps, often down to the second. These can reveal when you wrote it (3 AM the night before submission?) or that a file you claimed was unchanged was actually edited yesterday.
• Title, Subject, Keywords — set automatically by some Office templates and often contain internal project names.

For many documents, this is harmless. For a journalist receiving a "anonymous" leak, an HR resume from a "different" laptop, or a court submission, it can be a critical disclosure.

What this tool actually does

Your PDF is read into your browser's memory using pdf-lib, a JavaScript PDF library. The Info dictionary is enumerated, displayed to you, and on download we generate a new PDF file with the selected fields cleared. The original file on your disk is untouched.

We never upload anything. The browser tab does the entire job. You can verify this yourself: open DevTools → Network tab, run a clean, and watch — no outgoing requests.

What this tool does NOT do

A few important limits:

• It does not strip XMP metadata embedded in object streams. PDF spec has two parallel metadata systems: the Info dictionary (what we clean), which for most files is the noisier one, and the XMP dictionary.

• It does not remove text content, images, or comments. If your concern is hidden text, redacted-but-not-actually-redacted boxes, or revision history, you need a more aggressive tool.

• It does not preserve digital signatures. Any modification to the file's bytes invalidates the signature. If you need both a clean document AND a valid signature, sign *after* cleaning, not before.

What can't be cleaned (and why)

A few metadata fields cannot be removed — only replaced. This isn't a choice we made; it's a limit of the underlying PDF library. Worth knowing what's happening:

Producer (the PDF library signature). Every PDF carries a /Producer entry naming the software that wrote the final bytes. We use a JavaScript library called pdf-lib to read your file and write the cleaned copy. When pdf-lib saves a PDF, it stamps its own name — pdf-lib (https://github.com/Hopding/pdf-lib) — into Producer, overwriting whatever was there. There's no public API to suppress this. Byte-level surgery on the output to wipe Producer would risk corrupting PDFs that use modern compression features (Object Streams), and silent corruption is worse than a generic signature. The leak is small: seeing "pdf-lib" tells someone the file was passed through an open-source library; it doesn't identify you, your computer, or the document's actual history.

Creation date and modification date. The PDF spec allows these to be absent, but pdf-lib will not produce a PDF without them — and even if we could remove them, the absence itself is a signal ("this file had its dates stripped"). The honest move is to replace them with something that isn't your real timestamp. The tool gives you two strategies:

• January 1 of the original year (default). Keeps the year of the original document and zeroes out month, day, and time. Plausible — a PDF "from 2024" doesn't look suspicious. Loses the day-level detail that an investigator could correlate with other events.
• 1970-01-01 (Unix epoch). Completely neutral, but creates a recognizable pattern: anyone analyzing the file will know it was processed through a stripping tool. Use this when you'd rather signal "scrubbed" than reveal anything specific.

Best practice: keep both files

We download the cleaned version as yourdocument-clean.pdf. The original stays on your machine. Get in the habit of keeping both, at least temporarily — if you discover the cleaned version broke something you needed (a signature, a form field's metadata), you can always go back.