Remove PDF Metadata

Empty a PDF's document information dictionary and cut the catalogue's link to its XMP stream, then take back a clean PDF with every page as it was.

Runs on your devicePDFWorkingNo account needed

What this tool does

Remove PDF Metadata opens the document with pdf-lib and walks the information dictionary key by key, deleting each one it finds: Title, Author, Subject, Keywords, Creator, Producer, CreationDate and ModDate, together with any non-standard key the program that produced the file wrote in beside them. It is not matching against a fixed list of names, so a custom entry left behind by a workflow tool goes with the rest. The line above the download counts what went, and a chip beside it names the keys — that chip is cut off after forty characters, so a long list arrives trimmed.

Two settings decide the shape of the pass. Also remove the XMP metadata stream is on by default, and what it does is delete the /Metadata entry from the document catalogue — the pointer to the XML block where layout and publishing software keeps a second, longer copy of the same fields, plus edit history and document identifiers. Severing that pointer is what stops a viewer's properties panel from naming an author you thought you had removed, and leaving it in place is the ordinary way an anonymised PDF still gives itself away. Keep the creation and modification dates is off by default; switch it on when the document is a record and its filing date has to survive the clean-up.

Be exact about how far the XMP setting goes, because a privacy tool is worth nothing if you guess. The catalogue entry is deleted, so no PDF reader follows it any more — but pdf-lib writes back every object it parsed, and an object nothing points at is written back too. The orphaned XML packet therefore still travels inside the delivered bytes, where the strings command or a forensic parser can read it out. Against a reader, a search index or a properties dialog the stream is gone; against someone examining the raw file it is not.

The whole job happens in this browser tab. pdf-lib is loaded into the page, the file is read into memory, rewritten and handed straight back, so nothing is transmitted and the tool works with the network switched off. One deliberate detail sits behind that: the document is parsed with pdf-lib's own metadata update disabled, because the library otherwise stamps itself in as the Producer and writes a fresh ModDate the moment it opens a file. With every key being deleted that would wash out, but with the dates kept it would hand you today's timestamp dressed up as the document's own.

The narrow scope is the rest of what is worth reading before you rely on it. What is cleared is the document-level information dictionary and the catalogue's XMP pointer, and nothing else. A /Metadata stream attached to an individual page, XMP carried inside an embedded image, and the properties of a file attached to the document all survive the pass. So does everything a reader can see: the text, bookmarks, form field values, content sitting outside the crop box, and annotations — each sticky note and highlight stores its author name in its own dictionary on the page, so comments keep the name of whoever wrote them. The filename is not inside the file either. It is carried over with -clean added, so Q3-forecast-final-jsmith.pdf still announces a person. And because the document is re-serialised rather than patched, an existing digital signature will no longer verify.

How to use it

  1. Choose the PDF. One document is handled at a time, though you can queue several and each comes back as its own file. A PDF that asks for a password cannot be parsed, so run Unlock PDF over it first.
  2. Decide whether the dates stay. Keep the creation and modification dates is off, so CreationDate and ModDate go with everything else. Tick it when the timestamps are part of the record and only the names need to disappear.
  3. Leave the XMP stream switched on. Also remove the XMP metadata stream unhooks the copy Acrobat and InDesign write into the catalogue, which is what readers and indexes go looking for. Untick it only if something downstream reads that block on purpose.
  4. Read the summary, then take the file. It counts the entries removed and says whether the XMP stream was unhooked too, or reports that the document carried no information to remove — the same message you get when the dates were all it had and you chose to keep them. The output keeps the original name with -clean appended.

Processing & privacy

Runs on your device. Your file is processed directly in this browser.

Common use cases

Sending a document outside the organisation

Creator and Producer name the software that built and wrote the file, and Author is normally taken from whoever was signed in to it — three facts about your setup that the recipient did not ask for.

Publishing a form or a template

A PDF exported from an internal draft keeps the draft's title in its information dictionary, and that is the title a reader's window bar and a search engine's result both show, not the filename.

Handing over finished work

Creation and modification timestamps show when a deliverable was actually started and last touched. Clearing them removes a conversation you would rather not have about it.

Tidying a set of files before archiving

Queue the batch and each document comes back with an empty dictionary, so a full-text index built over the archive has nothing to catalogue but the pages themselves.

Questions

Is my PDF uploaded?
No. pdf-lib does the entire job inside this browser tab, so the document is read, rewritten and returned on your own machine. Turn the Wi-Fi off before you start and the pass runs exactly the same.
What exactly gets removed?
Every key present in the document information dictionary — the standard eight and any custom entry alongside them — and, unless you untick it, the catalogue's link to the XMP metadata stream. The dictionary itself stays in the file; it is left empty rather than deleted.
Is the XMP text itself gone from the file?
Not from the bytes, no. Deleting the catalogue entry means no reader follows it and no properties panel shows it, but pdf-lib copies through every object it parsed, so the unreferenced XML packet still sits inside the output and turns up in strings or a hex editor. When the file has to survive that kind of inspection, re-image it — PDF to JPG and then Image to PDF — instead of trusting this pass.
Will the cleaned file say it was made by pdf-lib?
No. That is exactly what the disabled metadata update prevents, so the library does not write itself in as the Producer while parsing. Open the result in PDF Metadata and the fields read back as empty.
Does it strip the author name from comments?
No. An annotation carries its author inside its own dictionary on the page rather than in the document information, so sticky notes, highlights and replies keep the names attached to them. Delete those annotations first if they matter.
Can I keep the dates and lose only the names?
Yes. Tick Keep the creation and modification dates and CreationDate and ModDate are left exactly as they were, while Title, Author, Subject, Keywords, Creator, Producer and anything custom are still deleted.
Does this hide what is written in the document?
No. Every page comes through unchanged, including text you would rather nobody read. Redact PDF covers the words you list and rasterises the page so they are gone for good, and Protect PDF puts a password and permission flags in front of the file.
What if I want to set new values instead of clearing them?
Use PDF Metadata. It reads the existing fields, and in its write mode it sets a new title, author, subject, keywords, creator or producer rather than emptying the dictionary.
Why was my file refused?
Either it is not a PDF, or it is password-protected — an encrypted document cannot be parsed at all, and the message says so and points at Unlock PDF.

Related tools

  • Merge PDF — Combine several PDFs into one document in the order you choose.
  • Split PDF — Split one PDF into separate documents by page range.
  • Compress PDF — Reduce PDF file size while preserving the quality that matters.
  • PDF to JPG — Export every page of a PDF as a JPG image.
  • JPG to PDF — Turn one or many JPG photos into a single PDF.
  • Protect PDF — Add a password and set permissions on a PDF.