Two places the details hide
The Info dictionary is the obvious one: title, author, subject, keywords, the producing application, and two timestamps for created and last modified. It’s a short list of plain text fields, and every reader can show it to you.
XMP is the one people don’t know about. It’s an XML block that can sit alongside the pages and record far more: the full editing history, the original file path, the camera or scanner model, sometimes coordinates. It’s often larger and more revealing than the Info dictionary, and nothing in a normal reader surfaces it.
What each field actually gives away
- Author
- The account name the software was licensed to. On a work machine this is often your full name, not the name on the document.
- Creator
- The application that authored the content: Word, InDesign, a scanner driver. Reveals the workflow behind the file.
- Producer
- The library or virtual printer that produced the PDF, usually with a version number. Combined with Creator, it identifies the toolchain.
- Creation and modification dates
- A timeline. When the file was made and when it was last touched, which isn’t always when the content stopped changing.
- Custom and company fields
- Whatever the originating system chose to write. Templates often stamp a firm name, a client code or a matter number here without anyone deciding to.
- XMP history
- The most revealing block when present. Original path, revision history, sometimes device information.
Where the risk actually sits
Metadata is a small leak, and it matters most in the cases where the document is going somewhere adversarial. A CV sent to a company you don’t want knowing your current employer. A redacted document where the author field names the person the redaction was hiding. A quote going to a competitor, carrying the version of the software that produced it.
It also matters cumulatively. One file with an author field isn’thing. A hundred files from the same source, each carrying the same fields, is a pattern, and it’s the pattern that gets noticed, not any single file.
See it before deciding
The tool reads the file and shows you the whole picture: the Info fields it found, the risk rating it assigns each one, and whether an XMP block is present. Nothing is modified at this stage, because it’s read-only and the original file is untouched on your disk.
Then choose one of two operations. Wipe everything removes the Info dictionary entirely and deletes the XMP block from the document catalog. Write specific fields lets you set the ones you want and clear the rest.
How you know it worked
Most tools tell you the metadata is gone. This one reopens the document it just wrote and walks its structure to confirm two things: the Info dictionary reference no longer exists in the trailer, and there’s no XMP stream in the catalog.
That’s a structural check rather than a text search, which matters more than it sounds. Metadata inside a compressed object stream is invisible to a plain text scan, so a tool that proves its work by grepping the output can report clean on a file that isn’t. Confirming the reference is gone doesn’t depend on how the bytes were stored.