What does your PDF give away?
A PDF carries far more than the page you can see: who wrote it, on what software, when it was edited, what was attached to it, and often the whole of its previous version. Drop one in and find out. The file is read in your browser and never sent anywhere.
This tool reports; it does not yet edit. It will tell you exactly what is in your document, and it will not hand you a modified one. Rewriting a PDF means rebuilding its internal object graph, and doing that imperfectly produces a file that opens fine today and fails in someone else's reader next month. For a contract or a filing that is a worse outcome than knowing the metadata is there. Removal is being built, and it will ship when it can be proved correct rather than when it appears to work.
What this finds
Earlier versions that never went away
This is the one that catches people. When you edit a PDF, most software does not rewrite the file — it appends the changes and leaves the previous version in place, because that is faster and lets readers recover damaged documents. The result is that a PDF which has been saved four times may contain all four versions.
If something was deleted, covered with a black rectangle, or corrected before one of those saves, the earlier text can still be sitting in the file. This is the mechanism behind essentially every "redacted document was not actually redacted" story you have read. The tool counts the revisions and tells you when there is more than one.
Who and what made the file
The document information dictionary usually holds an author name — taken from the account name on the machine, which is frequently a real person's full name nobody chose to publish. Alongside it sit the creating and producing applications, often with exact version numbers, and creation and modification timestamps accurate to the second.
The title field deserves particular attention: it is commonly the original filename rather than the document's visible heading, so a report presented as Annual Review may still identify itself as draft3_before_legal.docx.
Attachments, comments and scripts
- Embedded files. Entire documents can be attached inside a PDF and travel with it invisibly — the spreadsheet behind the summary, for instance.
- Comments and mark-up. Sticky notes, highlights and review remarks are stored as annotations. They are easy to hide from view and easy to forget before sending.
- JavaScript. PDFs can contain scripts that run when opened. Usually that is form logic; occasionally it is not.
- XMP records. A parallel XML metadata block that repeats much of the above and often adds a document identifier that survives editing, linking versions of a file to each other.
This is not redaction, and will never claim to be. Drawing a black box over text in a PDF editor hides it visually while leaving the text itself intact underneath, where anyone can select and copy it. People have lost cases that way. True redaction means removing the content, then verifying it is gone. If you are redacting a document that matters, use software built for that purpose and check the result.
Printing to PDF does not rescue you here either, and it is worth being clear about why: the print pipeline draws the rectangle and the text underneath it, and both end up in the new file, still selectable. Print to PDF fixes metadata and revision history. It does not fix content that was covered rather than deleted.
Common questions
Is my document really not uploaded?
Correct, and you do not have to take our word for it. This site is static files on GitHub Pages with no backend to receive anything, and the entire parser is public. Open your browser's developer tools, switch to the Network tab, and inspect a file — nothing appears. Or disconnect from the internet and keep using the page. See how to verify for more.
How do I actually remove what you found?
Not on this page — this tool only reads. The fix happens on your own computer. Open the document in whichever PDF reader you normally use, choose Print, and set the destination to Save as PDF (on Windows, Microsoft Print to PDF).
That writes an entirely new file from the printed pages. Nothing is carried over from the old one, so the earlier revisions and the original author and software fields are not so much removed as never written. Attachments, comments and bookmarks go the same way, which may or may not be what you want.
Then bring the new file back here and check it. Printing software writes metadata of its own, and on several systems that puts your account name straight back into the author field — precisely what you were trying to clear. Confirm rather than assume; it takes a few seconds.
Two other trade-offs worth knowing: some print paths turn pages into images, which loses selectable and searchable text, and this does nothing about text hidden under a black rectangle. See below.
Why can it read metadata that isn't in the file's text?
Modern PDF writers pack many objects into compressed streams, so the author name is genuinely not present as readable text anywhere in the raw bytes. Searching a PDF in a text editor will therefore miss it. This tool decompresses those streams using the browser's own built-in decompressor and reads what is inside.
My PDF is encrypted. Will this work?
Partly. Structure is still readable, but text strings may not be, so results will be incomplete — and the tool says so when it detects encryption rather than quietly reporting less than it should.
Is there a file size limit?
No. The work happens on your machine and costs us nothing, so there is no reason to impose one. Very large files simply take a moment longer.
Handling genuinely sensitive material? Save this page to your
computer (Ctrl/Cmd+S) and open it with your
network turned off. It works exactly the same. There is no licence check and no call
home, because there is nothing to call home to.