How to view and remove PDF metadata
- Add the PDF. Tap Choose PDF or drop the file onto the page. If it is password-protected, type the password and tap Open.
- Read what it reveals. The table lists each metadata field with its value, or “not set”. Below it you see whether an XMP packet is embedded and what it says about the app and person that made the file.
- Remove everything. Tap Remove all metadata. The tool deletes every Info entry and the XMP stream, writes a new PDF, re-reads it to confirm the fields are gone, and downloads it as filename-clean.pdf.
- Check what is still inside. Metadata is only one layer. Visible text, names in headers, comments and form fields stay as they are — use Redact PDF for those.
What it does
- Shows all nine standard Info dictionary fields plus any custom keys the producing app added
- Detects an embedded XMP metadata stream and surfaces its creator tool, creator name and document IDs
- Remove all: clears every Info field, deletes the XMP stream and the private PieceInfo block, then saves a new file
- Verifies the result by re-reading the cleaned file before you download it
- Opens password-protected PDFs when you know the password, and can keep the password on the cleaned copy
- Runs entirely on your device — the only network request fetches the open-source pdf-lib library
What metadata does a PDF carry?
A PDF file has two places for information about the document, separate from what you see on its pages. The first is the document Information dictionary — a short list of key/value pairs defined in the PDF standard (ISO 32000-1:2008, section 14.3.3). The second is an XMP metadata stream, an XML packet in Adobe's Extensible Metadata Platform format, referenced from the document catalog (section 14.3.2). Modern apps usually write both, and they often disagree with each other, which is why this tool shows them separately.
Which field leaks what, and does this tool remove it?
| Field | What it can reveal | Removed by this tool? |
|---|---|---|
| Title | The original document or file name, sometimes a draft name you renamed later | Yes |
| Author | The user name of the account that created the file — often a real name or e-mail | Yes |
| Subject, Keywords | Topic lines and tags typed into document properties, or inherited from a template | Yes |
| Creator | The app that made the document (Word, Pages, Google Docs, a scanner model) | Yes |
| Producer | The PDF engine and version, sometimes a licence holder's name | Yes |
| CreationDate, ModDate | When the file was made and last changed, with time zone — can contradict a date printed on the page | Yes |
| Trapped | A prepress flag; harmless but identifies print workflows | Yes |
| Custom Info keys | Anything the producing app added: company name, source document path, template IDs | Yes |
| XMP stream | Creator tool, author, document and instance IDs that link versions of the same file, edit history (xmpMM:History) | Yes — the whole stream object is deleted |
| PieceInfo | Private application data (Illustrator, InDesign, Photoshop save settings) | Yes |
| Page text, headers, footers | Names, addresses, account numbers that are simply printed on the page | No — use Redact PDF |
| Embedded font names | Which fonts (and sometimes which OS) you use | No |
| Annotations and form fields | Comment author names, sticky notes, typed form values | No |
| EXIF / XMP inside embedded images | Camera model, GPS location, capture time of a photo placed in the PDF | No — strip it from the photo before placing it (Photo metadata) |
| Attachments, bookmarks, links | Attached files, outline titles, link targets | No |
How does the removal work?
The tool parses the file with pdf-lib and opens it with its metadata update switched off, so the library does not stamp its own Producer and date on the way in. It then deletes every entry of the Info dictionary, removes the catalog's Metadata reference and the XMP stream object it points to, drops PieceInfo, and serialises a fresh file. Because pdf-lib rewrites the whole object table rather than appending an incremental update, the old values are not left behind in the file as dead objects — a trap with some editors that “clear” the properties but keep the previous revision in the file. The output is written with plain-text dictionaries so you can confirm the result yourself: open it in a text editor and search for /Author or x:xmpmeta.
When should I strip PDF metadata?
- Before sending a CV, a complaint, a tenancy application or any document where your software, your user name or your editing timeline is nobody's business.
- Before publishing a report or form: the Author field often names a staff member, not the organisation.
- When a scanned document should not reveal the scanner model, driver and scan time.
- Before sharing a file whose Title still carries its working name (“offer_v3_final_LOW.pdf”).
What are the limits?
- Document-level metadata only. Visible content is untouched by design — a name printed on page one stays on page one.
- Damaged files, or files that pdf-lib cannot parse, fail with a message rather than a partial result.
- Encrypted files open only with the password. Owner-password restrictions are not preserved in the cleaned copy.
- Digital signatures become invalid, because the file bytes change. That is true of any edit to a signed PDF.
- Large files (hundreds of megabytes) are rewritten in memory; a phone may run out before the save finishes.