Strip author, company and hidden history from Office and PDF files — in your browser
| What leaks | Where it lives | Example content | Removed |
|---|---|---|---|
| Author | docProps/core.xml (dc:creator) | Sign-in name of the Office account | Yes |
| Last modified by | docProps/core.xml (cp:lastModifiedBy) | "Microsoft Office User" or a colleague's name | Yes |
| Created / modified dates | docProps/core.xml (dcterms:created, dcterms:modified) | Exact timestamps, often two time zones apart | Yes |
| Revision number | docProps/core.xml (cp:revision) | Save count — 47 tells a story | Yes |
| Company | docProps/app.xml (Company) | Set at install time, appears in every file | Yes |
| Manager | docProps/app.xml (Manager) | Org-chart data inside a spreadsheet | Yes |
| Template | docProps/app.xml (Template) | Normal.dotm or a named HR form | Yes |
| Total editing time | docProps/app.xml (TotalTime) | Minutes of editing, in minutes | Yes |
| Custom properties | docProps/custom.xml | Sensitivity labels, client names, reviewer IDs | Whole part deleted |
| Preview thumbnail | docProps/thumbnail.jpeg | Snapshots deleted content, often old versions | Whole part deleted |
| PDF Info dictionary | Trailer /Info dict | Author, title, producer app, dates | Emptied |
| PDF XMP block | Catalog /Metadata stream | Dublin Core echo of the same fields | Stream removed |
| Tracked changes & comments | word/document.xml (content) | Not metadata — this tool never touches content | No (by design) |
Every "Yes" row above was verified by building files with those exact fields, cleaning them with this tool's code, and re-opening the packages to confirm the parts are gone. The manifest and relationships that pointed at deleted parts are rewritten too, so Word, Excel, PowerPoint and every PDF reader open the cleaned file without a repair prompt.
Word, Excel and PowerPoint files (anything ending in x) are ZIP packages. The text you see lives in one part; the file's description of itself lives in others — docProps/core.xml, docProps/app.xml, docProps/custom.xml and an optional thumbnail. Those property parts are what File Explorer, Google Drive and document review software read to show "Author: [email protected]" next to your resume.
This tool unpacks the package in memory with JSZip, replaces core.xml and app.xml with clean minimal versions, deletes custom.xml and the thumbnail, and rewrites the [Content_Types].xml manifest and root relationships so nothing references a missing part. For PDFs it loads the file with pdf-lib, empties the trailer Info dictionary and removes the XMP metadata stream from the catalog and any pages that carry one. Both libraries run locally in your tab; first use downloads about 350 KB of JavaScript, cached afterwards.
A performance-review .docx built with realistic fields came in at 8,185 bytes. Before cleaning, it exposed 18 items: the author (Jane Q. Author), last modified by, two timestamps, revision 7, title, subject, keywords and description in core.xml; Company (Acme Corp Legal Dept), Manager, Template, HyperlinkBase and 184 minutes of TotalTime in app.xml; 3 custom properties including a sensitivity label and a client name; and a preview thumbnail. After cleaning: 7,262 bytes, zero identity fields, the package manifest valid, and the body text untouched. The same run on a PDF with an 8-field Info dictionary (title, author, subject, keywords, creator, producer, two dates) produced a 903-byte file with an empty Info dictionary, no XMP stream, and not even a producer fingerprint left behind.
That last part deserves emphasis: many "cleaners" leave the producer field showing which tool did the cleaning. This one writes nothing in place of what it removes, so the cleaned file doesn't advertise how it was made.
In Word you can run File > Info > Check for Issues > Inspect Document and remove the properties it finds, but many versions keep some fields. This tool takes the surer route: it opens the .docx package, rewrites docProps/core.xml and docProps/app.xml with clean minimal versions, deletes docProps/custom.xml and any preview thumbnail, and fixes the package manifest. Your visible text, tables and formatting are untouched, and nothing is uploaded.
More than most people expect. A typical Word file stores the author (from the Office account that created it), last modified by, created and modified timestamps, revision number, total editing time, the company name and manager from app.xml, the template it came from, hyperlink bases, up to dozens of custom properties, and sometimes a preview thumbnail. In testing, a routine HR document carried 18 separate identifying items.
No, and this surprises people. A PDF carries its own Info dictionary with author, title, subject, keywords, producer and creator app, plus an XMP metadata block. Exporting a dirty Word file to PDF copies the author and title straight over. The PDF path in this tool empties the Info dictionary and removes the XMP stream so both copies are clean.
No, and that separation matters. Metadata is the file's description of itself; tracked changes, comments and hidden text are document content. This tool deliberately never touches content, so formatting and revisions survive intact. To remove revisions, accept or reject them in Word first, or rasterize the final file. For redacting visible material from a PDF, use a redaction tool that destroys the underlying text.
Not directly. Those 1997-2003 formats are binary OLE containers, not ZIP packages, so the in-browser approach does not apply. Open the file in Word, Excel or PowerPoint and save it as .docx, .xlsx or .pptx first, then clean it here. The tool tells you this rather than pretending to work.
No. The cleaning runs in your browser tab with JavaScript: JSZip rewrites the Office package in memory and pdf-lib rewrites the PDF structure, both on your machine. There is no server round trip, the tool keeps working offline after the libraries load, and it is a sensible choice for contracts, resumes, HR files and anything else you would not paste into a stranger's upload form.