The file never leaves your browser.
Drop a photo, a document or a PDF. It is read and cleaned on your own machine. Nothing is uploaded.
A photo knows where it was taken. A document knows who wrote it.
Photos
JPEG and PNG carry the camera and lens, the exact time, GPS coordinates to a few metres, the owner's name, the software used, and often a small copy of the original picture from before it was cropped.
Read about photos
Word, Excel, PowerPoint
Author, company, who last saved it, the revision count, total editing time, and the names on every comment and tracked change.
Read about documentsThe Info dictionary and the XMP packet: author, creator software, producer, creation and modification dates. Some of it sits in compressed streams, and the page says which.
Read about PDFThe browser does the work.
There is no server in the loop. Three steps, all of them on your machine.
How it works- 01
Drop
Drop a file on the panel, paste one, or browse. The browser hands the page a reference to the file; the file is not copied anywhere.
- 02
Read
The parser reads only the header bytes it needs: a JPEG's segments, a PNG's chunks, an Office package's properties, a PDF's Info dictionary. Every field is listed.
- 03
Download
The clean copy is assembled from slices of the original, with the metadata parts left out, and saved with a "-clean" suffix. The pixels and the text are byte for byte what they were.
Six formats today.
Every format, read and removed- JPEGAll of it. The ICC colour profile stays so the colours do not shift. If the photo was shot sideways, one orientation value is written back so it still displays the right way up.
- PNGAll of it. The colour profile stays.
- WordThe three property parts are emptied, the thumbnail is dropped and every zip entry's own timestamp is reset. Comment and tracked-change authors are shown, not removed: that is an edit to the document, so it stays your call.
- ExcelThe three property parts are emptied, the thumbnail is dropped, the zip timestamps are reset.
- PowerPointThe three property parts are emptied, the thumbnail is dropped, the zip timestamps are reset.
- PDFInfo values and an uncompressed XMP packet are blanked in place, same length, so nothing else in the file moves. A compressed XMP stream, which is what Word and Acrobat usually write, is shown and kept, and the page says so. Metadata inside compressed object streams is not read yet.
There is nothing here to trust.
A cleaner that uploads your file has to promise things: encrypted in transit, not stored, deleted after. FileSanity makes no promise because the file is never sent. There is no account, no cookie, no analytics script, and no server that could keep a log.
You can check rather than believe. Open the network panel, drop a file, and watch. The counter beside your result reads the same panel.
Security
Send the file, not the story of who made it.
Free, no account, no upload. Works offline once the page has loaded.