Choose a file to identify
Any file type is accepted. Only the first 4 KB are read — nothing is uploaded.
What this tool does
Most file formats begin with a recognizable byte sequence — often called "magic bytes" — regardless of what the file name or extension claims. This tool reads only the first 4 KB of a file, checks that prefix against a database of known signatures, and reports what it actually found.
The main thing it looks for is a mismatch: a file named "report.txt" that is actually a PNG image, a ".jpg" that is really a ZIP archive, or any other case where the extension disagrees with the content. If no binary signature matches, it checks whether the content decodes as UTF-8 or Shift_JIS text, and reports whether that text looks like JSON, CSV/TSV, HTML, XML or SVG.
Why your files stay on your device
Privacy here is structural, not a promise. There is no upload step because there is no server to upload to:
- Only the first 4 KB of the file are ever read — never the whole file.
- Identification runs entirely in your browser.
- The page is served as static files and makes no request with your file's data.
- The source is open and anyone can read it (MIT).
- It works offline, which is only possible because nothing leaves the device.
If you want to check for yourself, open your browser's Network panel while identifying a file — no request carries your data. Read the source.
How to use it
-
Choose a file
Click to select any file, or drop it anywhere on the page. One file at a time.
-
Read the result
See the detected format, its MIME type and common extension(s) — or, for text, the encoding and content type.
-
Watch for the warning
If the file's own extension doesn't match the detected format, a mismatch warning appears first.
FAQ
Is my file uploaded anywhere?
No. Only the first 4 KB are ever read, and identification runs entirely in your browser. There is no server component, so the file has no path off your device. The source is open and you can confirm this in your browser's Network panel.
What are "magic bytes"?
Most file formats start with a fixed, recognizable byte sequence (also called a file signature) — for example, PNG files always start with the same 8 bytes, and ZIP archives with "PK\x03\x04". This tool checks a file's first bytes against a database of known signatures (magic-bytes.js) to report the actual format, independent of its file name.
Why does it say a file could be several different formats?
ZIP-based container formats — ZIP, JAR, DOCX, XLSX, PPTX, ODT, APK and others — share the exact same first few bytes. Telling them apart usually requires looking inside the archive's central directory, which often falls outside the first 4 KB this tool reads, so it lists every format that shares the signature instead of guessing.
What if the file isn't a recognized binary format?
If no binary signature matches, the tool tries to decode the first 4 KB as UTF-8 text, then as Shift_JIS. If that succeeds, it reports whether the content looks like JSON, CSV/TSV, HTML, XML/SVG or plain text. If neither decodes, it reports "unknown binary data" — the bytes match no known format and are not valid text either.
Does it convert or open the file?
No. This tool only identifies what a file is — it doesn't convert, extract or view its contents. Where a matching runlocally tool exists (for example ZIP Viewer, CSV Viewer or Hex Viewer), the result links to it.
Does it work offline?
Yes. It's a PWA. After the first visit it is cached, so identification works without a network connection. You can also install it to your home screen.