Claude Text Cleaner

Private-use text guide · October 23, 2026

Does Claude Text Cleaner Remove Private-Use Characters?

Yes. This browser-only cleaner deletes code points in all three Unicode private-use ranges, even if a custom font displays one as a useful icon or letter. In a live test, A + U+E8A1 + B became AB. The report showed original length 3, clean length 2, removed characters 1, changed categories 1, and “Private-use characters” 1. Keep the original when the symbol carries meaning.

Try the browser cleaner

The code-point labels remain readable when the original glyph is unavailable in your font. This is a text change, independent of its appearance.

Why a visible symbol can disappear

Unicode's private-use FAQ defines these code points as having no standard interpretation. A particular font or application may give one a glyph and meaning; another font may show a box. The cleaner cannot know that private agreement. Its browser script removes U+E000–U+F8FF, U+F0000–U+FFFFD, and U+100000–U+10FFFD by code point, whether the glyph is visible or not.

A public question about U+E8A1 displaying as a box was resolved by finding the intended icon font. Another question about U+F0E1 in React reports a square in place of a font icon. Those examples illustrate why missing-glyph diagnosis and deleting a code point are separate decisions. They do not establish where any pasted character originated.

Check before copying the cleaned result

  1. Keep the original text, especially if it came from a document, icon font, or custom writing system.
  2. Paste a copy into the cleaner and review the “Private-use characters” count alongside the before-and-after text. A positive count means matched code points were deleted, not that a watermark was found.
  3. If the symbol matters, restore the original in an editor with its intended font. If you need portable text, choose an appropriate standard character or a textual label yourself; this cleaner does not infer a replacement.

The local report counts code points; it does not inspect fonts, detect the text's author, or identify a hidden watermark. For the arithmetic behind the report, see how to read removal counts. For another case where appearance and stored text differ, see the subdivision flag guide.