The hidden character detector shows the invisible characters lurking in text — zero-width spaces, non-breaking spaces, byte-order marks, right-to-left overrides and control codes — that break code, searches, passwords and spreadsheet formulas. Paste text to see every hidden character marked in place with its Unicode name and code point, then choose which kinds to remove and copy the cleaned text. It also warns about Trojan Source bidi tricks in code. Everything runs in your browser.
| Code point | Name | Usual source |
|---|---|---|
| U+200B | Zero width space | Web pages, CMS editors |
| U+00A0 | No-break space | Word, HTML |
| U+FEFF | Byte order mark | Files saved by Windows tools |
| U+200D | Zero width joiner | Emoji sequences |
| U+202E | Right-to-left override | Spoofed file names, Trojan Source |
| U+00AD | Soft hyphen | Hyphenated web text |
U+200B, a character with no visible width. It's used for line-break hints but often sneaks in when copying from web pages and breaks searches and code.
One usually contains a hidden character, such as a zero-width space or a non-breaking space instead of a normal one. Paste both here to see it.
Hiding bidirectional control characters in code so it displays differently from how it runs. The detector warns when they're present.
No. Joiners inside emoji like 👨💻 are kept unless you turn that option off.
No. It's checked in your browser.
The text is read one code point at a time — not one UTF-16 unit — so emoji and other characters outside the Basic Multilingual Plane are never split. Each code point is checked against a table of characters that are invisible or easy to mistake for something else: zero-width characters (U+200B, U+200C, U+200D, U+2060, the U+FEFF byte-order mark), unusual spaces (no-break, figure, thin, hair, em and ideographic spaces, line and paragraph separators), bidirectional controls (U+200E/F marks, U+202A–U+202E embeddings and overrides, U+2066–U+2069 isolates), C0 and C1 control characters other than tab and line breaks, other format characters (soft hyphen, invisible math operators, Hangul fillers, variation and tag characters, anything in Unicode category Cf) and typographic substitutes (curly quotes, en and em dashes, the ellipsis, minus and prime signs). Joiners and variation selectors that sit inside an emoji sequence are marked as such, because removing them splits 👨💻 into separate pictures. The text is redrawn with a labelled marker for every hit, and a summary table lists each character's code point, name, type and count. Bidirectional overrides and isolates trigger a Trojan Source warning (CVE-2021-42574). Cleaning removes the selected categories, turning odd spaces into ordinary spaces and smart punctuation into ASCII.