Invisible & problem character detector
Text copied from Word, Google Docs, a PDF, or a chat app often carries characters you can't see: zero-width spaces, non-breaking spaces, lookalike letters from another alphabet. Paste below to reveal them, and get a cleaned copy ready to use.
Your text never leaves your browser.
Paste text above to scan it for invisible, non-standard, or lookalike characters.
Line endings: —
This text is largely written in a non-Latin script, so homoglyph detection has been skipped to avoid flagging every letter of a genuinely non-English document.
| Clean | Category | Found | Actions |
|---|
Your text, marked
Invisible characters have no width to highlight, so each one is shown as a[MARKER] chip you can select for details. Curly quotes and dashes keep their own shape since they're already visible.
Cleaned copy
Untick a category above to keep it as-is. Zero-width characters, control characters, and trailing whitespace are removed by default; smart punctuation and homoglyphs are left alone unless you opt in, since both are usually intentional.
What this tool catches
Every character class below is genuinely invisible, ambiguous, or misleading on screen — that's the selection criterion. A checker that flagged ordinary punctuation would be noise; this one only lights up for characters that can pass a visual review and still break a build, a diff, a CSV import, or a search.
Zero-width characters
Zero-width spaces, non-joiners, joiners, and word joiners have no visible glyph at all. They commonly survive a copy-paste from Word or Google Docs and then silently break exact-match searches, URL slugs, or string comparisons that look identical in every editor. The one exception: a zero-width joiner sitting between two emoji is holding a sequence like 👨👩👧 together, not hiding anything, so this tool leaves those alone.
Homoglyphs
A Cyrillic "а" or a Greek "ο" can be typed, pasted, and read as an ordinary Latin letter, which is exactly what makes them useful for spoofing a domain or an identifier past a visual review. This is the most security-relevant finding on the page — seehomoglyph attacks in text for how the attack actually works. Genuinely multilingual documents are not homoglyph problems, so this check is skipped automatically once a document is substantially written in a non-Latin script.
Non-standard spaces
A non-breaking space looks identical to a normal space but doesn't wrap the same way, which breaks layouts and confuses word counts. It's also sometimes intentional — French typographic convention uses one before certain punctuation, and "10 km" is a legitimate non-breaking pair — so this tool flags it as worth a look rather than something automatically wrong.
Control and directional characters
Soft hyphens and bidirectional text marks are usually harmless, sometimes even intentional. Directional overrides are the exception: they can make a file namedexe.pdf display as fdp.exe, a real technique used to disguise a malicious file extension.
Line endings and trailing whitespace
A document mixing CRLF and LF line endings shows every line as changed in a version-control diff, even when nothing meaningful changed. Trailing whitespace is invisible in almost every editor and a constant source of noisy diffs and failed exact-match comparisons.