Zero-width spaces, joiners, soft hyphens, directional marks and Unicode tag characters render as nothing and survive every copy and paste. They break searches, corrupt diffs, fingerprint documents, and can carry text that only a machine will ever read. StripShot shows you where they are and removes them, while leaving emoji and complex scripts intact.
Last updated September 2026
Paste text below and click Scan. StripShot lists every invisible character by code point, highlights where each one sits, and removes them on Strip. Emoji sequences, flags and the joiners that Arabic, Persian and Indic scripts need are preserved.
Scans are always free. Cleaning counts toward the daily free strips; Pro is unlimited.
| Character | Code point | Why it matters | Action |
|---|---|---|---|
| Zero-width space | U+200B | Invisible word break, the classic fingerprinting carrier | Removed |
| Zero-width non-joiner and joiner | U+200C, U+200D | Needed inside emoji sequences and some scripts, otherwise invisible padding | Removed when free-floating, kept as glue |
| Word joiner | U+2060 | Invisible, prevents a line break | Removed |
| Byte order mark | U+FEFF | Leaks from files into pasted text | Removed |
| Soft hyphen | U+00AD | Invisible unless a line breaks there | Removed |
| Directional marks and overrides | U+200E, U+200F, U+202A to U+202E, U+2066 to U+2069 | Can reorder or hide text; used in filename spoofing | Removed |
| Variation selectors | U+FE00 to U+FE0F, U+E0100 to U+E01EF | Glyph selectors, meaningful after emoji and CJK, otherwise invisible | Kept after a base, removed alone |
| Unicode tag characters | U+E0000 to U+E007F | Invisible ASCII look-alikes used to smuggle text past human readers | Removed except inside flag sequences |
| Hangul and Khmer fillers, Mongolian selectors | U+115F, U+1160, U+3164, U+17B4, U+180B to U+180F | Invisible unless next to their own script | Kept in context, removed alone |
| Non-breaking and typographic spaces | U+00A0, U+2000 to U+200A, U+202F, U+205F, U+3000 | Look like spaces, break search and matching | Replaced with a normal space |
| Look-alike letters | Cyrillic а е о р с у х, Greek ο α, full-width A to z | Words that look right but do not match | Replaced with the Latin letter (option) |
A naive cleaner that deletes every zero-width joiner turns a family emoji into four separate people and breaks Persian words in half. StripShot checks the neighbours of each invisible character first: a joiner between two emoji, or between two letters of the same joining script, is kept. A joiner floating between two Latin letters is a carrier and goes. The rules are adapted from the MIT-licensed watermarks-remover project and tested against emoji, flags, Arabic, Devanagari, Khmer and Mongolian.
Removing invisible characters is deterministic and verifiable: the tool tells you what it removed and the re-scan proves it. It is not an AI detector and it is not an AI-detection bypass. Statistical text watermarks live in word choice, not in characters, and character cleaning does not touch them. The AI text watermark guide explains that boundary in detail.
Paste the text
Paste into the box above. The word count shows how many hidden characters it already sees.
Scan
Click Scan text. Each kind of invisible character is listed with its code point and count, and the highlighted view shows exactly where each one sits.
Strip
Click Strip. Free-floating invisibles are removed, unusual spaces become normal spaces, and, if the options are on, look-alike letters and typographic punctuation are normalized.
Copy
The cleaned text is re-scanned and shown. Copy it with one click.
Code points that render as nothing, or as an ordinary space, but still sit in the text. Zero-width space (U+200B), zero-width non-joiner and joiner (U+200C, U+200D), word joiner (U+2060), soft hyphen (U+00AD), the byte order mark (U+FEFF), directional marks and overrides (U+200E, U+200F, U+202A to U+202E, U+2066 to U+2069), variation selectors, and the Unicode tag block (U+E0000 to U+E007F) are the common ones.
Copy and paste from web pages, PDFs and Word documents; chat interfaces that emit formatting characters; text that was deliberately fingerprinted so a leaked copy can be traced; and, increasingly, hidden instructions smuggled into text with tag characters that people cannot see but language models can read.
No. The cleaner keeps the zero-width joiners that hold emoji families and flag sequences together, the variation selector that makes ✔ render as an emoji, and the joiners that Persian, Arabic, Hindi, Khmer and Mongolian text needs to shape correctly. Only free-floating invisibles are removed. Turn off Keep emoji and script joiners in Options if you want everything gone.
Only if you ask it to. Normalize punctuation swaps em dashes, curly quotes and the ellipsis character for their keyboard equivalents, and Fix look-alike letters replaces Cyrillic and Greek stand-ins with the Latin letters they imitate. Both are options you can turn off. Invisible characters and unusual spaces are always cleaned.
Yes for the invisible characters, which have no business in source code and break diffs and searches. Turn off Normalize punctuation before cleaning code so string literals keep their exact quotes.
Scan first. The result card shows your text with every hidden character marked as its code point in red and every look-alike letter in yellow, before anything is changed.
No. Scanning and cleaning run inside your browser tab. StripShot never sends text to a server or to a language model.
Hidden characters and smart punctuation in pasted output.
What can and cannot be removed, honestly.
The same cleaner applied inside Word files.
Every format on one page.
Pricing
No card needed to start. Upgrade anytime.