Zero-width format controls · local cleanup

Zero width space remover

Detect, highlight inline with exact U+XXXX badges, and strip hidden U+200B ZWSP, U+200C ZWNJ, standalone U+200D ZWJ, U+FEFF BOM, and U+2060 Word Joiner characters in your browser—without breaking valid emoji sequences.

Physical bytes vs. statistical signals

Zero-width cleanup is deterministic, free, and 100% local. If you also need authorized wording changes across long drafts or automated pipelines, sentence-level rewriting uses your account character balance on the pricing page.

Browser-local zero-width scanner

Inspect and strip hidden zero-width characters

100% browser-local scan
Physical Unicode scan12 anomaly code points flagged
Statistical token signalstatistical/vendor watermark unknown
Vendor verifier statusvendor verification unavailable

Detected codepoints in current buffer:

U+200B ZERO WIDTH SPACE (ZWSP) × 6U+2060 WORD JOINER × 3U+FEFF BYTE ORDER MARK / ZWNBSP × 1U+200C ZERO WIDTH NON-JOINER (ZWNJ) × 1U+200D ZERO WIDTH JOINER (ZWJ) × 1
Inline codepoint highlight map
[U+FEFF]Quarterly product brief:[U+200B] launch[U+200B] readiness[U+200B] is on[U+2060] track[U+2060] for Friday.[U+200C] Review[U+200B] the API[U+200D] schema[U+200B] before[U+2060] merging[U+200B] to main. Team check-in ❤️ stays at 10:00.

Need deeper wording changes or high-volume API cleanup?

Removing invisible Unicode strips literal hidden codepoints, but it never alters statistical token patterns in the visible wording. Run an optional authorized sentence-level rewrite below (numbers, URLs, emails, code spans, and quotes are placeholder-protected), or compare Pro, Scale bearer-API, and one-time character packs.

Checking your session…

Codepoint reference

Zero-width codepoints detected and stripped on this page

Each character below occupies zero horizontal pixels in standard UI fonts while adding three UTF-8 bytes per occurrence.

Hex CodepointUnicode NameHow It Enters LLM / Pasted TextEditor, ATS & Code Breakage Impact
U+200BZERO WIDTH SPACE (ZWSP)Inserted by rich-text web editors, chat UIs, markdown renderers, and scraped HTML soft-wrap boundaries.Breaks exact string comparisons, SQL/JSON keys, URL slugs, ATS resume keyword matching, and regex word boundaries.
U+FEFFZERO WIDTH NO-BREAK SPACE / BOMPrepended by UTF-8/UTF-16 file exports, clipboard wrappers, or concatenated document chunks mid-string.Causes JSON.parse syntax errors at byte 0, breaks CSV column headers, and corrupts shell shebang lines.
U+2060WORD JOINER (WJ)Added by typesetting pipelines and PDF-to-text extractors to glue punctuation to preceding words.Silently alters string byte length (3 UTF-8 bytes: E2 80 A0) and defeats full-text search tokenization.
U+200CZERO WIDTH NON-JOINER (ZWNJ)Legitimate in Persian, Arabic, and Indic script shaping; stray in English pastes from OCR or web templates.Splits English identifiers into two tokens in compilers and search indexes while remaining invisible.
U+200DZERO WIDTH JOINER (ZWJ)Required inside multi-person or profession emoji sequences; stray when copied outside pictograph pairs.Inflates character counts and breaks strict ASCII/identifier validators when pasted outside emoji.
Technical FAQ

Frequently asked questions

Does removing U+200B and U+FEFF send my text to a server?

No. The zero-width scan, inline badge map, and cleanup run entirely in your browser tab using JavaScript codepoint iteration. Text only leaves your browser if you explicitly confirm ownership and trigger an optional sentence-level rewrite.

Will stripping zero-width characters break multi-part emoji like family or heart icons?

Our forensic engine checks adjacent pictographs before flagging U+200D (Zero Width Joiner) or U+FE0F (Variation Selector-16). Emoji joiner sequences remain intact while standalone U+200D, U+200B, U+2060, and U+FEFF inside plain prose are stripped.

Why did my JSON parser or compiler fail on text that looks completely normal?

A single U+200B (UTF-8 bytes E2 80 8B) or mid-string U+FEFF (UTF-8 bytes EF BB BF) inside a property name or variable identifier has zero rendered width in most fonts, yet parsers treat it as an illegal identifier byte.

Does removing zero-width spaces remove a statistical AI text watermark?

No. Stripping zero-width characters removes literal hidden Unicode codepoints only. Statistical LLM watermarks live in the visible token choices rather than hidden bytes, and cannot be verified without the model vendor’s private detector.