Zero-Width Character Detector / Decoder
A zero-width character detector finds the invisible Unicode characters — zero width spaces, joiners, direction marks — that can sit inside ordinary-looking text, shows exactly where they are, and decodes any message they carry using the schemes steganography tools actually use, entirely in your browser.
- 330k text, base 4 (ZWNJ ZWJ PDF BOM)Best match
the flag is in plain sight
- binary bytes, ZWJ = 0, PDF = 1
UJ�9L�L
- 330k binary, base 4 (ZWNJ ZWJ PDF BOM)
t h e f l a g i s i n p l a i n s i g h t
- binary bytes, BOM = 0, PDF = 1
n{^� - binary bytes, ZWJ = 0, BOM = 1
@!�,
- binary per character, PDF = 0, ZWJ = 1, BOM between
斪-&ŦM
NZWNJZWNJZWNJZWNJZWJBOMZWJZWNJZWNJZWNJZWNJZWNJZWJPDFPDFZWNJZWNJZWNJZWNJZWNJZWJPDFZWJZWJZWNJZWNJZWNJZWNJZWNJPDFZWNJZWNJZWNJZWNJZWNJZWNJZWJPDFZWJPDFZWNJZWNJZWNJZWNJZWJPDFBOMZWNJZWNJZWNJZWNJZWNJZWJPDFZWNJZWJZWNJZWNJZWNJZWNJZWJPDFZWJBOMZWNJZWNJZWNJZWNJZWNJPDFZWNJZWNJZWNJZWNJZWNJZWNJZWJPDFPDFZWJZWNJZWNJZWNJZWNJZWJBOMZWNJBOMZWNJZWNJZWNJZWNJZWNJPDFZWNJZWNJZWNJZWNJZWNJZWNJZWJPDFPDFZWJZWNJZWNJZWNJZWNJZWJPDFBOMPDFZWNJZWNJZWNJZWNJZWNJPDFZWNJZWNJZWNJZWNJZWNJZWNJZWJBOMZWNJZWNJZWNJZWNJZWNJZWNJZWJPDFBOMZWNJZWNJZWNJZWNJZWNJZWJPDFZWNJZWJZWNJZWNJZWNJZWNJZWJPDFPDFZWJZWNJZWNJZWNJZWNJZWJPDFBOMPDFZWNJZWNJZWNJZWNJZWNJPDFZWNJZWNJZWNJZWNJZWNJZWNJZWJBOMZWNJBOMZWNJZWNJZWNJZWNJZWJPDFPDFZWJZWNJZWNJZWNJZWNJZWJPDFZWJBOMZWNJZWNJZWNJZWNJZWJPDFPDFZWNJZWNJZWNJZWNJZWNJZWJBOMZWJZWNJothing to see here, just a perfectly ordinary sentence.
About Zero-Width Character Detector / Decoder
Unicode has a family of characters that render as nothing at all. Most have real jobs: the zero width joiner glues 👨👩👧 into one family emoji, the non-joiner controls letter shapes in Persian and Indic scripts, and direction marks keep mixed Arabic and English in order. But because they're invisible, a sentence can carry hundreds of them without looking any different — which makes them a favourite of CTF challenges, a way to fingerprint which copy of a document leaked, and the reason two identical-looking usernames can be different strings. This tool shows them first: how many there are, which ones, and where each sits, marked inline, plus the text with all of them removed. Then it tries to decode them. Most zero-width puzzles are made with 330k's Unicode Steganography tool, which writes each character of the message in base N using N chosen invisible characters as digits, so the decoder reproduces that scheme exactly, including its choice of digit order. It also tries plain binary — one character for 0 and another for 1, eight bits per byte of UTF-8 — in both polarities, and binary with a third character separating one letter from the next. The readings are ranked by how English they look. Going the other way, it hides a message in a cover text either way: 330k-compatible, so that tool's decoder can read it, or as binary with zero width spaces and non-joiners. The result looks identical to the cover. Some platforms strip invisible characters when you paste, so it's worth checking a hidden message survives wherever you're putting it — and if you're receiving text rather than sending it, the visible-text-only copy is a quick way to clean it.
What Zero-Width Character Detector / Decoder does
- Counts and marks every invisible character: zero width spaces and joiners, direction marks, word joiners, BOM and more
- Decodes 330k Unicode Steganography output exactly, in text and binary modes
- Tries binary schemes in both polarities, with or without a separator character
- Ranks readings by how English they look
- Strips every invisible character, leaving only the visible text
- Hides a message in a cover text, 330k-compatible or as binary
When to reach for Zero-Width Character Detector / Decoder
- Solving a CTF challenge whose text is suspiciously long when you count characters
- Checking a document or message for invisible fingerprinting before you share it
- Finding out why two strings that look identical don't compare equal
- Cleaning invisible characters out of pasted text
How to use Zero-Width Character Detector / Decoder
- 01
Paste the text
The count and the inline markers show at once whether anything is hidden, and where.
- 02
Read the decoded message
330k's scheme and the binary schemes are all tried, most English-looking first.
- 03
Clean or keep
Copy the visible-text-only version to strip every invisible character.
- 04
Hide your own
Switch to Hide a message, write a cover text and a secret, and copy the result.
When to use Zero-Width Character Detector / Decoder vs alternatives
| Alternative | Use Zero-Width Character Detector / Decoder when… | Use the alternative when… |
|---|---|---|
| 330k's Unicode Steganography tool | you don't know which characters or settings were used, or the message isn't from 330k at all. | you know the exact character set and want its finer options for encoding. |
| Searching for \u200b in an editor | you need every kind of invisible character found, not just one, and the message decoded. | you're cleaning a file you're already editing. |
Frequently asked questions
How do I find zero-width characters in text?
How do I decode a hidden zero-width message?
Are zero-width characters always suspicious?
Can invisible characters reveal who leaked a document?
Will a zero-width message survive being pasted?
Related tools
Related concepts
Homograph attack
A homograph attack uses characters that look identical to others — Cyrillic а in place of Latin a — so that a different domain, username or identifier appears to be a trusted one. Browsers defend domain names by showing the xn-- form of names that look suspicious instead of the Unicode spelling.
Unicode normalization
Unicode normalization rewrites text into one standard sequence of code points, so strings that look identical — é stored as a single precomposed character, or as e followed by a combining accent — compare as equal. Unicode defines four normalization forms: NFC, NFD, NFKC and NFKD.