Unicode inspector
Paste a text and get one row per character: code point, Unicode name, general category, block, and a flag when it is invisible, a control, a lookalike letter or a fullwidth form. Totals in code points, UTF-16 units, UTF-8 bytes and graphemes.
Nothing is sent to a server: the text stays in your browser, the tool runs offline once the page is loaded. Ctrl+Enter runs it.
- Names come from a built-in table of about 4,700 characters (Latin, Greek, Cyrillic, punctuation, spaces, arrows, symbols, emoji blocks) plus rules for fullwidth, mathematical, CJK, Hangul, private use and tag characters
- General category is computed by the browser (Unicode property escapes), so it follows the browser's Unicode version
- Flags: invisible or format character, control, lookalike (Cyrillic or Greek letter that mimics a Latin one), fullwidth or mathematical letter
- Graphemes are counted with the browser's segmenter when available, so an emoji family counts as one
- Shows the first 3,000 rows; totals cover the whole text
What each row shows
A text is a sequence of code points; what you see on screen are graphemes, and what a file stores are bytes. Those three counts differ as soon as accents, emoji or invisible characters are involved. The inspector lays the sequence out one code point per row:
- Char: the character itself, or a short token (ZWSP, NBSP, SHY, LRM) when it has no visible shape.
- Code point: U+ followed by four to six hexadecimal digits.
- Name: the Unicode character name (LATIN SMALL LETTER A, ZERO WIDTH SPACE, CYRILLIC SMALL LETTER O).
- Category: the general category (letter, mark, number, punctuation, symbol, separator, format, control) as computed by your browser.
- Block: the Unicode block (Basic Latin, Latin-1 Supplement, General Punctuation, Cyrillic, Mathematical Alphanumeric Symbols).
- Flags: invisible, control, lookalike, fullwidth or mathematical.
Typical uses
Find out why two strings that look identical do not match (a Cyrillic о in one, a Latin o in the other). Find the character that breaks a JSON file or a CSV import. Understand why a tweet is longer than it looks (combining marks, variation selectors). Check what a copied product code really contains. Identify a symbol you cannot type. Verify that a text sent to a form field contains only ASCII when the field requires it: tick Hide plain ASCII and anything left is what you need to fix.
How to use it
- Paste the text (or a single character) and click Inspect.
- Tick Flagged characters only to keep the rows worth looking at, or Hide plain ASCII to see only what is outside the 7-bit range.
- Copy or download the report as tab-separated text.
Fixing what you found
The inspector shows; it does not change anything. To remove the invisible and lookalike characters it flags, use the invisible character remover, which uses the same tables and lets you choose category by category. To clean AI output in one pass, including Markdown and dashes, the AI text cleaner does it.
None of this touches a statistical watermark: that mark lives in the choice of words and survives any character clean-up. If that is your question, the AI text watermark remover hub explains which tool applies.
Privacy
The inspection runs in your browser; your text is not sent, stored or logged.
Unmarker is an independent product, not affiliated with or endorsed by Anthropic.
Questions people ask
What is the difference between code points, UTF-16 units and graphemes?
A code point is one Unicode character. JavaScript strings count UTF-16 units, so emoji and characters above U+FFFF count as two. A grapheme is what a reader sees as one character: an e with an accent written as two code points is one grapheme, and so is a family emoji made of several code points joined by U+200D.
Why is a name missing for some characters?
The built-in table covers about 4,700 common characters plus rules for large regular blocks (CJK, Hangul, mathematical, fullwidth, private use). Rare characters outside those show their block and category but no name.
What does the lookalike flag mean?
The character is a Cyrillic or Greek letter that looks like a Latin one (a, e, o, p, c, x, y, and others). Inside a Latin word this is a classic way to defeat searches and filters, or a copy-paste accident.
Can I inspect a single character?
Yes. Paste just that character; the row gives its code point, name, block and category.
Is the text uploaded?
No. The inspector is a script that runs in your browser and works offline once the page is loaded.