How to use
What it is
Paste a text before sending it to an AI, publishing it or using it in code, and the tool shows you what you cannot see: instructions hidden with invisible characters, words split to dodge filters, code that reads differently from how it runs, and fake letters from another alphabet. Everything happens in your browser: the text is neither sent nor saved.
Why it matters with AIs
A language model receives the characters, not what you see on screen. If a document, an email or a rules file (.cursorrules, AGENTS.md…) contains a sentence written with invisible characters, the AI may read it and obey it. Here it is decoded and shown as is.
What is NOT flagged
Several of these characters have legitimate uses and are not treated as suspicious: the one that joins the emoji of a family, the one that turns the text heart into its emoji version, the flags of Scotland or Wales (which are made of invisible tags), the one Persian uses between letters, or the direction marks of an Arabic or Hebrew text. They show up in the view but in grey, and are kept when cleaning.
Cleaning
The clean text removes everything invisible and suspicious. Letters from another alphabet are not changed: there is no safe way to know whether a «с» was meant to be a «c» or is a Russian word, so they are flagged and you decide.
