Invisible / Zero-Width Character Detector

Paste text and the tool highlights every whitespace and invisible Unicode character: regular spaces as middle-dots, tabs as arrows, newlines as pilcrows, NBSPs in amber, zero-width characters in red, and bidi / formatting marks in fuchsia. A summary lists every invisible code point it found, and a one-click clean removes them — handy for fixing broken JSON, mangled passwords, or copy-pasted code that won't compile.

The non-breaking space is the one that ruins a day, because it looks exactly like a space and is a different character. Word processors, chat clients and some websites insert it automatically, so text copied out of a document arrives with them scattered through it, and a parser that splits on whitespace does not recognise it. The symptom is a config file that fails on a line you have read twenty times, or a password that is rejected while appearing correct in every field.

Zero-width characters are worse because they leave no trace at all. The zero-width space, joiner and non-joiner have legitimate uses in scripts that need them, and they also survive a copy out of a rendered page where they were inserted for line-breaking. A string that compares unequal to an identical-looking one usually has one of these in it, and no amount of squinting will show it.

Bidirectional control marks are the class worth treating with suspicion in source code. They reorder how text is displayed without changing what it says, so a line can render in one order and compile in another, which is the basis of the Trojan Source attack published in 2021. Compilers and code hosts now warn about them; seeing one in a file you did not expect to contain any is worth a pause.

The clean strips the invisible characters and leaves ordinary spaces and newlines alone. That is the right default for a password or a JSON key and the wrong one for text in a language that relies on these marks, so check the summary before cleaning rather than after.