Line separator guide · November 1, 2026
Does Claude Text Cleaner Remove Line Separators?
No. The browser cleaner leaves U+2028 LINE SEPARATOR and U+2029 PARAGRAPH SEPARATOR in pasted text. In live tests, A + U+2028 + B and A + U+2029 + B each stayed three code points with zero removals. By contrast, A + U+200B + B became AB, with one removal. A zero-removal report only describes this cleaner's supported removal rules.
A | U+2028 | BUnchanged · 3 → 3 · removed 0A | U+2029 | BUnchanged · 3 → 3 · removed 0A | U+200B | B → ABChanged · 3 → 2 · removed 1Why can a line break remain after cleaning?
The browser script checks pasted text against fixed removal categories. U+2028 and U+2029 are outside those categories, so the cleaner copies them unchanged. U+200B is in a supported zero-width category. Unicode's line-breaking reference identifies U+2028 and U+2029 as line and paragraph separators. Their display and parsing can vary by destination.
An English-language public question describes trouble identifying a separator in a textarea; a commenter identifies U+2028. Another question asks how U+2028 relates to a rendered line break. Neither discussion reports a test of this product.
What should you do with the separator?
- Keep a copy of the original. Paste into the cleaner and compare its output and removal counts.
- Inspect the destination input. A developer can list code points with
Array.from(text, c => 'U+' + c.codePointAt(0).toString(16).toUpperCase()). This distinguishes U+2028 or U+2029 from an ordinary newline or U+200B. - Choose the destination-specific treatment. Preserve, replace, or escape the separator according to the receiving editor, file format, or parser. This cleaner does not repair arbitrary JSON or code.
The live browser tests were recorded October 31, 2026. The cleaner cannot establish where text came from, detect a watermark, or predict an AI-detector outcome.