A visible report for invisible problems
A word can look correct while containing a zero-width space that breaks a search. Two spaces can look identical while only one allows a line break. The cleaner reports character names and code points so you can inspect the cause instead of guessing from appearance.
Detection examines the original input before normalization. Each result lists supported characters that were removed, replaced, or kept, along with counts. The output preview lets you review the result before copying or downloading it.
Inspect the implementation or report an issue
The project source is available in the CleanPastedText GitHub repository. It includes the cleaning rules, automated tests, and the source for these guides. You can report a bug or suggest a correction on GitHub.
For a cleaning bug, share a short, non-sensitive example, the preset you used, and what you expected. A code point such as U+200B is more useful than a screenshot of an invisible character. Avoid posting private documents in a public issue.
What the results can tell you
A detection identifies a character; it does not identify its author or why it was inserted. Ordinary documents, languages, and emoji use many of the same characters. An empty report means the supported checks found no matches, not that the text has passed an exhaustive security review or an AI authorship test.
Review output carefully for multilingual text and code. Unicode normalization can change compatibility characters, and removing join controls can affect some writing systems. The cleaning methodology explains the settings and limitations. The privacy and data handling page explains what is processed locally and how site analytics work.