[^ ] is actually excluding characters not in the specified range, rather than excluding the whole string the way you wanted it too.

You could check if it contains the string or not before you try to run pattern matching on it, but the most straightforward, though probably not the most obvious way to do it is to use token functions.

First, you can use Split and treat the whole "Marked for archive" string as delimiter. That will break the string apart and generate a multi-value token. For ex, "String MARKED FOR ARCHIVE test" becomes a token with 2 values "String " and " test". Then you want to trim the leading and trailing spaces so entries with just the beginning or end piece don't cause issues as WF renames documents. Then you can put the name back together with space as a delimiter using indexing.