@RyanM You are correct in your belief that the start and end of the string, in the regex version we use, is not, necessarily, considered to be a word boundary. In order to accommodate that, and account for testing for the beginning and end of the string, the watched and blacklisted keywords use more complex bookending, which is
(?is)(?:^|\b|(?w:\b))(?:{})(?:\b|(?w:\b)|$)
. Please see
PR #2322 for a bit more detail.