A find-and-replace that runs entirely in the browser, with case matching and whole-word matching so a swap does not spill into longer words.
The classic find-and-replace mistake is replacing a short word and discovering it inside longer ones. Swap cat for dog and catalogue becomes dogalogue; swap is for was and this becomes thwas. Whole-word mode fixes this by requiring a word boundary on each side of the match, so only the standalone word is replaced. Turn it off when you genuinely want partial matches, changing a filename prefix across a list, or fixing a misspelling that appears inside compounds. Case matching is separate: leave it off to catch a term at the start of a sentence as well as mid-sentence, turn it on when The and the mean different things.
Everything you type in the find box is treated as literal text, not as a pattern. Characters that would mean something in a regular expression (a dot, a question mark, brackets, a backslash) are escaped before the search runs. This is a deliberate constraint. It means searching for a dot finds actual dots rather than matching every character, and it means no input you paste can construct a pattern that hangs the page. The trade-off is that you cannot do pattern work here, so a task like matching any four-digit number needs a different tool.
Whole-word mode was off, so the match ran on substrings. Turn it on and only the standalone word is replaced.
No, by design. The search is literal and regex characters are escaped, so a dot matches a dot. This keeps a pasted pattern from hanging the page.
Yes. Type the phrase in the find box and leave the replace box empty. Every match is removed.