Does it detect the input convention automatically?
Yes. The input is split into words by detecting camelCase boundaries, underscores, hyphens and dots, so you can paste snake_case, kebab-case, PascalCase or plain prose without telling it which. Acronym runs are handled too: getHTTPResponseCode splits as get / HTTP / Response / Code rather than mangling the acronym.
Why is Title Case not just capitalising every word?
Because that is not title case. AP style keeps minor words — a, an, the, and, of, to, in — lowercase unless they are the first or last word. "The Quick Brown Fox Jumps Over the Lazy Dog" is title case; capitalising "the" in the middle is not.
What happens with multi-line input?
Identifier conventions (camel, snake, kebab, and so on) are applied line by line, because an identifier is a single token — applying camelCase to a whole paragraph would silently delete every space. Prose conventions (Title, Sentence, UPPER, lower) apply to the full text.