Why letter case is a real problem, not a cosmetic one
Two case failures I keep running into, years apart. First: a release email with the subject typed in ALL CAPS, which every mail client treats as shouting - open rates droop, and the fix is sentence case, not a rewrite. Second: a refactor that renamed forty snake_case columns to camelCase, where one missed letter means a property that silently reads as undefined and fails three screens away from the cause. Both problems are one paste away from being solved here.
What makes a converter hard is not uppercasing - that is one method call. It is splitting words back out once case and separators have done their job. XMLHttpRequest is three words to a human and a wall of letters to a naive splitter. Runners of capitals are the trap: XML, HTTP, JSON are single words, so the parser has to know that the capital starting "Parser" belongs to the next word, not the acronym. Get it wrong and you ship x_m_l_http_request.
This tool's tokenizer treats a run of capitals as one acronym and hands the final capital to the next word when lowercase letters follow it. So XMLParser lands as xml_parser, getJSON as get_json, parseJSONData as parse_json_data. It also accepts mixed input - snake, kebab, spaces and camel all at once - because real-world identifiers are never one convention.
The writer's four
Sentence case capitalizes only after a sentence ends - period, question mark, exclamation, or a newline - which is the register people actually read email and documentation in. lowercase and UPPERCASE are the workhorses: normalizing data before a comparison, or feeding the punchline of a headline. Title Case capitalizes content words and lowercases the small ones - a, an, the, of, and, or, but, in, on, at, to, for, and friends - while keeping the first word of the line capitalized, which is what most style guides actually ask for. (Chicago and AP disagree at the margins; no converter settles that argument.)
The developer's five
camelCase for JavaScript variables and function names; PascalCase for classes, types and components; snake_case for Python and SQL column names; kebab-case for CSS classes, HTML attributes and URL slugs; CONSTANT_CASE for environment variables and true constants. The differences sound pedantic until a lint rule rejects your pull request over one of them, or a teammate greps for user_data and finds nothing because the codebase says userData.
One honest pattern to know: camel, snake and kebab outputs are single identifiers - spaces and punctuation between words collapse into the joiner, because that is what an identifier is. Paste a sentence into snake_case and you get the_quick_brown_fox, which is what you wanted for a slug and definitely not a sentence anymore. The writer's four keep all spacing intact. Nothing is judged, everything is shown before you copy it.
Limits, stated plainly
Software cannot read minds about proper nouns. Sentence case folds NASA to nasa and Title Case lifts iphone to Iphone, because the machine has no dictionary of names worth trusting at this scale - the honest move is to let you touch those up by hand. Chinese, Japanese and Korean text carries no case at all, so it passes through untouched (this tool is built for English text; pure Chinese input returns unchanged). And the input cap is 1MB, about half a million English words, set so a pasted novel cannot freeze the tab rather than for any deep technical reason.