like petnames, but funkier
Readable handles and higher-entropy codes from six curated pools — and, unusually, the arithmetic that says what they're worth. A three-word name feels about as unguessable as a password. It isn't, and the gap is where the bugs live. Source on GitHub.
A handle is three words in a fixed order — a descriptor, then a colour or action, then a place or thing — drawn from six pools curated for how they sound together. A code is the other shape: flat draws from every pool at once, any word in any position, more bits per word. Both are generated in your browser by the same library the package ships; no name you generate leaves this page — the only thing it fetches is the typeface.
Three structured slots — descriptor, colour or action, place or thing. Reads like a name. No word repeats.
Flat draws from every pool at once. Any word, any position — worth more bits per word, reads less like a name.
Turn pools off, pick how many words to analyse, then set the attack you're worried about. (Word length is set once, at the top of the page, and applies here too.) The last slider is the one people forget: an attacker doesn't want your name, they want any live one — so a big keyspace gets diluted by a big population.
Same settings as Take it apart. Pick what you'd use a code for and you get a yes or a no. Each use has a bar: the number of possible codes it needs before guessing stops being worth an attacker's time. The codes above are only examples; this panel measures the recipe, not any one code.
A bit is one coin-flip of randomness, and every extra bit doubles the number of possible codes — 10 bits is about a thousand, 20 about a million, 30 about a billion.