Testing purpose
Match the generated form to the component or dataset under test
Names help exercise lists and forms, short phrases fit labels and cards, sentences test wrapping, and paragraphs test long content. Usernames and hashtags need character and length rules from the destination. Generate empty, short, long, duplicate, and Unicode cases separately because a random average batch will not cover every boundary.
Synthetic identity
Generated names and handles are examples, not verified identities or available accounts
An output can accidentally match a real person, company, trademark, username, or offensive phrase. Do not attach private, financial, medical, or reputation-affecting attributes to synthetic identities. Check public-facing samples manually and label demo datasets clearly.
Randomness limits
Uniqueness in one batch does not guarantee global uniqueness or security
Prefer-unique mode can avoid obvious repeats within the generated set, but a finite vocabulary will still collide over time. Do not use a creative text generator for passwords, API tokens, identifiers, cryptographic keys, or statistical simulations that require a defined distribution.