What it does
POST /api/v1/safety/nickname reads a name the way a moderator would: part by part, in its language and in the setting where it is shown. It returns a signal about the name, not a verdict on the person, the categories it fell in, a short reason, and a recommendation. Your platform decides what to do.
A name is judged by what it says, not by a word list. The same letters can be a surname, a year, a place or a team, and a name the reading cannot place returns none rather than a guess.
Request
Response
Signals and recommendations
recommendation is one of allow, allow_and_watch, ask_to_rename or hold_for_review. It is never an emergency action: a name alone cannot carry that.
Categories
categories comes from a closed list and is empty unless the signal is review:
self_harm_promotionimpersonation_of_staff_or_supporthate_or_extremist_referencesexualisedseeks_older_contactharassment_of_named_personcontact_detail_or_off_platform_lureage_disclosure
rationale is written for a moderator and never reproduces a slur.
When the reading does not finish
If the reading cannot complete, the answer issignal: "none" with reading_complete: false and confidence: 0. That means nothing was analysed, not that nothing was found, so retry. An incomplete reading is not charged (credits_used: 0).