AI agent support: lightweight toxicity signal for user-generated text (insults, threats, harassment, profanity) with category-level scoring, for agents pre-filtering inbound messages. Local-only mini-lexicon, returns per-category match counts — coarse, indicative-only.

Payment Options