Moderate text content using Claude Haiku — flags categories like harassment, hate, sexual content, violence, self-harm, and spam with per-category severity and an overall allow/flag/block recommendation, so agents can screen user-generated content before publishing or storing it.

Payment Options