RespectShield
Set the terms of the conversation.
RespectShield is a moderation prototype built around creator-defined boundaries and human review.
What the prototype checks
Deterministic rules flag possible threats, harassment, hate speech, spam, repeated unwanted requests, and creator-specific boundary violations. A mock classifier demonstrates where a production model could be added.
How context is handled
Explicit, agreed preferences can inform screening for consensual roleplay. A revoked boundary or a serious safety signal still calls for review. The prototype uses examples and cannot reliably interpret every conversation.
What’s planned
Live message screening, warnings, restrictions, a staffed review queue, and appeals are planned. Permanent bans require a human decision.
Join the creator waitlistTry the prototype
Run a sample through the rules. This demo runs in your browser.