Guardrails

Guardrails are the safety and abuse controls that run around every clean. They protect you, other users, and the service itself. This page explains what those controls are; the rules they enforce live in our Usage Policy.

Input limits

• We accept text prompts only; empty submissions are rejected.
• Each prompt is capped at your plan’s character limit (2,000 on Guest and Free, 10,000 on Pro).
• The mode and model you choose are validated before anything runs.

Content safety

Before a prompt is cleaned, it is automatically screened for content that violates our Usage Policy — such as illegal material, content that exploits minors, threats or harassment, and other clearly harmful content. If a prompt is flagged, the clean is refused and nothing is sent on for processing. The screening is automated and may occasionally flag legitimate content; if that happens, revise and try again, or contact us.

Abuse protection

• Rate limits cap how many requests you can make in a short window, per plan.
• Usage limits apply a fair-use allowance (rolling 24-hour windows for Guest and Free; credit-based for Pro).
• We store a hashed (not raw) IP address with each clean, solely to detect and prevent abuse.

Your part

• Don’t submit prohibited content — see the Usage Policy for the full list.
• Keep secrets and other people’s personal data out of your prompts.
• Always verify a cleaned prompt before relying on it — AI models can make mistakes.

Reporting

Seen something that shouldn’t be here, or think a prompt was blocked by mistake? Reach our support contact (the same one listed in our Terms and Privacy Policy).