Prompt Shield Chat

A chatbot with a visible safety check that blocks jailbreaks and explains why.

How the shield works

Prompt check

Your message is screened for jailbreaks, harm and insecure code.

Chat reply

Only a safe prompt gets a chatbot reply.

Reply check

The reply is screened too before you see it.

Your message is sent to Muse Spark 1.3 to check it and write replies. This app does not keep it. History is saved on this device only.

Try a normal coding question, then try to trick the chatbot. Blocked messages show a category and a plain reason.

Your message0 of 2000 characters. Messages up to 2000 characters.

Safety checks are best effort. They can miss clever jailbreaks or block a harmless prompt by mistake.
Code security notes are a helpful explanation, not a certified security scan.
For teams that need full control, open protection models can be self hosted instead of using this hosted check. No outside moderation service is used here.

Send a message above. A safe reply or a blocked notice with its reason will appear here.

Every prompt and reply check from this device. Use it to see why something was allowed or blocked.

No checks yet. Send a message and its prompt and reply checks will be listed here.

Log counts: 0 checks, 0 blocked, 0 safe.