Verify a Prompt

This endpoint accepts a text prompt, strips PII, and checks it for prompt injection, returning an injection score.

post/wall

Request body

user_idstring

The user ID of the user who is submitting the prompt. This is used to flag suspicious users

session_idstring

The session ID of the user who is submitting the prompt. This is used to flag suspicious sessions

promptstring required

The text prompt to be verified.

scan_piiboolean

Whether to scan for PII in the prompt.

xml_tagstring

The XML tag that is used to escape user input in your prompt (this may have been generated with keep).

check_badwordsboolean

Whether to scan for badwords in the prompt.

fast_checkboolean

Whether to perform a fast check on the prompt instead of a full check.

Successful operation.

contains_piiboolean

Whether the prompt contains PII.

potential_jailbreakboolean

Whether the prompt contains a potential jailbreak.

potential_xml_escapingboolean

Whether the prompt contains potential XML escaping.

suspicious_userboolean

Whether the user is suspicious.

suspicious_sessionboolean

Whether the session is suspicious.

modified_promptstring

The prompt, modified in some way - e.g. by summarising it as per the configuration specified in the request.