asb-who-me

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). YES: the required workflow assigns “outside-in homework” to other people and then processes their returned free-text answers, which are outsider-authored content that the agent would read and include in the LLM context during “mine them exactly like prompt answers” / edge and tensions extraction.