Agency

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

• Third-party content exposure detected (high risk: 0.75). The required workflow ingests “client brief and preferences (from client file)” and “client messages/chaotic feedback” into the agent’s LLM context; since those client-provided texts are authored by an outsider (the client) and are not chosen by the operating user to be introduced as their own content, this creates an indirect prompt-injection surface via client-authored free text.