med-entertainment

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime workflow ingests user-supplied entertainment materials (e.g., press kits, embargoed critic notes, transcripts, and other text the user provides) into the agent’s LLM context; if any of that text is outsider-authored (e.g., studio press materials, embargoed press reviews, or other people’s transcripts), it becomes free text in the LLM context via the user’s input path.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)