ai-detect

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs the agent to "Read the paper the user points you to" and to "use web search to verify" every citation, which requires fetching and interpreting open/public third-party content (user-provided papers and web search results), exposing the agent to indirect prompt injection.