surprise-me

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

• Third-party content exposure detected (high risk: 0.80). The workflow explicitly instructs the agent to "search today's news" (public third-party web content) in the required fallback and also to "read the full SKILL.md body of each selected skill" and "follow each skill's instructions," meaning it ingests and acts on external/untrusted content that could influence actions and enable indirect prompt injection.