newsapi

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly instructs the agent to call NewsAPI (via Membrane actions like "search-everything" / "get-top-headlines" and proxy requests to newsapi.org) to fetch public news articles and sources, so the agent will ingest untrusted, third-party content (news sites/blogs) as part of its workflow which could contain instructions that influence behavior.