slop-eval

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

• Third-party content exposure detected (high risk: 0.75). The required workflow’s “Gather evidence” step can fetch and screenshot a user-supplied live URL (public web content), and that fetched page content (rendered text/DOM-derived artifacts) is then used to produce the report that the agent’s LLM context would reference—an outsider-authored free-text source.