playtest-report

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

• Third-party content exposure detected (high risk: 0.70). In “analyze [path]” mode, the workflow reads “raw notes at the provided path” (outsider-authored free-form notes from an arbitrary user-supplied path) and then fills the LLM prompt/template with those observations, creating an indirect prompt-injection path via the notes’ text.