cat-linguistic-style-matching

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

• Third-party content exposure detected (high risk: 0.75). The skill’s workflow ingests the target user’s and community members’ post text to compute function-word frequencies and baselines; if those community posts include outsider-authored free text (i.e., not written by the operating user), that text is fed into the LLM context via the runtime analysis/reporting path.