research-scout

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

• Third-party content exposure detected (medium risk: 0.30). This skill’s required runtime workflow performs research “search” across outsider-authored content (e.g., it generates queries and then extracts quotes/idea text from paper/blog/newsletter pages) so a queue/feed-style outsider submission can poison the ingested text before the agent selects anything specific.