devrel-content

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

• Third-party content exposure detected (medium risk: 0.30). The required workflow (“Phase 1: Research & Validation”) explicitly instructs the agent to search Google and “Search Reddit, HN, Stack Overflow” and also pull “Internal data” like “Support tickets, Discord questions, GitHub issues,” so outsider-authored free text can be ingested from these community/work-inbox sources during research.