information-search-knowledge-management

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

• Third-party content exposure detected (high risk: 0.90). The skill's "多渠道搜索" explicitly ingests and analyzes public community content (Reddit, Hacker News, X/Twitter) and open tech media as part of its workflow to read, verify, and save findings, exposing the agent to untrusted user-generated third-party content that could carry indirect prompt injection.