opinion-miner
Warn
Audited by Snyk on Apr 1, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 1.00). SKILL.md explicitly instructs the agent to fetch and scrape user-generated content from public third-party sites (Bilibili, Reddit, GitHub Issues) using webfetch/agent-browser in the "支持的数据来源" and "步骤 1: 爬取评论" sections, and mandates reading and using those comments to drive clustering/decisions, which could enable indirect prompt injection.
Issues (1)
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
Audit Metadata