llm-public-opinion-analytics-assistant

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly describes crawling public, user-generated sources ("crawls hot search rankings from 15 platforms including Weibo, Bilibili, Douyin") and contains examples (e.g., DetailCrawler.extract_content and OpinionAnalyzer.analyze_sentiment) showing that scraped third-party webpages/posts are ingested and fed to the LLM for analysis and alerting, which could allow indirect prompt injection from untrusted web content.