top-journals-methods-scout

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly fetches and ingests public third-party paper metadata and abstracts from the OpenAlex API (see SKILL.md "第二步：OpenAlex 抓取" and scripts/fetch_journal_papers.py), and the main agent reads and interprets those titles/abstracts/keywords to generate analyses, teaching recommendations, and drive further LLM-based encoding — so untrusted external content can directly influence agent decisions and tool use.