openakita/skills@baidu-scholar

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

• Third-party content exposure detected (high risk: 0.75). Runtime path: the skill calls run_skill_query(args, query) with a free-form Chinese query, and the Baidu Scholar/AppBuilder response (paper titles/authors/abstracts/citation lists) is returned as readable text into the agent’s LLM context; this content is authored by external parties (public academic sources), creating indirect prompt-injection exposure.