Security audit
Security checks for vulnerabilities and agentic risk
This is a simple course-guidance skill with some vague and overbroad wording, but no code, hidden behavior, persistence, credential use, or destructive capability.
Before installing, users should treat this as a Chinese-language course-selection helper and be aware that some described fields sound broader than a simple course guide. Do not provide account credentials or private education records unless a future version clearly explains why they are needed and how they are handled.
The natural-language content of the skill is entirely in Chinese and the description presents the skill as operating in that language, but there is no indication that users may choose another language or that the locale restriction is intentional and justified. Under the policy, language constraints should be opt-in or clearly documented as region-specific.
The stated purpose is to provide access guidance for NetEase/Bilibili open courses, but the screening items include pass rate, certificate value, and other evaluative criteria, while the update section references exam outlines, instructor teaching plans, and learning-effect feedback. These descriptions suggest a broader exam-preparation or training-advisory service rather than a narrowly scoped open-course guide.
The manifest description at L03 says the skill provides a guide for obtaining NetEase Open Courses/Bilibili knowledge-area courses. However, the documented return fields claim functionality such as mastery analysis, wrong-question notebook sync, lecture download progress, interactive Q&A status, mock exam transcripts, and degree-certification linkage, which goes well beyond a simple acquisition guide and implies learner account tracking or educational management features.
The trigger phrases are so generic that they could match ordinary user requests unrelated to this skill, causing over-invocation or routing the user into an unintended skill context. That increases the chance of prompt hijacking of normal conversations, user confusion, and inappropriate skill activation beyond its intended domain.
No suspicious patterns detected.
