Security audit
Security checks for vulnerabilities and agentic risk
This is a disclosed Chinese-language strategy/persona skill with no executable code, external access, persistence, or credential handling.
Installers should understand this skill is meant to emulate a public business-leadership perspective, not provide verified statements from the person. Consider narrowing triggers and adding language-choice guidance, but no harmful behavior was found in the artifacts.
The trigger list includes broad phrases such as “长寿时代”, “保险 + 医养”, and “92 派企业家”, which can plausibly appear in ordinary discussion without the user intending to invoke this persona skill. That creates unintended activation risk, causing the assistant to shift into role-based output and potentially override the user’s expected task framing or answer style.
This file’s natural-language content is effectively fixed to a single language/locale, which can violate a policy requiring user language choice or explicit justification for locale restrictions. The document does not mention that it is intentionally China/Chinese-language specific or provide any opt-in or alternative language guidance.
This markdown file is entirely presented in Chinese and does not offer any language choice or explain a region-specific requirement. Under the policy rule for language or locale constraints, this can be considered a natural-language policy violation because it implicitly fixes the skill content to a single language without user opt-in.
The activation instructions describe how to turn the skill on but do not define when it must stay off, and they provide no counterexamples. Without explicit scope boundaries, the agent may activate on ambiguous or indirect references, increasing the chance of unintended persona adoption in unrelated conversations.
L096 规定“激活后，我会以陈东升的第一人称视角回应，语言风格：理论化、逻辑清晰、引经据典”，且全文均以中文描述触发和使用方式。该文件没有明确说明可根据用户偏好的语言或地区设置进行切换，存在默认强制单一语言/locale 的倾向。
No suspicious patterns detected.
