Security audit
Security checks for vulnerabilities and agentic risk
This is a formatting and reasoning guide for answers, with no code execution, data access, persistence, or hidden high-impact behavior.
Install this if you want answers to be automatically structured and sometimes bilingual. Be aware that ordinary phrases like asking for a simple explanation or comparison may activate its style rules, and mixed Chinese-English prompts may receive bilingual replies.
The skill uses very common natural-language phrases like '简单说一下', '详细解释', '对比一下A和B', and '你怎么看' as activation/style-routing cues. Because these overlap heavily with ordinary user prompts, the skill can be unintentionally invoked or steer response structure without explicit consent, creating prompt-priority and behavior-routing risk rather than direct code-execution risk.
The skill instructs the agent to respond bilingually whenever the user mixes languages, even if the user did not explicitly request bilingual output. This can cause unintended disclosure amplification, longer outputs, and reduced user control over formatting or language preference, though the security impact is limited in this general Q&A context.
No suspicious patterns detected.
