Security audit
Security checks for vulnerabilities and agentic risk
This skill is a local Chinese dating-chat phrase helper with some overpromised analysis language, but no evidence of hidden access, persistence, network use, or destructive behavior.
Install only if you want a Chinese-language dating/chat phrase helper. Treat its emotion and intent labels as generic keyword-based suggestions, not real analysis of another person, and review any suggested message before sending it.
声明描述的是一个具备分析能力和上下文理解能力的聊天辅助技能，应根据具体聊天场景解释对方话语含义并给出定制化回复建议。实际代码只是一个CLI脚本，维护若干静态文案数组，并根据命令参数打印对应分类内容或随机挑选一句。虽然主题上都与恋爱聊天/回复建议有关，但核心行为与主用途存在实质差异：代码没有读取或解析用户与女生的真实对话文本，没有任何NLP、推断、打分、分析说明或风险提示逻辑。因此该描述不能准确代表代码实际行为，属于明显不匹配。
The skill metadata promises contextual analysis of a girl's emotion/intent plus 3-5 tailored replies with reasoning and pitfalls, but the implementation only emits hard-coded scripts. This is a specification-deception issue: users or upstream agents may rely on capabilities the code does not provide, causing misleading automation behavior and unsafe social guidance decisions.
The skill description includes broad natural-language trigger phrases such as general reply-help requests, which can overlap with ordinary conversation and cause unintended activation. In a multi-skill environment, ambiguous routing can expose users to the wrong skill behavior, produce confusing outputs, or let this skill intercept requests not meant for it.
Telling users they can invoke the skill without any command creates ambiguous activation conditions and increases the chance of accidental triggering. In systems that rely on heuristic matching, this broadens the attack surface for prompt-routing mistakes, where unrelated user messages may be captured and processed by this skill instead of the correct one.
The file’s user-facing description and all generated outputs are written exclusively in Chinese, indicating the skill is designed to operate in a single language. There is no opt-in, fallback, or justification that this is a region-specific tool, which conflicts with the language/locale policy criteria.
This JavaScript skill’s natural-language content, command names, and usage instructions are entirely in Chinese, and there is no indication that the user can choose another language or that the locale limitation is intentional for a region-specific skill. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation.
These sections expose generic bedtime/escalation and reply-script capabilities without any situational interpretation, despite the skill being advertised as context-aware advice. In this social-assistance context, that mismatch can push inappropriate or manipulative responses because users may assume the tool has analyzed the conversation state when it has not.
