Security audit
Security checks for vulnerabilities and agentic risk
The skill matches a 408 exam tutoring purpose, but it should be reviewed because it mandates ongoing study-profile archiving and external knowledge-base or web lookup without clear user consent, retention, or deletion controls.
Install only if you are comfortable with the tutor keeping an ongoing record of your study topics, weak points, review schedule, and cited resources. Avoid sharing sensitive personal details in questions, and prefer an implementation that asks before saving progress or sending queries to web search.
声明描述的是一个综合性的408考研辅导技能，核心能力包括基于用户知识库进行问答、检索、总结和个性化提醒；而代码实际只实现了一个独立的命令行脚本，用固定间隔计算复习计划。虽然代码与“艾宾浩斯遗忘曲线/费曼学习法”中的复习提醒主题部分相关，但它既不处理408学科内容，也不访问任何知识库、真题或用户资料，更没有问答、索引、总结或主动提醒能力。因此其实际行为只覆盖声明中的很小一部分概念，整体主用途与声明明显不符。
The activation criteria are broad enough to trigger during many ordinary study conversations, increasing the chance the skill activates without clear user intent. That can unexpectedly change response behavior and cause unnecessary retrieval, logging, or reminder workflows to run on benign user queries.
The skill instructs persistent tracking, archiving, and reminder generation for user learning data without any notice or consent mechanism. This creates privacy risk because study history, weaknesses, and interaction-derived profiles may be retained longer than the user expects.
The instructions explicitly require logging and summarizing dialogue content into a user-specific knowledge record. Persistent per-user profiling of conversation content can expose sensitive educational history and creates a larger privacy/security footprint if those records are later accessed or leaked.
The skill directs the agent to search a knowledge base and potentially the web without warning users that their questions may be transmitted to external systems. If a user includes sensitive study notes or personal details, those could be shared beyond the immediate chat context.
Mandating that every Q&A be recorded and all conversation knowledge points be extracted into tracking artifacts goes beyond what many users would reasonably expect from a tutoring interaction. This increases the amount and granularity of retained personal data, raising privacy and secondary-use risks.
Requiring end-of-conversation updates to a structured user knowledge map and mandatory summary generation institutionalizes persistent user profiling as a default behavior. In the context of an educational skill, this is more dangerous because it continuously aggregates weaknesses, review cadence, and learning patterns without any visible privacy boundary.
The entire document is written in Chinese and presents guidance without offering any language or locale choice. Under the stated policy, forcing a specific language without user opt-in can be a natural-language policy violation unless the locale constraint is explicitly documented and justified.
The document title and all instructional content are written exclusively in Chinese, including the directive to update the knowledge index after each conversation. Under the policy, forcing a specific language without user opt-in or clear region-specific justification is a natural-language policy violation.
The module docstring, usage text, and user-facing output are all presented only in Chinese. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale restriction is explicitly documented and justified.
The weekday labels and command-line messages printed to users are fixed in Chinese, and there is no mechanism to select another language. This enforces a specific locale at runtime without user choice or justification.
SQP-3 applies to all file types and covers language or locale policy violations. This markdown file presents all headings and guidance exclusively in Chinese, with no opt-in, alternative locale, or justification that the skill is intended only for Chinese-speaking users.
