Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed Chinese academic paper-reading workflow that processes user-provided papers, with no evidence of hidden execution, exfiltration, persistence, or privilege escalation.
Before installing, be aware that this skill is optimized for Chinese paper translation and explanation and may process an entire supplied paper by default. Provide only documents you are allowed to process, and explicitly ask for a limited section or another output language when needed.
The activation description is broad enough to match many ordinary paper-related interactions, which can cause the skill to trigger without clear user intent. This creates a scope-control problem: the agent may begin translating, analyzing, or processing files when the user only wanted a narrow answer, increasing the chance of unintended file handling or over-collection of content.
The instruction to automatically enter the workflow whenever a user provides a paper file removes an explicit consent boundary and lacks limiting conditions. In practice, this can cause the agent to perform full-document extraction and transformation by default, even when the user may only want a small excerpt explained, leading to unnecessary processing of potentially sensitive or copyrighted material.
Mandating Chinese output without user choice can override user preferences and reduce transparency about how results will be presented. While not directly enabling code execution or exfiltration, it is still a policy and usability weakness because it may cause the agent to disregard explicit user language needs and produce inaccessible or unexpected output.
This markdown checklist repeatedly requires content to be translated or rendered in Chinese, such as translating figure titles/captions to Chinese and ensuring Chinese fluency. Because the file does not offer user opt-in or explain a region-specific requirement, it appears to impose a specific language policy through natural-language instructions.
No suspicious patterns detected.
