Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed short-video script and storyboard helper with no executable code, persistence, credential handling, or hidden system changes.
Install this if you want a Chinese-language workflow for short-video scripts and detailed Vidu-ready storyboard prompts. Review generated English prompts and any web-searched character references before approving handoff to video generation, especially if likeness or brand accuracy matters.
The skill entry uses broad trigger terms such as '脚本创作/分镜/编导/内容创作', which are generic enough to match many ordinary creative requests. In an agent system, overly broad routing can cause this skill to activate unexpectedly, override more appropriate tools, or induce unintended downstream actions such as web searching or handing content to video-generation components without clear user intent.
Mandating English output without user choice creates a prompt-routing and usability safety issue because it can override the user's language expectations and cause hidden transformations of user intent. In multilingual agent workflows, forced language conversion may also degrade accuracy, lead to miscommunication in generated prompts, or cause the system to pass content to downstream tools in a form the user did not approve.
The explicit requirement that all prompts must be in English forces a language choice without opt-in and can bypass user control over how their content is represented. While not a classic security flaw, it is a genuine safety and policy concern in agentic systems because it can introduce ambiguity, mistranslation, and reduced transparency before content is sent to other components like video generation.
No suspicious patterns detected.
