Security audit
Security checks for vulnerabilities and agentic risk
This is a plain writing skill for beauty-product short-video sales scripts, with no code execution, network access, persistence, or hidden data handling.
Installers should understand this skill is designed to create persuasive beauty-product sales copy, including purchase calls to action. Review generated claims for advertising accuracy and platform rules, especially for cosmetics or skincare effects, but the artifact itself does not show unsafe system access or hidden behavior.
The trigger phrases are broad enough to match many ordinary script-writing requests, which can cause the skill to activate outside its intended beauty-commerce scope. That creates routing and prompt-selection risk: users asking for generic writing help may be unexpectedly steered into sales-script behavior, leading to irrelevant, manipulative, or policy-misaligned outputs.
The activation logic explicitly says to invoke automatically and '无需额外提问', but it does not define when the skill must not activate. In an agentic system, ambiguous auto-invocation increases the chance of incorrect tool/skill selection, causing the model to bypass clarification and produce persuasive commerce content for requests that may be generic, incomplete, or outside intended scope.
No suspicious patterns detected.
