Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only design skill for recreating SVG or Canvas artwork from reference images, with no hidden execution, persistence, credential use, or network behavior.
Install this if you want agents to use measurement-heavy workflows for recreating icons, card faces, UI screenshots, or similar raster references as SVG/Canvas. Expect it to read and analyze images you provide and possibly create project-local verification scripts or comparison pages when the task calls for it.
The skill content is written entirely in Chinese and the finding indicates it forces Chinese-language presentation without user opt-in. This can override the user's preferred language, reduce transparency of the agent's actions, and create a prompt-scope conflict where the agent follows skill-local language constraints rather than user instructions.
The skill's activation guidance is very broad, with multiple natural-language examples like '按这张图的图案改' and '照着这个图做图标' that can match many ordinary design requests. Over-broad triggers can cause the agent to invoke this specialized workflow unnecessarily, leading to inappropriate tool usage, scope creep, or accidental processing of user images when a simpler response was intended.
No suspicious patterns detected.
