Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed design-project skill generator that creates local project Markdown files and a generated design workflow, with no evidence of deception, exfiltration, destructive behavior, or hidden persistence.
Install this only if you want an agent to create and maintain local Markdown files for a design project. Review the generated preview before approving installation, and be aware that generated project skills may activate on broad design-continuation phrases and will persist progress files in the project directory.
The trigger conditions are very broad and instruct the system to prioritize this skill for generic design intent, which can cause it to activate in many normal conversations where the user did not clearly request filesystem-generating behavior. Over-broad activation increases the chance of unintended tool use, unnecessary data collection, and premature file-generation workflows.
The trigger description instructs activation on very broad phrases such as '继续', '接下来', '开始设计', and generally prioritizes this skill whenever design intent is inferred. This can cause unintended invocation in unrelated or only loosely related conversations, leading the agent to read/write project files, shift workflow state, or steer the user into a persistent design process without clear user intent.
The trigger text explicitly says the skill should activate on broad, common phrases like “继续”, “接下来”, and “开始设计”, which are highly ambiguous and likely to appear in unrelated conversations. In an agentic context, overbroad activation can cause unintended tool use, file reads/writes, and workflow takeover when the user did not actually intend to enter this design-management process.
The README presents activation examples like "帮我生成一个设计 skill" and "创建设计项目" without narrowing context or listing exclusion conditions. These phrases are broad enough to match ordinary conversation about design work, which could cause unintended skill activation.
The skill explicitly instructs scanning the user's home directory for global skill paths, which extends access beyond the current workspace and is not necessary to generate or install a project-local design skill. This broadens filesystem visibility, may expose unrelated personal or configuration data, and violates least-privilege expectations for a content-generation workflow.
The skill directs creation of directories and files in the workspace and /tmp, plus packaging and moving ZIP archives, but does not require an explicit safety notice describing the concrete filesystem changes before execution. Even though it asks the user to choose a delivery mode, the lack of a clear pre-execution warning and change summary increases the risk of users authorizing operations without understanding their scope.
This markdown file contains natural-language instructions that effectively force a specific language/locale for users of the skill. Under the policy, language constraints should either be optional for the user or clearly justified as region-specific; neither is present here.
文件中的自然语言说明全部以中文给出，且描述未提供语言选择、用户偏好继承或仅限中文场景的说明。按规则，这构成可能的语言/区域策略问题，因为技能默认强制特定语言而没有用户选择机制。
The skill instructs the agent to automatically create the project directory if it does not exist and later mandates automatic saving and overwrite updates for multiple files. Without explicit user confirmation, preview, or safeguards, this can lead to unintended filesystem modifications, overwriting prior work, or creating persistent artifacts the user did not expect.
This markdown file contains all instructional content in Chinese and does not mention that the language is optional or targeted to a specific audience. Under the policy, forcing a specific language without user opt-in can be a natural-language policy violation.
No suspicious patterns detected.
