Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only skill for mobile responsive layout help, with broad activation wording but no hidden execution, data access, persistence, or destructive behavior.
Installers should expect this skill to activate for common frontend layout and responsive-design requests. If precise routing matters, narrow the trigger phrases or disable implicit invocation, but no malicious or high-risk behavior was found in the artifacts.
The trigger phrases are extremely generic and mirror ordinary user requests about mobile responsiveness, layout, and practical workflows. In systems that auto-invoke skills from natural-language matching, this can cause unintended activation, routing user requests into this skill when they did not explicitly choose it, which may lead to incorrect task handling or prompt-surface expansion.
The README lists natural-language trigger phrases such as "Help me ..." and "I need a practical workflow ..." that are broad conversational patterns and could overlap with ordinary user requests. The file does not provide clear scope boundaries, exclusion conditions, or negative examples to distinguish when this skill should activate versus when it should not.
The trigger keywords are broad enough to match many ordinary frontend or content-related requests, which can cause the skill to activate outside its intended scope. Overbroad activation increases the chance of prompt/skill hijacking at the routing layer, where this skill may preempt more appropriate skills or inject irrelevant workflow instructions into unrelated tasks.
The example trigger phrases use very general language ('Help me', 'I need a practical workflow') around a broad requirement statement, which trains or encourages invocation on commonplace requests rather than a narrowly bounded task. This makes misrouting more likely and can lead to unintended skill selection, reducing isolation between skills and potentially exposing users to irrelevant or conflicting instructions.
The trigger keywords are broad and overlap with common frontend discussion terms such as 'layout', 'responsive design', and 'frontend'. This can cause the skill to activate in unrelated conversations, increasing the chance of inappropriate context capture, workflow hijacking, or user confusion when a more suitable skill or direct response should have been used.
The skill enables implicit invocation without any visible activation guardrails, narrowing criteria, or user-confirmation requirements. That can cause the agent to invoke this skill opportunistically on loosely related requests, increasing the chance of unintended prompt routing, over-collection of user context, or execution of skill-specific workflows the user did not explicitly request.
The invocation guidance lists keywords and trigger sentences but does not define when the skill should not activate, how much semantic overlap is required, or whether explicit user intent is necessary. In a common domain like frontend/UI help, this ambiguity increases accidental or overly aggressive routing, which can be abused or simply lead to incorrect tool selection.
The trigger sentence is broad enough to match ordinary user phrasing about getting help with mobile responsiveness, which can cause the skill to activate when the user did not explicitly intend to invoke it. Unintended activation can route conversations into this skill's workflow unexpectedly, reducing user control and potentially crowding out more appropriate skills.
This is a Chinese README, but core descriptive and trigger content is presented in English, including the invocation phrases. That can impose a language expectation on users without explicit opt-in or a documented language-selection policy.
The example trigger phrases encourage invocation for a broad problem statement but do not define exclusion criteria or when the skill should not run. Without explicit non-trigger guidance, orchestration systems or users may over-apply the skill, leading to unnecessary activation and lower-quality or mis-scoped assistance.
No suspicious patterns detected.
