Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only skill for improving error messages, with broad auto-invocation wording but no hidden, destructive, or privileged behavior.
Reasonable to install if you want help rewriting application error messages or building support-facing troubleshooting workflows. Be aware that its current metadata may trigger on broad debugging or support requests, so explicit invocation is preferable when you only want it used for error-message work.
The manifest description uses very broad activation terms such as work-productivity, debugging, support, analysis, and implementation support, which can match a large share of ordinary user requests. This creates an overbroad invocation surface where the skill may activate in situations not specifically about improving error messages, increasing the chance of unintended routing or misuse.
The trigger sentences are broad and generic enough that the skill may activate for many ordinary support, debugging, or productivity requests without clear user intent. In an agent environment, over-broad activation can cause inappropriate routing, unexpected instruction injection into unrelated tasks, or the skill taking over conversations where the user did not explicitly request it.
The trigger phrase is broad enough to match ordinary user requests for help with error messages or workflows, which can cause the skill to activate outside clearly intended contexts. In an agent system, ambiguous activation boundaries can lead to prompt-routing mistakes, unexpected behavior, or unreviewed skill influence over general conversations.
The invocation example uses a vague request template that resembles normal conversational help-seeking rather than a bounded tool invocation. This increases the chance of accidental activation or trigger collision with other skills, which is especially risky in multi-skill agent environments where routing decisions may rely on lightweight phrase matching.
The trigger keywords are generic operational terms that commonly appear in normal support and development conversations. Because they are not specific to the skill's narrow purpose, they can cause the skill to be selected for many unrelated requests, undermining least-privilege routing and creating opportunities for prompt-surface expansion.
The example trigger sentences are built from everyday phrases and broad requests, without clear boundaries indicating when the skill should or should not activate. This encourages overmatching and makes it easier for unrelated user requests to satisfy the trigger examples, further broadening the skill's effective scope.
该文件整体以 zh-CN 语言环境提供内容，且未见提示用户可选择其他语言或说明该语言限制的适用范围。按照规则，若技能强制特定语言而没有用户 opt-in 或合理且明确的地区性约束说明，应视为自然语言策略问题。
触发关键词包含“work-productivity”“support”“debugging”等高度泛化词汇，容易与大量普通对话重叠，导致技能在非目标场景被误触发。误触发会把用户请求路由到不相关的技能，造成错误建议、上下文混淆和自动化流程偏转，尤其在多技能代理环境中更明显。
示例触发句使用了非常通用的表达，例如“I need a practical workflow...”和“Help me...”，但没有给出哪些相似请求不应触发该技能，导致触发边界模糊。这会放大误匹配概率，使系统把广泛的生产力或支持类请求错误归入该技能。
The default prompt uses broad, everyday phrasing such as helping with productivity, debugging, support, and practical workflows, which overlaps with many normal user requests. In combination with implicit invocation, this makes accidental or over-broad routing more likely, causing the skill to engage in contexts the user did not specifically intend.
The skill enables implicit invocation while providing no narrowly scoped trigger conditions beyond a broad productivity/error-message theme. This can cause the agent to auto-select the skill for ordinary troubleshooting or support requests, expanding its reach beyond clearly intended contexts and increasing the chance of prompt-surface abuse or unintended workflow interference.
The trigger phrase is so broad that it can match ordinary user requests about help, debugging, support, or practical workflows, causing the skill to activate outside its intended scope. Over-broad activation can hijack unrelated conversations, inject off-topic instructions, and increase the attack surface for prompt-routing abuse or unintended disclosure in contexts where the skill was not explicitly requested.
The file description states that SKILL.md contains English skill instructions while the Chinese version is in a separate file, but this README does not indicate that users can choose their preferred language or locale. This can be a natural-language policy concern because the default documentation presentation privileges a specific language without explicit opt-in.
