Security audit
Security checks for vulnerabilities and agentic risk
This is a local-LLM setup guidance skill with overly broad activation wording but no hidden code, persistence, credential handling, or privileged behavior.
This skill appears safe to install for local LLM setup advice. Be aware that it may trigger more often than intended because its activation wording is broad; review any generated setup commands, downloads, or hardware recommendations before acting on them.
The trigger sentence is so broad and awkwardly templated that it can match user requests far outside a narrowly defined local-LLM setup use case, increasing the chance this skill activates when it should not. Over-broad activation can route unrelated prompts into this skill, causing unintended behavior, misleading advice, or policy bypass opportunities through misclassification of user intent.
The trigger sentence is phrased as a broad, natural-language request and lacks tight activation boundaries. In an agent system, this can cause accidental or inappropriate skill invocation, leading the agent to apply this skill outside its intended scope and potentially override more suitable or safer behaviors.
This trigger uses a generic 'I need a practical workflow for...' pattern that is ambiguous and easy to match across many unrelated user requests. Such overbroad matching increases prompt-routing risk, where the skill may be invoked in contexts it was not designed to handle, reducing reliability and potentially exposing downstream workflows to inappropriate instructions.
The trigger phrases are broad and generic enough that the skill could activate for loosely related requests about local AI, workflows, privacy, or consumer hardware. This can cause unintended invocation, leading the agent to apply this skill in the wrong context and potentially produce irrelevant, misleading, or overconfident setup guidance.
The trigger examples are broad, natural-language phrases that could match ordinary user requests and cause this skill to activate when it was not explicitly intended. Unintended invocation can steer conversations toward this skill's workflow and outputs, increasing the chance of misrouting, prompt-scope confusion, or inappropriate handling of requests that only loosely relate to local LLM setup.
The trigger keywords are broad and weakly bounded, including generic terms like "privacy" and common technical phrases that may appear in many unrelated requests. This can cause accidental invocation of the skill, leading the agent to apply specialized local-LLM guidance when the user did not intend it, which increases the risk of misrouting, irrelevant advice, or unintended workflow changes.
The sample trigger phrases use very generic formulations like "Help me" and "I need a practical workflow for...", which are common across many user requests and do not clearly delimit when this skill should activate. If downstream routing or prompt matching relies on these examples, the skill may be selected for loosely related tasks, causing prompt-scope confusion and reducing the reliability of skill selection.
The default prompt is phrased broadly enough that ordinary user requests about software, data, local LLMs, consumer GPUs, CPU inference, workflows, or implementation help could trigger implicit invocation without strong boundaries. This can cause the skill to activate in contexts the user did not explicitly intend, increasing the chance of prompt-routing mistakes, over-collection of user context, or inappropriate delegation to a specialized skill.
The usage signals section lacks precise boundaries for when the skill should and should not activate, and several triggers are generic enough to overlap with ordinary requests for help. In a routing system, ambiguous trigger definitions can cause prompt misrouting, accidental invocation, and broaden the skill's effective authority beyond its intended domain.
该文件为 SKILL.zh-CN.md，内容整体以简体中文呈现，但未见向用户提供语言/地区选择或说明该语言限制的适用范围。按 SQP-3，若技能隐含强制特定语言而无用户选择或正当限定，可能构成语言/地区策略问题。
No suspicious patterns detected.
