Security audit
Security checks for vulnerabilities and agentic risk
This is a text-only AI-writing risk review skill with disclosed limits and no evidence of hidden access, persistence, or unsafe execution.
Installers should understand that this skill produces heuristic AI-writing risk assessments, not proof of authorship or misconduct. It is best suited for Chinese-language review workflows unless the user asks for another language, and any high-stakes use should rely on additional evidence such as drafts, sources, revision history, or human review.
The skill metadata and default prompt are written to force Chinese output regardless of the user's language preference. This can degrade usability, cause misunderstanding of sensitive evaluative results, and create consent/localization issues when the user expects another language or when downstream systems assume language alignment.
The default invocation prompt is generic and can be triggered on arbitrary text without clearly constraining when the skill should be used or what inputs are appropriate. In an agent environment, overly broad activation increases the chance of misrouting unrelated user requests into AI-authorship adjudication, producing unwarranted accusations, privacy-sensitive analysis, or unreliable conclusions outside the intended scope.
This markdown template is entirely written as a Chinese-only reporting format and does not indicate that users may choose another language or locale. Under the stated policy, forcing a specific language without user opt-in is a natural-language policy violation unless the regional constraint is explicitly documented and justified.
No suspicious patterns detected.
