Security audit
Security checks for vulnerabilities and agentic risk
This is a markdown-only writing analysis skill whose disclosed behavior matches its purpose, with the main caution being that URL inputs may cause remote content fetching.
Install only if you want an AI-writing-style evaluator. Paste article text directly when privacy matters, and do not submit private, internal, or sensitive URLs unless you are comfortable with the agent fetching that content. Treat the AI-authorship verdicts as writing heuristics, not proof of authorship.
The suggested activation phrase "这篇文章有没有AI味" is conversational and broad enough that users might say it in ordinary discussion, making it unclear when the skill should activate versus when the user is just asking a general question. The README does not provide tighter trigger constraints, alternate exact invocations, or negative examples to narrow scope.
The README states that users can paste a URL and the skill will automatically fetch article content, but it does not warn users that this triggers network access and remote content retrieval. This can surprise users, expose internal or sensitive URLs if accepted blindly, and increase risk from fetching untrusted content or SSRF-like behavior depending on how the skill is implemented.
The trigger language is broad enough to activate on generic requests about whether text 'looks AI-written' or how to reduce AI flavor, which can cause the skill to run outside a narrowly intended scope. This is not code-execution dangerous, but it can lead to unintended invocation, misrouting, and user surprise, especially when adjacent to general editing or writing-analysis tasks.
The English 'When to Use' guidance overlaps with common requests about tone, authorship, and writing quality, making activation boundaries fuzzy. In agent environments, overly broad triggers can cause the wrong skill to seize control of a conversation and produce analyses the user did not explicitly request.
The skill instructs the agent to always output in the detected article language, removing user control over response language. This can override the user's accessibility, policy, or workflow needs, and in multilingual settings may produce unusable output or bypass higher-level UX expectations about honoring the user's chosen language.
This markdown file is entirely written as Chinese-only evaluation guidance, beginning with a Chinese title and instructions for calibration. For an all-file-types language policy check, this is a locale/language constraint presented without any opt-in, alternative language option, or stated region-specific justification.
This markdown file contains user-facing guidance only in Chinese, and there is no indication that the skill is region-specific or that users can opt into this locale. Under the language/locale policy rule, forcing a specific language without user choice can be a natural-language policy violation.
No suspicious patterns detected.
