Security audit
Security checks for vulnerabilities and agentic risk
This is a simple text-rewriting skill with no code execution, credentials, persistence, or hidden data access found.
Use this skill when you explicitly want text rewritten to sound less AI-generated. Review the result for factual accuracy, source preservation, and unintended language or tone changes before publishing or submitting it.
The phrase "Or invoke directly when editing documents" does not specify what exact trigger phrase, command, or context should activate the skill. Because it lacks constraints or negative examples, it creates ambiguity about when the skill should run versus when ordinary document editing should not invoke it.
The skill body is written entirely in Chinese and presents the behavior as a fixed-language experience, without indicating that the output should follow the user's language. In a mixed-language environment, this can lead to unintended language switching, user confusion, and unsafe routing where content is transformed into the wrong language or loses fidelity during rewriting.
The invocation guidance is extremely broad: users are told to send any AI-generated text and receive a rewritten version. That can overlap with ordinary writing/editing requests and cause the skill to trigger in situations where the user did not explicitly ask for AI-humanization, increasing the chance of unintended transformation of content or bypass of higher-level review flows.
No suspicious patterns detected.
