Security audit
Security checks for vulnerabilities and agentic risk
This is a Chinese communication-polishing skill that only provides wording and reply suggestions, with no code execution or hidden system access.
Install this if you want Chinese-language help making messages sound more tactful. Be aware that common Chinese requests like asking how to reply may route into this skill, so review generated wording before sending it, especially in sensitive work or relationship contexts.
The trigger phrases include very common language such as '怎么回' and '润色一下', which can easily appear in ordinary conversation and cause the skill to activate when the user did not explicitly intend to invoke it. Unintended activation can override more appropriate system behavior, leading to response hijacking, confusion, or misuse of the skill in contexts where it should not run.
The manifest description is entirely in Chinese and all trigger phrases are Chinese-only, implying the skill operates in a fixed language. There is no indication that users may choose another language or that the Chinese-only scope is an intentional, justified regional limitation.
The fallback logic uses broad semantic inference like 'first person/imperative' or 'describes an awkward situation' without clear exclusion rules, so many unrelated user requests could be classified into one of the skill modes. This increases the chance of accidental routing and makes the skill behavior hard to bound, which is risky in an agent setting where skills should activate only under well-defined conditions.
