Security audit
Security checks for vulnerabilities and agentic risk
This is a simple English-learning coaching skill with no code execution, data access, persistence, or hidden behavior.
Install this if you want an English practice helper. Be aware that it may be invoked for broad English-learning requests and its practice prompts default to English, so ask for Chinese or bilingual explanations if that better fits your learning style.
The description says the skill should be used whenever the user needs speaking practice, listening improvement, exam prep, or simply wants to communicate in English. These are very broad conditions without clear trigger boundaries or negative examples, which could cause unintended invocation for many ordinary language-learning conversations.
The prompt instructs the assistant to act as an English conversation partner and continue the interaction in that mode, but it does not offer a language choice or confirm that the user wants responses in English. This may violate language/locale policy if the skill is invoked for users who have not explicitly opted into English-only interaction.
The writing prompt directs the assistant to coach and review writing in English but does not state that English-only feedback is optional or user-selected. Because the file otherwise targets a Chinese-speaking audience, the lack of explicit opt-in for language mode creates a policy concern around forced language choice.
No suspicious patterns detected.
