Security audit
Security checks for vulnerabilities and agentic risk
This is a markdown-only tutoring skill with broad educational scope and no executable code, persistence, credential access, or hidden data flow.
This skill appears safe to install as a tutoring aid. Be aware that it may activate for broad learning requests, including regulated subjects like medicine, finance, and law, so treat those outputs as educational explanations rather than professional advice. In mixed-language conversations, specify your preferred output language if you do not want Chinese.
The trigger guidance is extremely broad and encourages invocation for vague phrases like not understanding a topic or needing to get up to speed fast. That can cause the skill to activate in contexts where a more specialized or safer skill should handle the request, increasing the chance of misrouting sensitive domains such as medicine, law, or finance into a generic tutoring workflow.
Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.
## Phase 2: Teach — Structured Dialogue
Work through the subject conversationally, one concept at a time.
Never lecture for more than a few paragraphs without pausing to check in.
### The Teaching Loop (repeat for each concept)
The skill instructs the agent to default to Chinese for mixed-language conversations unless the user specifies otherwise. This can cause the generated tutorial to appear in an unexpected language, leading to user confusion, incorrect comprehension, or accidental mishandling of important educational content.
The line instructs authors to use Template B's bilingual approach or write fully in Chinese for Chinese-speaking audiences. Because the policy scope includes language or locale constraints, this is a mild natural-language policy concern unless the skill makes user language preference an explicit opt-in choice.
No suspicious patterns detected.