Security audit
Security checks for vulnerabilities and agentic risk
TeacherKit is a disclosed education prompt-generation skill with no evidence of data exfiltration, persistence, credential access, or destructive behavior.
Install only if you want a Chinese-first teaching assistant for lesson plans, quizzes, syllabi, slide outlines, and teaching scripts. Review or adjust the language setting and be aware that broad education terms may invoke the skill, but the inspected artifacts do not show hidden network access, credential use, persistence, or destructive actions.
Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
}
)
return prompt
async def generate_quiz(
self,
Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
}
)
return prompt
async def generate_quiz(
self,
Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
}
)
return prompt
async def generate_quiz(
self,
Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
}
)
return prompt
async def generate_quiz(
self,
Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
}
)
return prompt
async def generate_quiz(
self,
The skill metadata advertises invocation on the broad everyday term "备课," which can appear in many ordinary education-related conversations without the user explicitly asking to invoke a tool-like skill. This creates an overbroad trigger surface that may cause unintended activation, leading to irrelevant or surprising behavior and possible context leakage into the skill flow.
The skill instructs the agent to auto-select functions from broad natural-language descriptions, including multiple loosely defined keywords like "备课", "考试", and "PPT." This ambiguity increases the chance of unintended invocation or wrong function selection, which can route user conversations into the wrong workflow and produce outputs based on assumptions the user did not confirm.
This markdown file contains natural-language descriptions of skill behavior and output principles entirely in Chinese, including what the skill generates. Under the policy, forcing a specific language without user opt-in can be a locale/language policy violation, and there is no indication here that users may choose another language.
The user valve sets language to 中文 by default, which imposes a specific output language unless the user changes the setting. This is reinforced by helper logic that falls back to Chinese output when no explicit language is provided.
The _get_lang_instruction method returns an instruction to output entirely in Chinese for any language value other than explicit English or bilingual selections. This can violate language-choice policy because users who do not opt in are still steered into a specific locale.
No suspicious patterns detected.