Security audit
Security checks for vulnerabilities and agentic risk
The skill is mostly a disclosed STEM tutoring package, but it explicitly offers help lowering AI-content detection and includes student experiments without safety guardrails.
Review before installing. The tutoring content itself is low-risk and non-executable, but the publisher should remove the AI-detector-evasion offering, narrow activation triggers to explicit tutor requests, and add safety warnings and adult-supervision guidance for home experiments.
The skill explicitly advertises reducing AIGC/AI-content detection, which is an evasion-oriented capability aimed at bypassing institutional or platform integrity controls. Even though framed as 'polishing,' the stated goal is to lower detection risk, enabling academic dishonesty, policy circumvention, or misrepresentation of AI-generated work as human-authored.
Suspicious Unicode normalization or mixed-script content
The README centers the skill around Chinese-language tutoring and the primary usage example addresses the agent in Chinese, but it does not state that users may choose their preferred language. Because the policy requires avoiding forced language or locale constraints unless clearly optional or justified, this is a natural-language policy concern.
The activation example shows a very natural phrase ('@浩云学长 我二次函数不会') and describes automatic multi-step behavior, which can encourage broad invocation patterns and unintended triggering in ordinary educational chat. In agent platforms, overly loose activation cues can cause the skill to activate when the user did not explicitly intend to enter this workflow, leading to context hijacking or unnecessary agent actions.
The euphemistic 'professional polishing' section still directly offers optimization of AI-generated content to reduce detection, which is a classic evasion pattern. In an education-oriented skill, this is more dangerous because it can facilitate cheating, undermine assessment integrity, and normalize deceptive use by students seeking to disguise generated assignments.
The search keyword list includes broad conversational phrases such as variants of 'won’t/can’t do this topic,' which can match ordinary user speech and cause the skill to trigger outside its intended scope. Overbroad activation increases the chance that users are routed into this skill unintentionally, including when they did not ask for tutoring or for the add-on polishing behavior.
The visible natural-language metadata is entirely in Chinese for the primary name, title, and description, and the file does not indicate any user language choice or opt-in for interaction language. Under the stated policy, a skill that appears to assume a specific language without offering selection can be a locale/language policy concern.
The trigger list contains very broad subject terms such as '理科', '数学', '物理', and '化学', which can cause the skill to activate in many unrelated educational conversations. This creates unintended routing risk: users may be redirected into this skill without explicitly asking for it, reducing user control and potentially exposing them to irrelevant or biased guidance.
Suspicious Unicode normalization or mixed-script content
The household experiment section includes potentially hazardous activities such as using a laser pointer, batteries/wires/bulbs, glassware, and reactive materials like vinegar and baking soda, but provides no safety warnings, adult supervision guidance, PPE advice, or precautions against eye, burn, breakage, or ingestion hazards. In an education skill likely aimed at children or students, omission of safety guidance materially increases the chance of accidental injury or property damage.
This markdown file presents the skill's output templates almost entirely in Chinese, which effectively imposes a specific language on users. The policy allows locale or language constraints only when users are given a choice or when the constraint is clearly documented and justified; neither appears here.
The phrase "Auto cross-subject triggers" suggests the skill may trigger associations automatically, but the README does not explain under what exact conditions this happens or how users can avoid it. This ambiguity increases the risk of unintended behavior because the trigger scope is not clearly bounded in the documentation.
该知识库从标题到主要内容均默认使用中文进行说明和教学，未见关于语言偏好、切换机制或适用受众范围的说明。按规则，若技能强制特定语言而没有用户选择或明确的区域性正当说明，可构成自然语言层面的语言/locale 政策问题。
No suspicious patterns detected.
