Security audit
Security checks for vulnerabilities and agentic risk
This is a markdown-only product strategy review skill with no code execution, data access, persistence, or hidden system changes.
Before installing, users should know this skill is aimed at Chinese-language product strategy critique and may activate on broad requests to evaluate a product idea. It appears safe from a security standpoint because it ships only markdown guidance and does not run code or access data.
The skill metadata is written as Chinese-only context and does not offer user language choice, which can force responses into a language the user did not request. While not a direct security exploit, this is a control and safety issue because it can reduce user comprehension of important caveats, recommendations, or risk statements, especially in mixed-language environments.
The activation phrases are very broad and match ordinary requests such as 'help me look at this product direction' or 'evaluate this idea', which makes unintended invocation likely during normal conversation. This can cause the system to route user queries into this specialized persona when the user did not explicitly request it, leading to scope drift, unwanted behavior, or policy bypass if the skill has different response rules than the base assistant.
This markdown file is written fully in Chinese and does not indicate that users may choose another language or that the skill is intentionally limited to a Chinese-speaking context. Under the stated policy, forcing a specific language without user opt-in is a natural-language policy violation.
No suspicious patterns detected.
