Security audit
Security checks for vulnerabilities and agentic risk
This is a Chinese writing and thinking prompt skill with no code execution, network access, persistence, or hidden data transfer, though users should only give it files they intend the agent to read.
Install only if you want a Chinese-language writing/thinking assistant. When using the file-path option, provide only notes or drafts you intentionally want analyzed; pasted text is safer for sensitive material.
The skill is designed to auto-trigger on broad user intent even when the user does not explicitly invoke the skill. This can cause unintended interception of ordinary writing, critique, or brainstorming requests and silently apply hidden behavior from referenced mode files, reducing user control and making prompt-routing easier to manipulate.
The routing table uses common conversational phrases like '帮我看看' or '发散一下' as automatic activation signals, which are likely to appear in normal chat outside a deliberate skill invocation. This creates ambiguous routing and increases the chance that the agent follows skill-specific constraints or reads additional references when the user only intended a casual request.
This markdown file specifies the skill behavior exclusively in Chinese and does not indicate that users may choose another language. Under the natural-language policy rule, forcing a specific language without user opt-in is a policy concern.
The entire skill specification is written as a Chinese-only operating mode and presents no option for users to choose another language. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless the constraint is explicitly justified.
The skill explicitly tells the agent to read user-provided file paths, which can enable unintended local file access if the runtime permits filesystem reads. Without restrictions, path validation, or user-facing warnings, an attacker could try to coerce access to sensitive local files, turning a content-writing skill into a data exposure path.
The skill content is entirely written to operate in Chinese and does not provide any mechanism to detect, preserve, or ask for the user's preferred language. In practice, this can cause unintended language switching, reduce usability, and create prompt-control issues where the assistant ignores user language context, though it is not a direct security compromise.
The skill content is entirely written in Chinese and instructs a fixed stylistic output without indicating that the language should follow the user's preference. This can cause the agent to respond in an unexpected locale, reducing usability and potentially causing misunderstanding for users who did not opt into Chinese output.
