Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed Trump news summarizer with no code execution, persistence, credential access, or hidden high-impact behavior.
Before installing, understand that this skill is designed to browse and summarize recent Trump-related news and will default to Chinese output. Users who want another language or a narrower source set may need to override those preferences explicitly.
Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
信息截至 {datetime}，来源包括：{source_list}
### Output Guidelines
- **Language**: All output in Chinese (中文), including headlines and analysis
- **Tone**: Objective and factual; present multiple perspectives when available
The trigger description is broad enough to activate on common-language requests such as general Trump-related queries, which increases the chance of unintended skill invocation. Over-broad routing can override user intent, cause unnecessary browsing or source aggregation, and lead the assistant to follow this skill’s rigid output behavior when a simpler or different response was wanted.
The skill overview states it will produce a Chinese-language summary report regardless of the user’s requested language, which can override normal assistant behavior and user preference. Forced language output is risky because it can degrade usability, cause policy or safety messaging to be delivered in an unintended language, and create confusing or inaccessible responses.
The explicit output guideline mandating that all output be in Chinese hard-codes response language without user opt-in. In skill systems, rigid formatting and language constraints can act as prompt-level control that conflicts with higher-level user expectations and may suppress clear safety communication if the user does not understand Chinese.
The reference table explicitly labels the primary recommended sources as English-language (EN), and the suggested search queries are predominantly English with only one Chinese example. This creates a natural-language locale bias without documenting user choice or explaining that the skill is intended only for English-language news research.
No suspicious patterns detected.