Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed investment research workflow with expected internet research, sentiment analysis, risk calculation, and PDF generation behavior.
Before installing, confirm that you want a Chinese-oriented investment research workflow that may query public web/news/social sources through dependent skills and create PDF reports. Do not provide confidential client data, private holdings, or nonpublic investment theses unless you have reviewed the dependent skills' data handling.
The skill description and all user-facing examples are written exclusively in Chinese, and there is no indication that users may interact in other languages or choose a preferred locale. Under the stated policy, forcing a specific language without user opt-in is a natural-language policy violation unless the regional constraint is clearly documented and justified.
This markdown file contains user-facing instructions and examples primarily in Chinese, including the trigger guidance and operational description, without stating that the skill is Chinese-only or offering a language/locale choice. Under the language/locale policy rule, forcing a specific language without user opt-in is a natural-language policy concern.
The workflow explicitly performs network collection, social-media sentiment analysis, and report generation, but the description lacks a clear warning about what user/query data, fetched content, and potentially sensitive investment research inputs may be transmitted to external services. In a chained multi-skill workflow, this increases the risk of unintended disclosure of sensitive research topics, portfolio assumptions, or client-related information because users are not clearly informed before data leaves the local conversation context.
The trigger set includes broad generic phrases such as '投资研究', '行业分析', 'investment analysis', and 'portfolio research', which can match many ordinary user requests and cause this workflow to activate unexpectedly. Because the workflow chains research, sentiment analysis, risk calculation, and PDF generation, unintended activation can lead to unnecessary processing, accidental use of downstream skills, and confusing or unauthorized report generation in contexts where the user did not explicitly request this workflow.
The manifest requires a Chinese-font-capable Linux environment for PDF generation, which implicitly constrains output presentation to a specific locale without any indication of user preference negotiation. In this workflow, the forced locale is not directly a security exploit, but it can create misleading assumptions about user intent, reduce accessibility, and cause inappropriate rendering or disclosure of localized content when a different language or formatting context was expected.
No suspicious patterns detected.
