chinese-journal-of-management
Pass
Audited by Gen Agent Trust Hub on Jul 24, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill uses relative path references (
../../resources/source-basis.mdand../../resources/official-source-map.md) to access external documentation. While standard for organized repositories, using path traversal to access content outside the skill's specific directory is a pattern associated with data exposure. - [PROMPT_INJECTION]: The skill processes user-supplied manuscripts to provide feedback, creating an indirect prompt injection surface. Content within untrusted documents could attempt to manipulate the agent's advisory behavior, though the skill itself contains no malicious directives.
Audit Metadata