king-hu-perspective
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [SAFE]: The skill operates as a high-level persona and aesthetic framework for film analysis. It does not utilize system tools, perform network requests, or execute code. All instructions are focused on guiding the AI's creative output style and mimicking the director's philosophy.
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection as it processes untrusted user data. Ingestion points: User-provided storyboard descriptions, scene scripts, and creative queries processed in Phase 1 and 2. Boundary markers: Absent. The skill does not use specific delimiters to wrap user input or provide instructions to ignore embedded commands within the input. Capability inventory: None. The skill does not request any
allowed-toolsand does not contain subprocess, file-write, or network operations. Sanitization: Absent. There is no evidence of filtering or validation for the incoming creative text. - [DATA_EXFILTRATION]: The skill includes absolute local file paths in its 'Appendix' (e.g.,
/Users/baimengke/Documents/...). This reveals the author's local directory structure and system username, which is a form of metadata exposure, although no sensitive credentials are included and no network transmission is initiated by the skill itself.
Audit Metadata