horror-design
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references multiple educational video tutorials on Douyin (a well-known video platform) for cinematography and lighting techniques. These links are documented in references/SOURCE-LEDGER.md for source verification purposes.\n- [INDIRECT_PROMPT_INJECTION]: The skill transforms user trigger words into structured visual parameters for downstream AI platforms.\n
- Ingestion points: User trigger words defined in SKILL.md.\n
- Boundary markers: The output prompt templates in Section 7 do not implement delimiters or explicit instructions to ignore embedded commands in user content.\n
- Capability inventory: The skill generates formatted parameters for multiple image and video generation tools (MJ, SD, Kling, Jimeng).\n
- Sanitization: There is no documented validation or sanitization mechanism for user-supplied triggers.
Audit Metadata