seedance-vocab-zh
Warn
Audited by Gen Agent Trust Hub on Apr 23, 2026
Risk Level: MEDIUMPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill includes instructions and vocabulary specifically designed to circumvent AI content safety filters.
- The description explicitly mentions 'filter-safe vocabulary for action, weapons, and combat' used for 'verifying term accuracy' or 'translating production concepts'.
- Section D.15 is titled 'Production Context Tokens (Filter Shield)', which provides specific phrases to mask the intent of prompts from safety filters.
- Sections D.7 ('Safe Action & Combat Vocabulary') and D.8 ('Safe Weapon Vocabulary') provide euphemisms (e.g., 'Elongated steel instrument' for sword) intended to bypass restrictions on violent or restricted content.
Audit Metadata