character-continuity-bible
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user data—such as character descriptions, prop functions, and scene details—which is then interpolated into structured prompt templates for downstream AI video generation. While the skill includes instructions to maintain safety boundaries (e.g., 'no pornographic framing'), it lacks explicit sanitization or strict boundary markers (delimiters) to prevent maliciously crafted user input from influencing or overriding the generated prompt's logic.
- Ingestion points: User briefs and reference images described in
SKILL.mdandreferences/continuity-template.md. - Boundary markers: Absent; user text is directly integrated into the 'Continuity Bible' output.
- Capability inventory: Generates complex text prompts for AI video/image models.
- Sanitization: Not observed in the instruction set.
Audit Metadata