seedance-director
Pass
Audited by Gen Agent Trust Hub on Apr 7, 2026
Risk Level: SAFENO_CODEPROMPT_INJECTION
Full Analysis
- [NO_CODE]: The skill is composed entirely of natural language instructions and examples, with no accompanying scripts or binaries.
- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its core function of processing user-supplied scene descriptions.
- Ingestion points: User-provided plain text descriptions and images parsed as scene inventory.
- Boundary markers: The skill lacks explicit delimiters or instructions to ignore control sequences embedded within the user input.
- Capability inventory: Limited to generating text prompts in a specific JSON format; no system-level or network access is requested.
- Sanitization: Output is constrained to a fixed JSON schema, which provides structural integrity but does not filter the content of the generated prompts for malicious instructions.
Audit Metadata