seedance-video
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
- [NO_CODE]: The skill consists entirely of Markdown instructions and YAML metadata. It includes no executable code, shell scripts, or binary files.
- [SAFE]: The skill references fal.ai, a well-known AI model hosting service, for purposes of API endpoint documentation and schema resolution. These references do not involve untrusted code execution or sensitive data exfiltration.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to handle user-provided multimedia references (images, video, audio) to construct video generation requests. While this creates a data ingestion surface, the risk is minimal and inherent to the skill's primary function of multimodal video planning.
- Ingestion points: User-supplied image, video, and audio references and text prompts (SKILL.md).
- Boundary markers: Not explicitly defined in the provided instruction templates.
- Capability inventory: Construction of API requests to Seedance providers and use of the show_video display tool.
- Sanitization: No specific data sanitization or validation protocols are described in the instructions.
Audit Metadata