seedance-2-0
Pass
Audited by Gen Agent Trust Hub on Apr 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the Bun runtime to execute a TypeScript-based CLI tool (
seedance2.0.ts) that processes user arguments and interacts with the file system. - [EXTERNAL_DOWNLOADS]: Automatically downloads generated video files from ByteDance's official cloud storage to the local workspace upon task completion.
- [DATA_EXFILTRATION]: Transmits user-provided prompts and media URLs to ByteDance's Volcengine API endpoints (
ark.cn-beijing.volces.com) for processing. This is the intended functionality of the skill and utilizes a user-provided API key. - [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface (Category 8).
- Ingestion points: User-supplied video prompts and reference media URLs (images, videos, audio) entered via CLI arguments and processed in
seedance2.0.ts. - Boundary markers: Absent; inputs are directly embedded into the JSON payload sent to the external AI model.
- Capability inventory: The script has the capability to perform network requests (
fetch) and write files to the local disk (fs.writeFileSync). - Sanitization: No input validation or sanitization is performed on the prompt text or media URLs before they are passed to the downstream API.
Audit Metadata