seedance
Pass
Audited by Gen Agent Trust Hub on Apr 30, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: References external installation instructions for the platform's CLI tool from the vendor's GitHub repository (inference-sh/skills).
- [COMMAND_EXECUTION]: Instructs the agent to use the
beltCLI via the Bash tool to perform video generation tasks on the inference.sh platform. - [PROMPT_INJECTION]: Surface for indirect prompt injection identified through the processing of untrusted external content.
- Ingestion points: The skill accepts text prompts and remote URLs for images, videos, and audio files as input parameters for the
beltcommand. - Boundary markers: None observed in the command structures to isolate external inputs from instructions.
- Capability inventory: The skill uses the Bash tool to execute system commands.
- Sanitization: No explicit sanitization or validation of the remote URLs or prompt content is performed within the provided instructions.
Audit Metadata