seedance
Pass
Audited by Gen Agent Trust Hub on Mar 29, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill instructions (specifically in
SKILL.mdandreferences/compliance.md) direct the agent to 'route around' or 'bypass' the content filters of the third-party Seedance platform. This is achieved through descriptive substitution (e.g., replacing 'Iron Man' with 'red gold metal armor warrior'). While this is a form of filter circumvention, it is a functional feature of the skill's intended purpose and does not target the host agent's safety guidelines. Additionally, the skill explicitly instructs the agent to refuse politically sensitive content. - [EXTERNAL_DOWNLOADS]: The documentation and installation instructions in
README.mdreference external resources, including official ByteDance Lark documentation and the vendor's own GitHub repository. These references are used for installation and informational purposes and follow standard practices for skill distribution.
Audit Metadata