seedance

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides examples of using the belt CLI tool, which is the official interface for the inference.sh platform. The commands (belt login, belt app run) are standard usage for the service and do not contain malicious payloads.
  • [EXTERNAL_DOWNLOADS]: The skill contains links and installation instructions for official resources, including the belt-sh/cli package and documentation from inference.sh and raw.githubusercontent.com/inference-sh/. These are well-known, trusted sources for this specific platform and the downloads are for intended functionality.
  • [INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it processes user-provided prompts and external URLs for images, videos, and audio. However, this is the primary purpose of a video generation skill, and the risk is mitigated by the platform's standard AI guardrails and the fact that these are inputs to a generation model rather than a general-purpose agent loop.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 02:20 PM
Security Audit — agent-trust-hub — seedance