ai-avatar-video

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as a legitimate technical guide for the inference.sh AI video generation platform.
  • All external links, including documentation and installation scripts, point to verified domains (inference.sh) and official GitHub repositories (github.com/inference-sh).
  • [COMMAND_EXECUTION]: The skill defines clear boundaries for shell interactions.
  • The allowed-tools configuration 'Bash(belt *)' ensures the agent can only execute the specific 'belt' utility, preventing arbitrary command execution.
  • [INDIRECT_PROMPT_INJECTION]: The skill includes patterns for processing remote media files.
  • Ingestion points: The skill demonstrates passing image and audio URLs to the 'belt app run' command in SKILL.md.
  • Capability inventory: File processing and network interactions are performed by the 'belt' binary, which is the only authorized tool.
  • Sanitization: Inputs are wrapped in structured JSON payloads, which helps prevent command injection and clearly separates data from the executable command structure.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 01:13 PM
Security Audit — agent-trust-hub — ai-avatar-video