image-to-video

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references and directs the agent to fetch installation documentation from a remote GitHub repository at raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md.
  • [COMMAND_EXECUTION]: The skill utilizes the belt CLI tool to perform model inference and media merging tasks, which involves executing shell commands with user-defined parameters.
  • [INDIRECT_PROMPT_INJECTION]: The skill demonstrates a workflow where untrusted user input (prompts) is interpolated into shell commands, creating a surface for indirect injection.
  • Ingestion points: User-provided text prompts and image file paths defined in the SKILL.md usage examples.
  • Boundary markers: No delimiters or safety instructions are provided to separate user-supplied prompt data from the command structure.
  • Capability inventory: Access to the Bash tool to run the belt CLI for remote AI inference and local file processing.
  • Sanitization: The skill lacks explicit sanitization or validation logic for external strings before they are passed to the CLI tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 10:10 AM
Security Audit — agent-trust-hub — image-to-video