The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the belt CLI tool from a third-party source (belt-sh organization on GitHub). It also references external installation instructions at https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md which is outside the list of trusted vendors.
[PROMPT_INJECTION]: The skill examples and parameter definitions include a disable_safety_filter option set to true. This configuration is used to instruct the underlying AI model to bypass its default content safety guidelines.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided content (voice_script, video_prompt, voice_prompt) which is then interpolated into shell commands via the belt CLI tool.
Ingestion points: voice_script, video_prompt, voice_prompt, image, and audio parameters within the belt app run command in SKILL.md.
Boundary markers: None observed in the command templates or documentation to delimit untrusted input.
Capability inventory: Execution of the belt CLI tool using the Bash tool to perform network-based inference.
Sanitization: The skill does not define specific sanitization or escaping logic for the user-provided strings before they are passed to the CLI.

p-video-avatar