The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute belt CLI commands for model inference, such as belt app run pruna/p-video.- [EXTERNAL_DOWNLOADS]: The skill references the installation of external components from the inference-sh GitHub repository and the inference.sh service. It directs users to fetch installation instructions from https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md.- [INDIRECT_PROMPT_INJECTION]: The skill exhibits a potential surface for command injection during the interpolation of user-supplied prompts into shell commands.
Ingestion points: User-provided strings for video prompts, images, or audio URLs are passed via the --input flag in SKILL.md examples.
Boundary markers: The examples use single quotes and JSON structures, but do not provide explicit instructions for the agent to escape shell metacharacters in user input.
Capability inventory: The skill utilizes Bash(belt *) which allows for arbitrary shell command execution within the scoped tool.
Sanitization: There is no evidence of automated sanitization or escaping of the user-provided prompt before it is interpolated into the belt app run command string.

p-video