The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided prompts for video generation, which creates an entry point for untrusted data to influence agent behavior or downstream model output.
Ingestion points: Input parameters (e.g., "prompt") in the Bash command examples within SKILL.md.
Boundary markers: No explicit delimiters or guardrail instructions are used to separate user input from system-level instructions in the command templates.
Capability inventory: The skill utilizes the Bash tool to execute belt CLI commands for model inference.
Sanitization: The skill does not describe any validation or filtering mechanisms for the input JSON objects.
[EXTERNAL_DOWNLOADS]: The skill instructions direct the installation of external components from belt-sh and reference documentation from the inference-sh GitHub repository.
[COMMAND_EXECUTION]: The skill requires execution of the belt CLI tool via shell commands to interact with remote AI models.

p-video