The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references installation instructions and configuration files hosted on the official GitHub repository for the inference-sh platform.
[COMMAND_EXECUTION]: The skill facilitates the execution of the belt CLI tool to run AI models. This is restricted via the allowed-tools: Bash(belt *) configuration, which limits execution to specific subcommands of the belt tool.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a proxy for video generation by ingesting user-provided prompts, images, and video URLs.
Ingestion points: User input enters the agent context through the prompt, reference_images, and video parameters in SKILL.md examples.
Boundary markers: The skill uses structured JSON objects within the --input flag to pass data to the belt CLI, providing a degree of separation between data and instructions.
Capability inventory: The skill uses the belt command to interface with remote APIs.
Sanitization: The skill relies on the underlying belt CLI and inference.sh backend to handle sanitization of generated content.

happyhorse