The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for using the belt CLI tool, including belt login for authentication and various belt app run commands for generating video and audio content via remote AI models.
[COMMAND_EXECUTION]: The skill instructs the agent to install multiple external skills using the npx skills add command, which introduces third-party code and capabilities from the belt-sh and inference-sh repositories.
[EXTERNAL_DOWNLOADS]: The skill references an external setup guide and documentation hosted on GitHub at raw.githubusercontent.com/inference-sh/skills/.
[INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by ingesting natural language instructions that are passed directly to media generation tools.
Ingestion points: Prompt templates for video scenes and text-to-speech scripts located within SKILL.md.
Boundary markers: The instructions do not utilize specific delimiters or isolation techniques to prevent the generated prompts from influencing the agent's behavior.
Capability inventory: The skill uses shell execution (Bash(belt *)) to interact with external AI applications, providing a bridge between untrusted data and active capabilities.
Sanitization: No evidence of input validation, escaping, or filtering is present in the instructions before content is interpolated into CLI commands.

explainer-video-guide