The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documents how to use the belt CLI tool to interface with various AI models (such as FLUX, Veo, and Claude) for generating media content. These commands are standard operations within the supported ecosystem for social media automation.
[EXTERNAL_DOWNLOADS]: The documentation references installation commands for the belt CLI and other related skills. These resources are hosted on GitHub and are essential for the skill's functionality.
[INDIRECT_PROMPT_INJECTION]: The provided bash examples include patterns where user-defined variables are interpolated into prompts for AI models. While this represents a potential surface for indirect injection if the variables contain adversarial content, it is a routine part of prompt-driven automation.
Ingestion points: Variable assignments for content topics and concepts (e.g., CONCEPT and TOPICS) within the bash script examples in SKILL.md.
Boundary markers: Commands use JSON-formatted input strings to encapsulate user data, providing some structure.
Capability inventory: The skill examples demonstrate calling the belt command to run remote inference apps, which process the interpolated prompts.
Sanitization: The examples do not include explicit logic for sanitizing or escaping the user-provided variables before they are used in prompts.

ai-social-media-content