The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides links to installation instructions and documentation hosted on the official GitHub repository for inference-sh and the platform's primary domain. These resources are standard for setting up the required CLI environment for the skill's functionality.
[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow where user-supplied text (such as song prompts and lyrics) is passed to the belt CLI for processing by AI models.
Ingestion points: User-provided strings enter the context through the prompt and lyrics fields in the JSON input for the belt app run command in SKILL.md.
Boundary markers: The input is structured within JSON payloads, which provides a technical boundary between user data and command parameters, though there are no instructions for the agent to use specific delimiters to prevent the underlying music model from following instructions embedded in the prompt.
Capability inventory: The skill is scoped to the belt CLI tool, which is used specifically for interacting with remote inference APIs for audio generation.
Sanitization: No explicit sanitization or filtering of the user-provided prompt is performed within the skill; the system relies on the platform's API-side safety guardrails for the generated content.

ai-music-generation