The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text prompts and character dialogue to generate speech via various AI models. While this is the intended purpose, it creates a surface where embedded instructions in the processed data could attempt to influence the AI models' output or behavior.
Ingestion points: Text inputs provided to belt app run commands (e.g., prompt and text fields) in SKILL.md.
Boundary markers: Absent. The examples do not demonstrate the use of delimiters or 'ignore' instructions for the voice models.
Capability inventory: The skill uses the Bash tool to execute the belt CLI for remote inference.
Sanitization: None described in the documentation.
[EXTERNAL_DOWNLOADS]: The skill references external resources from the inference.sh platform and its GitHub repository for CLI installation and documentation.
Evidence: References to https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md and https://inference.sh for tooling and model playgrounds.
[COMMAND_EXECUTION]: The skill demonstrates the use of the belt CLI through the Bash tool to interact with remote AI applications.
Evidence: Multiple examples using belt app run with various model IDs such as infsh/kokoro-tts and inworld/text-to-speech-2.

ai-voice-cloning