The Agent Skills Directory

[SAFE]: No security issues detected. The skill's instructions and code snippets represent standard and safe usage of Resemble AI services.
[EXTERNAL_DOWNLOADS]: The skill mentions installing the chatterbox-tts Python package and downloading model weights from Hugging Face (ResembleAI/chatterbox). These are official resources for the described functionality.
[DATA_EXPOSURE]: The skill provides examples for API authentication using environment variables ($RESEMBLE_API_KEY) and placeholders (YOUR_API_KEY), which is a recommended security practice to avoid hardcoding credentials.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-supplied text for synthesis. It includes specific guidance on QA, text segmentation, and monitoring for artifacts like hallucinated continuation to manage the quality of the generated audio. As a text-to-speech utility, the risk of input manipulation is addressed through production heuristics and human review checkpoints. (Severity: SAFE).

resemble-chatterbox