The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references downloading AI models and development tools from well-known and reputable services such as the Hugging Face Model Hub and official GitHub repositories (e.g., Apple and the MLX community). These are standard practices in the AI development ecosystem.
[COMMAND_EXECUTION]: The documentation includes standard terminal commands for model quantization using pip install mlx-lm and mlx_lm.convert. These are legitimate developer tools used locally for model preparation and do not indicate malicious intent.
[PROMPT_INJECTION]: The skill outlines patterns for building chat and vision interfaces that process untrusted user data. While this creates a potential surface for indirect prompt injection within the resulting applications, the skill includes appropriate guidance on input validation and error handling to manage these risks.

local-ai-models