The Agent Skills Directory

[SAFE]: The skill serves as a technical assistant for the Unsloth library, a legitimate tool for fine-tuning LLMs. The instructions provided are consistent with standard developer workflows.
[SAFE]: The 'Agent operating procedure' includes strong safety and integrity guidelines, such as verifying the environment before execution, running small-scale tests ('smoke runs'), and explicitly instructing the agent never to fabricate results or APIs.
[SAFE]: The dependencies listed in the YAML frontmatter—unsloth, torch, transformers, trl, datasets, and peft—are standard and expected packages for the machine learning training use case.
[SAFE]: No obfuscation, data exfiltration patterns, or unauthorized command execution triggers were found in the skill metadata or instructions. The reference files provided are static documentation indices and links.

unsloth