The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill includes instructions to download pre-trained tokenizer models and datasets from the HuggingFace Hub (huggingface.co). These are standard library operations targeting a well-known service.
[INDIRECT_PROMPT_INJECTION]: The skill demonstrates how to train tokenizers using external files or data iterators. This ingestion of untrusted data is the primary purpose of the library and is documented using standard security practices for the tokenization pipeline.
[COMMAND_EXECUTION]: The skill provides standard command-line instructions for installing official Python packages (tokenizers, transformers, datasets) via pip.
[SAFE]: No malicious patterns, obfuscation, persistence mechanisms, or unauthorized privilege escalation attempts were detected in the skill instructions or reference materials.

huggingface-tokenizers