The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (SAFE): The skill references downloading datasets (C4, Wikitext, The Stack) and model weights from the Hugging Face Hub. These are standard, reputable sources for machine learning development.
[COMMAND_EXECUTION] (SAFE): Documentation provides standard shell commands for environment setup, including pip install with specific index URLs and docker run instructions for TGI. These are informational and do not run automatically.
[REMOTE_CODE_EXECUTION] (SAFE): No patterns were detected involving the piping of remote scripts to shells or the dynamic execution of untrusted external code.
[DATA_EXFILTRATION] (SAFE): No sensitive file access or unauthorized network exfiltration patterns were observed; network usage is limited to standard library-level data loading.

gptq