scikit-learn

Pass

Audited by Gen Agent Trust Hub on Apr 9, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides high-quality, comprehensive documentation for the scikit-learn machine learning library, covering classification, regression, clustering, and preprocessing.
  • [EXTERNAL_DOWNLOADS]: The skill instructions suggest installing well-known and trusted Python packages such as scikit-learn, pandas, numpy, and matplotlib via the 'uv' package manager.
  • [COMMAND_EXECUTION]: Provided scripts (classification_pipeline.py and clustering_analysis.py) perform standard local machine learning workflows using publicly available datasets (e.g., scikit-learn's breast cancer and iris datasets). No dangerous shell commands or network exfiltration patterns were identified.
  • [DATA_EXFILTRATION]: No instances of sensitive data harvesting or unauthorized network transmissions were found. The skill focuses on local data processing.
  • [PROMPT_INJECTION]: The instructions are strictly focused on technical guidance for scikit-learn and do not contain any patterns attempting to override agent behavior or safety filters.
  • [REMOTE_CODE_EXECUTION]: There are no patterns involving the download and immediate execution of scripts from remote servers.
  • [SAFE]: Documentation of model persistence using 'joblib' and 'pickle' follows standard data science practices for saving and loading locally trained models.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 9, 2026, 10:12 PM