atai-newton-omega-model-data-prep
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements standard data science workflows including linear imputation, out-of-time train/test splitting, and dimensionality reduction (PCA) using reputable libraries like NumPy, Pandas, and Scikit-Learn.
- [SAFE]: Analysis of the source code in
data_preprocessor.py,data_splitter.py, andfeature_preparer.pyconfirms that the skill only performs local data transformations and does not initiate network requests or access sensitive system files. - [SAFE]: The skill uses lazy imports for plotting functionality (
matplotlib), which is a secure and standard practice for handling optional runtime dependencies. - [SAFE]: The instructions in
SKILL.mdare consistent with the technical implementation and do not contain any prompt injection patterns or attempts to bypass safety guardrails.
Audit Metadata