tabular-eda
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill performs legitimate data analysis tasks using trusted libraries (pandas, scikit-learn, numpy, matplotlib, marimo). The logic provided for detecting target leakage, high cardinality, and redundant features follows standard data science best practices.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external tabular data (CSV/Parquet), which constitutes a potential ingestion surface for untrusted data.
- Ingestion points: Reads tabular data provided by the user (as described in
SKILL.md). - Boundary markers: None implemented for the data content itself.
- Capability inventory: The skill performs statistical calculations, generates plots, and prints findings. It does not possess network exfiltration or dangerous command execution capabilities.
- Sanitization: Standard data type validation is performed (e.g.,
infer_target_type), but no explicit sanitization for natural language instructions embedded within data is present as the skill's primary focus is numerical and categorical analysis. Given the limited capabilities, the risk is minimal.
Audit Metadata