feature-rules
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill documents the use of local Python scripts (
build_team_game_features.py,feature_preview.py,legality_report.py) to process data. These scripts are invoked via standard shell commands and operate on local data files provided as arguments. - [EXTERNAL_DOWNLOADS]: The Python scripts depend on standard libraries such as
pandasandnumpy. The code includes user instructions for installing these packages viapip. No remote scripts or unverifiable dependencies are downloaded during execution. - [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided data files (CSV, Parquet, JSON) for statistical processing. The scripts perform structured data operations (sorting, shifting, grouping) and do not pass untrusted data into prompt contexts for interpretation, effectively mitigating injection risks.
- [DATA_EXFILTRATION]: All data processing is performed locally. The scripts lack network capabilities and only write results to user-specified local paths.
Audit Metadata