predictive-modeling-best-practices
Pass
Audited by Gen Agent Trust Hub on Jun 26, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's primary function is to provide guidance and scripts for statistical analysis in ecology. A thorough review of the provided R and Python scripts shows no evidence of malicious behavior, such as unauthorized network access, file exfiltration, or persistence mechanisms.
- [EXTERNAL_DOWNLOADS]: The documentation mentions the use of the
rgbiflibrary in R to download occurrence records from the Global Biodiversity Information Facility (GBIF). GBIF is a well-known and trusted international network for biodiversity data, and such downloads are standard practice in ecological modeling. - [COMMAND_EXECUTION]: The skill includes R and Python scripts designed for command-line execution (e.g.,
collinearity_check.py). These scripts perform localized data processing on user-provided CSV files and do not invoke arbitrary shell commands or external processes in a dangerous manner. - [DATA_EXFILTRATION]: No patterns associated with data exfiltration were found. Network operations are limited to documented interactions with well-known scientific services, and no sensitive system files or credentials are accessed.
Audit Metadata