The Agent Skills Directory

[SAFE]: The skill follows standard data science practices for model training and evaluation. All identified external resources and behaviors align with the stated purpose of performing statistical modeling and data analysis.
[EXTERNAL_DOWNLOADS]: The documentation provides instructions for installing necessary R dependencies (optparse, data.table, Matrix, xgboost) from the official Comprehensive R Archive Network (CRAN) repository. This is an established and well-known service for R package distribution.
[COMMAND_EXECUTION]: The skill utilizes Rscript to execute the provided analysis scripts. These operations are limited to local data processing, model fitting, and file output within the user-specified directory.
[PROMPT_INJECTION]: The skill processes external tabular data, which represents an attack surface for indirect prompt injection.
Ingestion points: External tabular data files (CSV, TXT, TSV) are ingested via the --data_file parameter and processed by the read_data function in scripts/utils.R.
Boundary markers: No specific delimiters are used to wrap the processed data content, as the data is handled numerically and structurally by R scripts rather than being directly interpolated into a natural language prompt for the agent to interpret.
Capability inventory: The skill performs file reading, local file writing for results (tables and figures), and logging of session information. No network or system-level administrative operations are performed on the processed data.
Sanitization: Predictors are sanitized in scripts/functions.R using the sanitize_predictors function, which handles missing values and ensures categorical variables are correctly typed. Column names are processed with make.names to ensure they are valid identifiers.

xgboost-analysis