The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface as it processes external CSV files (expression matrices, group labels, and feature lists) provided by the user.
Ingestion points: The read_expression_matrix, read_group_file, and read_feature_list functions in scripts/io.R handle external file input.
Boundary markers: The agent interacts with these files via CLI arguments; no specific prompt boundary markers are used for the content of the data files.
Capability inventory: The skill is capable of reading and writing to the local filesystem and performing statistical computation via the glmnet R package.
Sanitization: The implementation provides high-quality sanitization, including numeric verification of matrices (storage.mode(matrix_data) <- "double"), duplicate detection, and strict validation of group labels in scripts/functions.R to ensure only binary classification is performed.

elastic-net-feature-selection