The Agent Skills Directory

[DYNAMIC_EXECUTION]: The R script scripts/rob2_plot.R contains a fallback mechanism that uses system2 to execute a local Python script (scripts/rob2_plot.py). The command is constructed dynamically using a computed path for the Python script and arguments (csv_path, save_name, output_dir) derived from user input.
[COMMAND_EXECUTION]: The skill instructions and scripts involve the execution of shell commands (Rscript, python) and file system operations like directory creation and file writing to generate plotting artifacts.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from a user-provided CSV file and passes metadata (file paths and output names) into the execution environment.
Ingestion points: CSV files are read by read.csv in scripts/rob2_plot.R and pd.read_csv in scripts/rob2_plot.py.
Boundary markers: None identified to separate data from execution instructions during file processing.
Capability inventory: The skill has capabilities for file reading, file system writing (PNG outputs), and subprocess spawning via R's system2 function.
Sanitization: While the R script validates specific domain assessment values (e.g., 'Low', 'High'), there is no explicit sanitization of the file paths or output names against path traversal or command injection characters before they are passed to the subprocess call.

meta-rob-plot