The Agent Skills Directory

DYNAMIC_EXECUTION (MEDIUM): The script dynamically generates a StarBasic macro (Module1.xba) and writes it to the local filesystem at runtime before execution. This is a form of runtime code generation.\n- COMMAND_EXECUTION (MEDIUM): The script executes system commands via subprocess.run to invoke the soffice binary. Although it uses list-based arguments to prevent shell injection, it triggers the execution of a custom macro generated by the script.\n- PERSISTENCE (MEDIUM): The script modifies the user's LibreOffice configuration directory by installing a persistent macro in the standard application path. This artifact remains in the user's application profile after the script completes execution.\n- INDIRECT_PROMPT_INJECTION (LOW): The skill processes external Excel files which could contain malicious formulas designed to exploit the recalculation engine or the script's error-scanning logic. \n
Ingestion points: Excel file input (filename parameter in recalc.py).\n
Boundary markers: Absent; the script does not validate the content or source of the Excel file.\n
Capability inventory: File system write (macro creation), subprocess.run execution (LibreOffice), and openpyxl file reading.\n
Sanitization: Absent; the script extracts cell values directly to check for error strings without sanitizing the input data.

xlsx