The Agent Skills Directory

[COMMAND_EXECUTION]: The recalc.py script executes the soffice (LibreOffice) binary using subprocess.run to recalculate formulas. The command is constructed using a list of arguments, preventing shell injection vulnerabilities from malicious filenames.
[DYNAMIC_EXECUTION]: The script recalc.py automatically generates a LibreOffice Basic macro (Module1.xba) and stores it in the user's application configuration directory. This is a legitimate functional requirement to facilitate headless recalculation and saving of Excel documents.
[INDIRECT_PROMPT_INJECTION]: The skill processes spreadsheet files which are an external data source, presenting an attack surface for indirect prompt injection.
Ingestion points: Spreadsheet data is loaded via pd.read_excel in SKILL.md and load_workbook in recalc.py.
Boundary markers: The instructions do not specify any delimiters or warnings to ignore potentially malicious instructions embedded in spreadsheet cells.
Capability inventory: The skill possesses capabilities for file system modification and local command execution through the soffice binary.
Sanitization: There is no evidence of cell value sanitization or validation before the data is processed by the agent.

spreadsheet-processor