The Agent Skills Directory

[COMMAND_EXECUTION]: The recalc.py script uses subprocess.run() to invoke the soffice (LibreOffice) binary and system timeout utilities. While these calls use argument lists to mitigate shell injection, they represent the execution of external binaries based on agent-provided filenames.
[DYNAMIC_EXECUTION]: The recalc.py script dynamically generates and writes a StarBasic macro (Module1.xba) to the local filesystem to enable LibreOffice formula recalculation. This is a form of runtime code generation and execution.
[PERSISTENCE]: The skill modifies the user's LibreOffice configuration by writing the RecalculateAndSave macro to the application's standard library directory. This modification persists across sessions and affects the application's global environment.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process spreadsheets (.xlsx, .csv, etc.), which are untrusted external data sources that could contain malicious instructions.
Ingestion points: Spreadsheet files are read in recalc.py (via openpyxl) and via instructions in SKILL.md using pandas.read_excel and load_workbook.
Boundary markers: None. The skill does not define delimiters or warnings to separate data content from instructions.
Capability inventory: The skill can write files to the local system (wb.save) and execute shell commands via recalc.py.
Sanitization: None. The skill does not specify any sanitization or validation of the contents of the spreadsheets it processes.

xlsx