The Agent Skills Directory

[COMMAND_EXECUTION]: The recalc.py script executes system commands using subprocess.run to interact with LibreOffice (soffice) and timeout utilities (timeout/gtimeout). These calls use argument lists rather than shell strings, which mitigates shell injection risks.
[DYNAMIC_EXECUTION]: To enable headless formula recalculation, recalc.py automatically generates a LibreOffice Basic macro file (Module1.xba) and saves it to the user's application configuration directory (e.g., ~/.config/libreoffice/). The macro content is hardcoded within the script and is used to trigger the calculateAll() and store() methods in LibreOffice.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external spreadsheet files (.xlsx, .csv, etc.), which represents an attack surface where malicious instructions embedded in cell data could attempt to influence the agent's logic.
Ingestion points: Data is loaded using pd.read_excel() and load_workbook() as shown in SKILL.md.
Boundary markers: The instructions do not specify the use of delimiters or warnings to ignore instructions found within the data.
Capability inventory: The agent has the ability to execute shell commands (via recalc.py) and write files to the local filesystem.
Sanitization: No explicit sanitization or validation of cell content is implemented in the provided examples.

xlsx