The Agent Skills Directory

[COMMAND_EXECUTION]: The recalc.py script uses subprocess.run to execute system commands such as soffice (LibreOffice), timeout, and gtimeout. These commands are used to perform headless operations on spreadsheet files.
[DYNAMIC_EXECUTION]: The script generates a StarBasic macro (Module1.xba) at runtime and saves it to the local user's LibreOffice configuration directory (~/.config/libreoffice/ or ~/Library/Application Support/LibreOffice/). This macro is then executed by the LibreOffice process to recalculate formulas.
[INDIRECT_PROMPT_INJECTION]: The skill processes external spreadsheet data, creating a potential vector for indirect prompt injection attacks.
Ingestion points: The skill reads data from .xlsx, .csv, and .tsv files using pandas.read_excel and openpyxl.load_workbook as indicated in SKILL.md.
Boundary markers: No specific delimiters or safety instructions are defined to separate untrusted data from processing instructions.
Capability inventory: The skill allows the agent to write files, perform data analysis, and execute the recalc.py script which interacts with the host OS and office software.
Sanitization: There is no evidence of sanitization or filtering of cell contents to prevent malicious instructions from being interpreted by the agent during analysis.

xlsx