The Agent Skills Directory

[PERSISTENCE]: The recalc.py script installs a persistent LibreOffice macro by writing to ~/.config/libreoffice/4/user/basic/Standard/Module1.xba (Linux) or ~/Library/Application Support/LibreOffice/4/user/basic/Standard/Module1.xba (macOS). This configuration persists across sessions and could overwrite existing user-defined macros stored in the default module location.
[DYNAMIC_EXECUTION]: The skill dynamically generates StarBasic macro code (stored as a string within recalc.py) and writes it to the local filesystem. This code is then executed by the LibreOffice process via a command-line trigger.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external spreadsheet files (XLSX, CSV, TSV). Malicious content within these files could attempt to influence the agent's behavior or subsequent actions.
Ingestion points: Data is loaded via pd.read_excel() and load_workbook() as described in SKILL.md.
Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are utilized when processing cell data.
Capability inventory: The skill can execute subprocesses (soffice) and write files to the system.
Sanitization: There is no evidence of content filtering or escaping for spreadsheet data before it is processed by the agent.
[COMMAND_EXECUTION]: The recalc.py script utilizes the subprocess module to invoke system binaries such as soffice, timeout, and gtimeout to facilitate the formula recalculation workflow.

xlsx