The Agent Skills Directory

[COMMAND_EXECUTION]: The recalc.py script executes system commands using subprocess.run, specifically calling soffice (LibreOffice) and timeout (or gtimeout on macOS). These commands are used to trigger formula recalculation in a headless office environment.
[COMMAND_EXECUTION]: The skill modifies the local host configuration by writing a LibreOffice Basic macro (Module1.xba) to the user's application configuration directory (e.g., ~/.config/libreoffice/ or ~/Library/Application Support/LibreOffice/). This persistent modification is required for the skill's calculation features but represents a change to the local environment.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes untrusted external spreadsheet data.
Ingestion points: Data is ingested from user-provided Excel, CSV, and TSV files using pandas and openpyxl as detailed in SKILL.md and recalc.py.
Boundary markers: No boundary markers or explicit instructions are provided to the agent to treat data within cells as untrusted or to ignore instructions embedded in the data.
Capability inventory: The skill is capable of writing files to the local filesystem and executing system commands through the recalc.py script.
Sanitization: No sanitization, validation, or filtering of the spreadsheet content is performed before processing or recalculation.
[PROMPT_INJECTION]: There is a metadata discrepancy: the skill author is identified as cam10001110101, while the LICENSE.txt file claims copyright by Anthropic, PBC. This inconsistency could be misleading regarding the skill's provenance.

xlsx