skills/jd-opensource/joysafeter/xlsx/Gen Agent Trust Hub

xlsx

Pass

Audited by Gen Agent Trust Hub on Sep 28, 2026

Risk Level: SAFECOMMAND_EXECUTIONPERSISTENCEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The recalc.py script dynamically generates a LibreOffice Basic macro file (Module1.xba) at runtime in the user application configuration directory to facilitate spreadsheet recalculation tasks.
  • [COMMAND_EXECUTION]: The recalc.py script invokes the soffice (LibreOffice) binary via subprocess.run to execute the generated macro against Excel files.
  • [PERSISTENCE]: The skill establishes persistence in the local environment by writing a macro configuration file to the LibreOffice user profile directory (~/.config/libreoffice or ~/Library/Application Support/LibreOffice/).
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from spreadsheet files (XLSX, CSV, TSV) using pandas and openpyxl. It lacks explicit boundary delimiters or data sanitization logic to mitigate potential prompt injection through cell content, while maintaining capabilities for file system modification and command execution. Evidence Chain: (1) Ingestion points: SKILL.md (pd.read_excel, load_workbook). (2) Boundary markers: Absent. (3) Capability inventory: recalc.py (subprocess.run), openpyxl/pandas (file writing). (4) Sanitization: Absent.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 28, 2026, 02:38 PM