calc

Pass

Audited by Gen Agent Trust Hub on Jun 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes external spreadsheet files (ODS, XLSX, CSV), creating an indirect prompt injection surface. Data within these files could contain hidden instructions targeting the agent.
  • Ingestion points: File paths provided to soffice and Python libraries (ezodf, uno) in SKILL.md.
  • Boundary markers: Absent; there are no instructions to delimit or ignore content within the files being processed.
  • Capability inventory: The skill executes shell commands (soffice, killall) and Python scripts capable of file and system interaction.
  • Sanitization: The instructions do not include validation or sanitization of spreadsheet content before processing.
  • [COMMAND_EXECUTION]: The skill executes shell commands via soffice for headless format conversion and spreadsheet management. It also includes the killall command for process management during troubleshooting.
  • [EXTERNAL_DOWNLOADS]: The skill recommends the installation of standard, well-known Python packages (ezodf, odfpy, pandas) via the pip package manager.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 24, 2026, 05:18 PM
Security Audit — agent-trust-hub — calc