skills/4444j99/a-i--skills/xlsx/Gen Agent Trust Hub

xlsx

Warn

Audited by Gen Agent Trust Hub on Oct 6, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDYNAMIC_EXECUTIONPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The script scripts/office/soffice.py dynamically writes C source code to a temporary file and compiles it into a shared object (.so) library using gcc. This library is then injected into the execution of LibreOffice via the LD_PRELOAD environment variable to shim network-related system calls (socket, listen, accept). While this is intended to enable LibreOffice in restricted sandbox environments, runtime compilation and syscall shimming are high-risk behaviors.
  • [COMMAND_EXECUTION]: The skill executes multiple shell commands via subprocess.run, including gcc for compilation, git diff for redlining validation, and soffice (LibreOffice) for spreadsheet recalculation and document processing.
  • [PERSISTENCE]: The recalc.py and scripts/recalc.py scripts modify the local user's LibreOffice configuration by creating a StarBasic macro file (Module1.xba) in the application's configuration directory. This macro remains present across sessions to facilitate automated formula recalculation.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted Office documents (.xlsx, .docx, .pptx) which provide a large attack surface for indirect injection. Maliciously crafted documents could attempt to exploit vulnerabilities in the underlying processing libraries (pandas, openpyxl) or the LibreOffice engine. The skill employs defusedxml in some components to mitigate XML-based attacks.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 6, 2026, 04:22 PM