Skills
skills/eigent-ai/agent-skills/docx/Gen Agent Trust Hub

docx

Warn

Audited by Gen Agent Trust Hub on Apr 8, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [DYNAMIC_EXECUTION]: scripts/office/soffice.py writes C source code to a temporary file and compiles it into a shared library using gcc at runtime.
  • [DYNAMIC_EXECUTION]: scripts/office/soffice.py uses the LD_PRELOAD environment variable to inject the compiled shared library into the soffice process.
  • [DYNAMIC_EXECUTION]: scripts/accept_changes.py generates a LibreOffice Basic macro and saves it to the user's profile before executing it via a specialized vnd.sun.star.script URI.
  • [COMMAND_EXECUTION]: The skill invokes various external command-line tools including soffice, pandoc, pdftoppm, and gcc to perform its tasks.
  • [EXTERNAL_DOWNLOADS]: SKILL.md contains instructions to install the well-known docx library from the npm registry.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted Word documents and possesses capabilities such as command execution and runtime compilation. 1. Ingestion points: scripts/office/unpack.py (reads .docx files). 2. Boundary markers: Present in SKILL.md via specific task instructions. 3. Capability inventory: subprocess.run calls across multiple scripts for system tool execution. 4. Sanitization: Employs defusedxml for secure XML processing.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 8, 2026, 10:20 AM
Security Audit — agent-trust-hub — docx