skills/binhmuc/autobot-review/docx/Gen Agent Trust Hub

docx

Pass

Audited by Gen Agent Trust Hub on Oct 9, 2026

Risk Level: SAFEPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PRIVILEGE_ESCALATION]: The skill instructions guide the agent to install system dependencies such as pandoc, libreoffice, and poppler-utils using sudo apt-get install. These tools are retrieved from well-known official repository services.
  • [INDIRECT_PROMPT_INJECTION]: The skill's document processing logic is vulnerable to a Zip Slip path traversal attack during document unpacking.
  • Ingestion points: ZIP-based Office archives are extracted using zipfile.extractall() in ooxml/scripts/unpack.py (line 15), ooxml/scripts/validation/base.py (line 625), and ooxml/scripts/validation/redlining.py (line 54).
  • Boundary markers: The skill does not implement boundary markers or instructions to validate extracted ZIP entry paths against traversal sequences.
  • Capability inventory: The skill uses subprocess.run to execute system utilities and has the capability to write to arbitrary file system locations via the unvalidated extraction process.
  • Sanitization: While the skill correctly uses defusedxml to sanitize XML content, it fails to sanitize ZIP entry filenames, which could allow a malicious document to overwrite scripts or configuration files in the workspace.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 9, 2026, 04:35 AM