skills/apconw/aix-db/docx/Gen Agent Trust Hub

docx

Warn

Audited by Gen Agent Trust Hub on Jul 30, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill makes extensive use of the Python subprocess module to orchestrate external binaries required for document processing.
  • scripts/office/soffice.py invokes gcc to compile a technical system shim.
  • scripts/accept_changes.py and scripts/office/soffice.py execute soffice (LibreOffice) to perform doc/docx conversions.
  • scripts/office/validators/redlining.py runs git diff for comparing document versions.
  • SKILL.md documents the use of pandoc and pdftoppm for text and image extraction.
  • [REMOTE_CODE_EXECUTION]: The skill performs dynamic execution by generating and compiling local code at runtime.
  • scripts/office/soffice.py contains a C source string that is written to a temporary file and compiled into a shared library using gcc during execution.
  • The resulting library is then injected into the LibreOffice process via the LD_PRELOAD environment variable to hook system socket calls, facilitating operation in sandboxed environments.
  • [PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection as it is designed to ingest and process external document content.
  • Ingestion points: scripts/office/unpack.py reads user-provided .docx files and extracts their internal XML content for the agent to process.
  • Boundary markers: The provided instructions do not include delimiters or specific guidance for the agent to distinguish between its own operational instructions and data found within the document XML.
  • Capability inventory: The skill has high privileges including subprocess execution (including a compiler), file system write access, and process injection capabilities.
  • Sanitization: Although the skill employs defusedxml to protect against certain XML vulnerabilities like XXE, it does not sanitize natural language content that could be used to manipulate the agent's behavior.
  • [EXTERNAL_DOWNLOADS]: The skill instructions in SKILL.md direct the user or agent to install the docx Node.js package from a public registry. This is a well-known service and the dependency is consistent with the skill's document generation purpose.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 30, 2026, 04:07 PM
Security Audit — agent-trust-hub — docx