skills/abcnuts/manus-skills/docx/Gen Agent Trust Hub

docx

Warn

Audited by Gen Agent Trust Hub on Jun 27, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes several external system binaries using Python's subprocess module for document processing and validation.
  • scripts/accept_changes.py and scripts/office/soffice.py invoke soffice (LibreOffice) to perform document conversions and track-change acceptance.
  • scripts/office/validators/redlining.py invokes git to generate word diffs for content validation.
  • scripts/office/soffice.py invokes gcc to compile a C source file into a shared object library at runtime.
  • [REMOTE_CODE_EXECUTION]: scripts/office/soffice.py implements a process injection technique to modify the runtime behavior of LibreOffice.
  • The script dynamically generates C source code and compiles it into a shared library (lo_socket_shim.so) in the temporary directory.
  • It uses the LD_PRELOAD environment variable to force the loading of this shared library into the soffice process. This technique is used to intercept and override standard socket system calls (like socket, listen, and accept) to bypass AF_UNIX socket restrictions often found in sandboxed environments.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 27, 2026, 02:53 PM
Security Audit — agent-trust-hub — docx