subsystem-edit

Warn

Audited by Gen Agent Trust Hub on May 16, 2026

Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: Both the PowerShell and Python scripts in this skill are configured to execute an external validation tool located in a relative sibling directory.
  • In scripts/subsystem-edit.ps1, the script attempts to execute subsystem-validate.ps1 from a path constructed as ../../subsystem-validate/scripts/ using powershell.exe.
  • In scripts/subsystem-edit.py, the script uses subprocess.run to execute subsystem-validate.py from a path constructed as ../../subsystem-validate/scripts/.
  • This pattern of executing code from computed relative paths outside of the skill's own package directory is a security concern as it relies on the integrity of the surrounding environment.
  • [PROMPT_INJECTION]: Static analysis detected homoglyph characters in SKILL.md. These characters are part of Cyrillic strings used for legitimate Russian localization within the 1C Enterprise context (e.g., "Продажи", "Товары", "Справочник") and appear to be benign documentation rather than an attempt to bypass agent instructions.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 16, 2026, 11:10 AM
Security Audit — agent-trust-hub — subsystem-edit