skill-creator

Pass

Audited by Gen Agent Trust Hub on Apr 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/init_skill.py performs file system operations including directory creation and file writing based on user-supplied arguments. It also programmatically sets execution permissions (0o755) on dynamically created Python scripts.\n- [COMMAND_EXECUTION]: The scripts/init_skill.py script generates new executable content by interpolating user-provided skill names into pre-defined Python templates. This runtime generation of scripts and subsequent file system modification is a core feature of the skill creator.\n- [COMMAND_EXECUTION]: The skill enables writing to local file system paths specified by the user through command-line arguments in scripts/init_skill.py. While the script checks if a directory exists before creation, it lacks rigorous sanitization of the path argument before performing file system operations.\n- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface in scripts/init_skill.py where untrusted user input is interpolated into script templates. Evidence Chain:\n
  • Ingestion point: sys.argv in scripts/init_skill.py\n
  • Boundary markers: Absent during template interpolation\n
  • Capability inventory: File system writes and permission modifications in scripts/init_skill.py\n
  • Sanitization: Input validation is absent during initialization, though separate validation logic exists in scripts/quick_validate.py.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 7, 2026, 09:23 AM
Security Audit — agent-trust-hub — skill-creator