skill-system-foundry

Pass

Audited by Gen Agent Trust Hub on Mar 16, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns or security vulnerabilities were detected in the skill's instructions, metadata, or scripts.
  • [COMMAND_EXECUTION]: The skill includes several Python scripts (validate_skill.py, audit_skill_system.py, scaffold.py, bundle.py) designed to manage the skill lifecycle. These scripts use standard library modules (os, shutil, zipfile) for file manipulation and directory traversal. Analysis shows they are well-bounded to the local filesystem and do not execute external payloads or perform unauthorized command injection.
  • [DATA_EXFILTRATION]: The bundle.py script and its associated library (scripts/lib/bundling.py) include logic to package skills into archives. This logic implements significant security safeguards, including an exclude_patterns list in configuration.yaml that prevents the inclusion of sensitive directories (like .git, __pycache__, and platform-specific hidden files) in the generated bundles. Furthermore, the script uses a is_within_directory check to prevent path traversal attacks, ensuring the bundler cannot be manipulated into reading files outside the defined system root.
  • [INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing untrusted data (user-defined skill names and descriptions). However, it mitigates this through strict validation logic in scripts/lib/validation.py, which enforces character limits, naming conventions, and prohibits specific keywords related to known AI vendor restrictions.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 16, 2026, 05:41 PM