The Agent Skills Directory

[SAFE]: The skill adopts a conservative security posture, enforcing local processing of legal materials and explicitly requiring user authorization for any network or publishing actions as documented in references/prior-art-and-trust.md.
[SAFE]: Robust file system protections are implemented in scripts/legal_meta_security.py; the resolve_regular_file and safe_relative_path functions prevent path traversal and symlink-based attacks by verifying every path component and rejecting parent directory markers.
[SAFE]: A proactive security scanning mechanism is included in scripts/validate_legal_skill.py, which uses regular expressions (SECRET_PATTERNS) to audit packages for leaked credentials and absolute user paths.
[COMMAND_EXECUTION]: The skill uses Python's subprocess module in scripts/legal_meta.py and various test files to orchestrate its own toolchain. These executions are secure as they use list-based arguments, the current Python interpreter (sys.executable), and are restricted to known internal script paths.

legal-meta-skill