The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/close_feature.py uses subprocess.run to execute a local validation script (validate_verification.py). This is a safe implementation as it passes arguments as a list, avoids spawning a shell, and strictly validates the feature argument against a lowercase alphanumeric slug pattern to prevent injection.
[SAFE]: The skill performs directory cleanup using shutil.rmtree in scripts/close_feature.py. This functionality is restricted by security checks that prevent operations on symlinks and ensure the target directory resides strictly within the expected .specs/features path, mitigating path traversal risks.
[SAFE]: The scripts remediation.py and scripts/review_convergence.py implement state transition logic and audit logging using standard libraries (json, hashlib, tempfile). Atomic file writes are performed safely using temporary files and directory existence checks.

wtk-ship