The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes local utility scripts (scripts/check.py, scripts/skill_checks.py) for validating the skill's package contract and architecture reports. A test script (scripts/skill_checks_test.py) invokes the validator via subprocess.run using the local Python interpreter, which is a standard development practice for CLI tools and does not incorporate untrusted external input.
[DATA_EXFILTRATION]: No network activity was detected. The skill features a static analyzer (scripts/check.py) that checks for the presence of absolute host/global paths (e.g., /Users, /home, /tmp, ~/.ssh) within the skill's own files to prevent accidental information leakage or environment dependencies.
[PROMPT_INJECTION]: Instructions within the skill and its reference material (e.g., references/08-failure-modes.md) are designed to guide the agent toward rigorous evidence-based analysis and specifically discourage bypassing validation gates, suppressing failures, or accepting social flattery as a substitute for architectural evidence.
[REMOTE_CODE_EXECUTION]: No remote code execution patterns, such as downloading and piping scripts to a shell, were found. The installation script (scripts/install.sh) is a standard utility for local deployment and file management.

architecture-design