The Agent Skills Directory

[COMMAND_EXECUTION]: The check.py script generates shell command strings (e.g., rm, ln, mv, cp) to assist users in fixing directory inconsistencies. The script correctly uses shlex.quote() to sanitize all path variables, preventing command injection if directory names contain special characters. The skill instructions direct the agent to execute the audit script and report the output for user review, rather than performing silent modifications.
[DYNAMIC_EXECUTION]: The tests/test_check.py file uses the importlib library to dynamically load the check.py script for unit testing. This is a standard and appropriate implementation for testing standalone scripts without package installation.
[INDIRECT_PROMPT_INJECTION]: The skill audits directory names in .agents/skills and .claude/skills to detect layout issues. While processing external filenames represents a theoretical attack surface for indirect prompt injection, the risk is minimal as the skill does not read file content and uses explicit instructional boundaries in SKILL.md (e.g., <goal> and reporting rules) to ensure neutral output.
[COMMAND_EXECUTION]: The unit tests in tests/test_check.py use subprocess.run() to execute the check.py script being tested. This use case is isolated to the development/test environment and uses safe argument passing (list format) rather than a shell string.

skill-link-check