The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes local Python scripts (check_sdd_structure.py, check_doc_artifacts.py, and count_text_size.py) to automate document validation and size analysis. These scripts are part of the skill's own distributed package and do not require external dependencies or remote calls.\n- [SAFE]: The skill enforces a mandatory Output Root Safety policy that strictly limits file operations to the repository root. It explicitly rejects absolute paths, parent directory traversal (..), and access to sensitive system directories such as .git or /etc/.\n- [SAFE]: A critical security rule in the instructions prevents the agent from building shell commands by interpolating user-provided paths, effectively mitigating the risk of command injection through user-supplied inputs.\n- [SAFE]: The skill contains no patterns associated with data exfiltration, credential harvesting, or remote code execution. All processing is local and confined to the repository's documentation and structure.

software-design-doc