The Agent Skills Directory

[SAFE]: The skill is a meta-utility for improving agent skill quality, focusing on predictability, hierarchy, and efficiency. It does not contain any malicious code, obfuscation, or data exfiltration patterns.
[COMMAND_EXECUTION]: The skill includes a local validation script (check.sh) designed to be run by the agent. A technical audit of the script confirms it uses standard Unix utilities (grep, sed, find, wc) to perform static analysis on local files. The script is properly quoted to prevent shell injection and does not perform any network operations or execute third-party scripts found within the target directory.
[PROMPT_INJECTION]: The instructions in SKILL.md and the glossary files are focused on technical best practices. No patterns were found that attempt to bypass AI safety filters or override system-level constraints.
[EXTERNAL_DOWNLOADS]: While the skill contains a reference to an external source in its metadata (metadata.derived-from), it does not programmatically download or execute external content. The check.sh script specifically identifies and ignores external HTTP/HTTPS links during its link-integrity check.

writing-skills