The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands to perform its core functions. It invokes the claude CLI to run evaluation queries and optimization loops, and uses the lsof utility to identify and manage local network ports for the result viewer.
[EXTERNAL_DOWNLOADS]: The HTML review interface loads the SheetJS library from a public Content Delivery Network to enable spreadsheet rendering. This is a legitimate use of a well-known technology service.
[COMMAND_EXECUTION]: The review server utility in eval-viewer/generate_review.py programmatically terminates local processes using os.kill to ensure it can bind to its designated network port (default 3117).
[DYNAMIC_EXECUTION]: The skill dynamically generates temporary skill definition files in the .claude/commands/ directory. These files are subsequently loaded and executed by the claude CLI to facilitate testing and optimization of new skills.

skill-creator