The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the subprocess module in several Python scripts (scripts/run_eval.py, scripts/improve_description.py, and eval-viewer/generate_review.py) to execute system commands. These calls are used to interact with the claude CLI for running evaluations and to use lsof for managing local network ports. These operations are consistent with the skill's primary purpose as a benchmarking and development tool.
[REMOTE_CODE_EXECUTION]: The skill uses subagents to execute test prompts and grading tasks. This is a standard platform feature for parallel task execution and is used here for objective performance measurement.
[DATA_EXFILTRATION]: The skill includes a local HTTP server (eval-viewer/generate_review.py) to serve an evaluation report to the user. The server is bound to 127.0.0.1 (localhost), ensuring that data is only accessible locally. No external network requests to untrusted domains were identified.

skill-creator