The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Python's subprocess module to invoke standard command-line tools such as claude (for triggering evaluations), gitleaks (for security scanning), and lsof (to manage local server ports). These calls are localized and necessary for the skill's primary function of developing and testing other skills.
[EXTERNAL_DOWNLOADS]: The skill utilizes the uv tool to manage environment dependencies. It dynamically installs well-known Python packages like PyYAML and the anthropic SDK from standard registries to ensure validation and optimization scripts have the necessary environment. This is a standard development practice.
[DYNAMIC_EXECUTION]: A local web server is implemented in eval-viewer/generate_review.py using the standard HTTPServer library. This server provides a user interface for qualitative review of test results and manages feedback by writing to a local feedback.json file. The server is restricted to localhost and serves the user's own evaluation data.

skill-creator