The Agent Skills Directory

[COMMAND_EXECUTION]: Several scripts in the scripts/ directory, including autoevolve.py, grade.py, run-eval.py, and test-triggers.py, use the subprocess module to invoke the claude or claude-code command-line interface. This is a functional requirement used to automate the testing of skill triggers and the evaluation of output quality by interacting with the agent platform.
[SAFE]: The autoevolve.py script implements an optimization loop that modifies the SKILL.md file on the local file system. It updates the instructions based on AI-generated suggestions to improve performance against user-defined test cases. This self-modification is part of the skill's intended purpose for autonomous skill evolution.
[SAFE]: The scaffold.py script performs file system operations to generate new skill directories and boilerplate files, including template Python scripts. These operations are restricted to the output path provided by the user and are typical for scaffolding tools.
[SAFE]: Reference documentation within the skill, such as references/spec.md, describes platform features like the !command syntax for dynamic context injection. This information is provided for educational purposes for skill authors and does not represent a malicious implementation within this skill.

skill-creator