The Agent Skills Directory

[PROMPT_INJECTION] (SAFE): No prompt injection patterns were detected in the documentation or instructions. The language used in examples describes standard operational behavior without attempting to override safety filters.- [DATA_EXFILTRATION] (SAFE): The script reads local files such as .git/HEAD and package.json to gather project metadata. It does not contain network calls (e.g., curl, requests) and no data is exfiltrated to external endpoints.- [COMMAND_EXECUTION] (SAFE): The script generates and modifies configuration files and adds scripts to package.json. It does not execute any system commands via subprocess or os.system; execution of the generated scripts is left to the user.- [EXTERNAL_DOWNLOADS] (SAFE): The documentation suggests installing well-known industry packages (Chromatic, Percy) and provides GitHub Action templates from trusted sources (chromaui/action). The script itself does not perform any automated downloads.

visual-regression