The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local Python scripts (search.py, core.py) to query a CSV-based design database. These scripts use standard library modules (argparse, csv, re) to provide formatted design recommendations. No arbitrary command execution or shell injection vectors were detected.
[DATA_EXPOSURE]: No sensitive data access or hardcoded credentials were found. The skill processes UI tokens (hex codes, spacing values, font names) and design principles.
[EXTERNAL_DOWNLOADS]: The documentation contains numerous references to official documentation URLs (e.g., angular.dev, tailwindcss.com, nextjs.org, developer.apple.com). These are treated as well-known/trusted sources for developer guidance. The skill instructions for installing Python (via brew, apt, or winget) are standard environment prerequisites and are clearly presented to the user.
[PROMPT_INJECTION]: Instructions include hierarchical logic where page-specific rules can 'override' master design files. This is a standard design system pattern and does not represent an attempt to bypass agent safety or system instructions.
[DYNAMIC_EXECUTION]: The design_system.py script generates Markdown and CSS content based on the selected design style. This generation is template-based and does not use eval() or exec() on untrusted input.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user queries to provide design advice. It implements boundary markers (Markdown headers, code blocks, and the 'Master + Overrides' pattern) to ensure the agent understands the context of the design rules.

ui-ux-pro-max