The Agent Skills Directory

[SAFE]: The skill's primary function is the forensic documentation of existing SaaS visual systems. It operates strictly on the provided codebase files to extract design tokens (colors, spacing, typography) and component specifications.
[COMMAND_EXECUTION]: The skill uses standard, non-malicious shell commands such as grep, find, and ls to locate and extract styling patterns (e.g., Tailwind classes, CSS variables, and component source files). These commands are used as intended for static analysis of the UI evidence.
[DATA_EXPOSURE]: While the skill reads source code files (HTML, CSS, TSX, JSX) to extract design data, it does not access sensitive directories such as .ssh, .aws, or .env. All extracted information is stored locally in a .design-soul/ directory within the target codebase root for user review.
[EXTERNAL_DOWNLOADS]: No remote downloads or automated network requests are performed. The skill references well-known design libraries (shadcn/ui, Radix, Lucide) and accessibility tools for documentation context only.
[PROMPT_INJECTION]: The instructions in SKILL.md and sub-prompts in the references/ directory are focused on maintaining documentation accuracy and adherence to the visual system. No attempts to bypass safety filters or override agent instructions were detected.

extract-saas-design