The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the gemini CLI tool to perform research on current UI/UX libraries and uses the open command to launch the generated design-catalog.html file in the user's default browser.
[EXTERNAL_DOWNLOADS]: The visual catalog generated by the skill incorporates external scripts and stylesheets from well-known services, specifically Tailwind CSS via cdn.tailwindcss.com and daisyUI via cdn.jsdelivr.net.
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection due to its reliance on untrusted external data and local project files.
Ingestion points: Live research results obtained from the gemini CLI or WebSearch, as well as local configuration files like brand.yaml, tailwind.config.ts, and DESIGN_MEMORY.md (specifically within file SKILL.md).
Boundary markers: No delimiters or safety instructions are provided to the agent to distinguish between its own logic and the instructions potentially embedded in the ingested research data or project files.
Capability inventory: The skill has the ability to write files to the project root (design-catalog.html, component directories), execute shell commands (gemini, open), and invoke secondary tools (/visual-qa, /design-tokens).
Sanitization: There is no evidence of sanitization or validation performed on the research results or project file contents before they are used to generate code or HTML.

design