The Agent Skills Directory

[OBFUSCATION]: The file demo/index.html contains a large Base64-encoded string (encodedHtml) which holds the complete HTML content for the demonstration. While this is used to package the demo payload, the use of encoding hides the underlying code from plain-text inspection.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided prompts to generate complex UI designs, as demonstrated in demo/PROMPT.md. This architecture creates a vulnerability surface where malicious instructions could potentially be injected through untrusted data input.
Ingestion points: User-supplied design prompts in demo/PROMPT.md and SKILL.md usage.
Boundary markers: Absent; there are no explicit delimiters or warnings to ignore instructions within the design data.
Capability inventory: The skill facilitates rendering of HTML/CSS layouts and execution of client-side scripts (GSAP, Three.js) as seen in demo/index.html.
Sanitization: Absent; the skill does not specify any sanitization or validation of the generated UI code.
[EXTERNAL_DOWNLOADS]: The demo references external resources from well-known technology services. These downloads are documented neutrally:
Fetches styling assets from Google Fonts (fonts.googleapis.com).
Loads animation and 3D visualization libraries from Cloudflare's CDN (cdnjs.cloudflare.com).
Uses the Tailwind CSS utility-first framework from its official CDN (cdn.tailwindcss.com).

corner-diagonals