The Agent Skills Directory

[OBFUSCATION]: The file demo/index.html contains a large Base64-encoded string (encodedHtml) that hides the complete HTML and JavaScript source for the design demo from plain-text scanning.
[DYNAMIC_EXECUTION]: The skill implementation in demo/index.html uses JavaScript to decode the Base64 content at runtime and inject it into an iframe's srcdoc attribute for execution.
[EXTERNAL_DOWNLOADS]: The demo implementation fetches configuration and runtime scripts from well-known services, including the Tailwind CSS CDN and Iconify's public registry. It also fetches typography from Google Fonts and Fontshare.
[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface for indirect prompt injection by encouraging the agent to process data from external reference URLs (e.g., from neuform.ai).
Ingestion points: External design reference URLs are ingested via the instructions in demo/PROMPT.md.
Boundary markers: No specific delimiters or safety warnings are present to isolate the agent from instructions potentially embedded in the fetched external design content.
Capability inventory: The skill allows the agent to generate and execute full HTML and JavaScript implementations.
Sanitization: No evidence of sanitization or filtering is provided for the content retrieved from external sources.

blue-cloudy-clean-modern