The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill establishes a framework for generating brand assets based on user-provided discovery data, creating a potential surface for indirect injection if ingested data contains instructions.
Ingestion points: User-provided values for discovery variables such as {{brand-name}}, {{brand-tagline}}, and personality traits defined in references/brand-package.md.
Boundary markers: The template uses double-curly-brace interpolation within SVG, CSS, and Markdown files without explicit instructions for the agent to sanitize or escape the content.
Capability inventory: The agent is instructed to create a directory structure and write multiple files to a brand-assets/ folder in references/brand-package.md.
Sanitization: No explicit validation or sanitization logic is provided for the user-supplied strings before they are interpolated into the final output files.
[DYNAMIC_EXECUTION]: The skill generates executable styling and vector graphic content at runtime through templating.
Script generation: The agent is guided to build design-tokens.css and several SVG logo files by combining pre-defined templates with dynamic user input.
SVG Filters: The generated CSS includes a grain overlay effect utilizing an SVG feTurbulence filter for visual texture.
[OBFUSCATION]: The skill uses standard URL/percent encoding for inline assets.
Evidence: In references/brand-package.md, the grain-overlay component uses a data URI where SVG tags are URL-encoded (e.g., %3Csvg, %3E).
Context: This is standard CSS implementation for inline SVG filters and is consistent with the skill's purpose of providing visual identity tools.

elite-brand-design