The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest external project data—specifically 'Page Specifications', 'Trigger Maps', and 'Design System' files (e.g., in output_folder/C-Scenarios/ and docs/D-Design-System/)—to drive prompt generation for asset creation and tool calls to Figma.
Ingestion points: Project specifications (.md), Trigger Maps (.md), and Design System tokens (.md/.yaml) located in the configured project output folder.
Boundary markers: The skill employs a disciplined multi-model framework (the 'Content Creation Workshop') that uses structured YAML blocks for data synthesis and refinement, which acts as a structural boundary.
Capability inventory: The skill invokes MCP tools like mcp2_import-html to inject generated HTML/CSS into Figma and performs file writes to the {output_folder}/E-Assets/ directory.
Sanitization: The workflow includes a 'Generate and Validate' step (step-04-generate-validate.md) that checks HTML structure, styling compatibility (e.g., hex conversion, pixel mapping), and content completeness before tool execution.
[EXTERNAL_DOWNLOADS]: The documentation and integration guides (data/mcp-server-integration.md, data/figma-mcp-integration.md) recommend the installation of external NPM packages to facilitate the Figma-to-code pipeline.
Packages: Recommends installing @modelcontextprotocol/server-figma and @wds/figma-mcp-server via NPM/npx.
Trust: These packages are part of the intended architecture for the Whiteport Design System (WDS) and the Model Context Protocol (MCP) ecosystem.

wds-6-asset-generation