The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external design data from Figma which may contain untrusted content or instructions embedded in layer names or metadata.
Ingestion points: Figma MCP tools figma_get_metadata and figma_get_design_context (SKILL.md Phase 1).
Boundary markers: The skill requires the agent to generate and present 'Decomposition Tables' (Table A and B) for user confirmation before proceeding to code generation (SKILL.md Phase 1g).
Capability inventory: The skill can execute shell commands via npx (Phase 0, 3), perform network requests/file writes via curl (references/asset-extraction.md), and execute JavaScript in a browser context via playwright_browser_evaluate (references/validation-patterns.md).
Sanitization: No explicit automated sanitization of Figma metadata is described; the primary mitigation is the mandatory user review of the design plan.
[COMMAND_EXECUTION]: The skill instructs the agent to execute various CLI commands including npx for project scaffolding and configuration, and curl for downloading assets. These commands target official tools from the vendor (Ignite UI) and well-known services (Figma).
[DYNAMIC_EXECUTION]: The skill uses playwright_browser_evaluate to execute dynamically generated JavaScript function strings within a browser environment to measure and validate UI elements (references/validation-patterns.md).
[EXTERNAL_DOWNLOADS]: The skill downloads project templates and MCP server packages via npx and fetches image assets from Figma's official API using curl (references/asset-extraction.md). These sources are well-known and reputable services.

igniteui-angular-figma-to-app