The Agent Skills Directory

[SAFE]: The skill incorporates extensive safety and quality controls, including a 'Safety Rules' section that prohibits hardcoding secrets, mandates WCAG 2.1 AA accessibility, and requires respect for intellectual property.
[EXTERNAL_DOWNLOADS]: The workflow involves fetching design specifications from the Figma API, which is a well-known service. The implementation instructions use environment variables for authentication and target official API endpoints.
[COMMAND_EXECUTION]: The skill utilizes standard command-line tools such as curl and unzip for processing design files. Additionally, it includes a validation script (validate_skill.py) that performs static analysis on generated code to detect accessibility and design token issues without executing the code itself.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it processes data from external design tools and screenshots to generate code implementation.
Ingestion points: Design data is ingested via Figma API responses, Sketch archive files, and image screenshots as described in SKILL.md and references/design-tokens-guide.md.
Boundary markers: The skill defines explicit quality tiers (Production, MVP, Prototype) and a detailed implementation checklist to guide the agent, though it does not specify character-based delimiters for the raw design JSON.
Capability inventory: Capabilities include network access via curl for design data and standard file system operations for asset extraction.
Sanitization: The skill emphasizes structural normalization and tokenization of ingested data rather than treating the data as executable logic.

design-to-code