The Agent Skills Directory

[SAFE]: The skill provides instructions and references for design system governance. It promotes best practices like semantic naming, token lifecycle management, and clear separation between primitives and semantic layers. No malicious command execution or unauthorized data access patterns were found.
[INDIRECT_PROMPT_INJECTION]: As a design system tool, the skill naturally ingests external data such as design tokens and Figma sync data. While this constitutes an attack surface, the instructions emphasize structural validation against the DTCG specification and manual review of AI-generated content, which are effective mitigations for this specific use case.
Ingestion points: External token definitions in DTCG JSON, CSS, and Figma sync data (reference/figma-sync.md).
Boundary markers: Mandatory adherence to the W3C DTCG v2025.10 format and semantic intent mapping to provide structure and role-based constraints (reference/token-system.md).
Capability inventory: File system operations for writing token artifacts (CSS/JSON) and integration with standard CI/CD workflows like GitHub Actions for synchronization (reference/figma-sync.md).
Sanitization: The workflow includes human-in-the-loop validation for generated tokens and formal lifecycle gates for system changes (SKILL.md, reference/token-lifecycle.md).

muse