W3C publishes Design Tokens 2.0 draft to standardize token semantics for AI workflows

Design · 3 min read

Design Tokens 2.0 builds on prior token formats by adding fields that express token intent, derivation (for AI-generated tokens), and a confidence score indicating how reliably a token matches documented brand constraints. The draft also standardizes transforms for platform-specific conversions.

The spec includes guidance on representing dynamic tokens that respond to context (e.g., density-aware spacing or accessibility-driven color adjustments) so design systems and models can interpret intent consistently. It also suggests audit fields for tracking which tools or models created or modified tokens.

W3C is soliciting feedback from tool vendors, design teams, and AI providers to refine the draft ahead of a formal recommendation. Proponents say the standard will make design-token-driven AI generation safer and more predictable across ecosystems.