Skip to content

AI-ready design systems, compared with proof

Reviewed

Work with Simbolik (opens in a new tab)
Work with Simbolik (opens in a new tab)

Sources

Sources for Best AI Design Systems

Answer in brief

Every claim in this reference rests on a public page: a design system’s own docs, its official GitHub, its npm page, its maker’s blog, or the standard a rating looks for. This page lists each source once, with what it supports and the entries that cite it, then the comparison’s 48 ratings, each linked to its proof.

What counts as a source

A source is a page published by the system’s makers or by the body that wrote a standard. Lists made by third parties never count, however well ranked, because they cannot show what a system ships. Where no public page shows a thing, the rating is No and the entry says so.

Sources describe themselves, not Simbolik

A system’s own docs support what that system offers. They do not endorse Simbolik or this reference, and a link from here does not mean its makers know of it.

Every source the entries cite

  1. Define success criteria and build evaluations, Anthropic (opens in a new tab)

    Test sets that mirror the real task with its edge cases, graded automatically where possible, with more cases preferred over fewer hand-graded ones, and a held-out set.

    Cited in A test set: prompts that measure an agent against your design system, A bounded pilot with an AI agent, Quality and checks; checked

  2. Web Content Accessibility Guidelines 2.2 (opens in a new tab)

    The contrast ratio of at least 4.5:1 for normal text that a script can check.

    Cited in A test set: prompts that measure an agent against your design system, Acceptance: checking an agent’s output against the design system; checked

  3. Primer MCP server (opens in a new tab)

    Primer’s lint_css tool, which checks CSS for Primer’s design token usage.

    Cited in A test set: prompts that measure an agent against your design system, Agent access through a design system MCP server, Acceptance: checking an agent’s output against the design system, Conformance check, GitHub Primer; checked

  4. Simbolik Design System, check your site (opens in a new tab)

    A checker that lists every value and component written by hand, with a report scripts and agents can read.

    Cited in A test set: prompts that measure an agent against your design system, Acceptance: checking an agent’s output against the design system, Conformance check, Design drift; checked

  5. Simbolik (opens in a new tab)

    The same brief given to the same model with and without a design system, shown as three pairs.

    Cited in A test set: prompts that measure an agent against your design system, Best AI design systems: the category guide, Simbolik and the best AI design systems, How the related websites connect; checked

  6. Model Context Protocol, understanding MCP servers (opens in a new tab)

    An MCP server’s tools, resources and prompts, and tools defined with typed inputs and outputs.

    Cited in Agent access through a design system MCP server, Agent access and docs, A procurement brief for an AI-ready design system, MCP server; checked

  7. Carbon MCP, onboarding and setup (opens in a new tab)

    IBM Carbon’s hosted server, its IBMid sign-in and the carbon-builder skill that goes with it.

    Cited in Agent access through a design system MCP server, How to choose a design system for AI agents, Written guidance for agents: AGENTS.md, rules files and skills, The AI-ready design system checklist, The cost and scope of making a design system AI-ready, Best AI Design Systems (2026) – AI-Ready Systems That Agents Actually Follow, How Best AI Design Systems is assessed, IBM Carbon; checked

  8. Carbon MCP on GitHub (opens in a new tab)

    The Terms of Use that govern Carbon’s hosted MCP service.

    Cited in Agent access through a design system MCP server, Licenses and usage rights: what an agent may reuse, IBM Carbon; checked

  9. Ant Design, MCP server (opens in a new tab)

    Ant Design’s official server, started with antd mcp from its CLI.

    Cited in Agent access through a design system MCP server, Ant Design; checked

  10. shadcn/ui, MCP server (opens in a new tab)

    shadcn/ui’s official server, which browses, searches and installs components.

    Cited in Agent access through a design system MCP server, shadcn/ui; checked

  11. Chakra UI MCP server (opens in a new tab)

    Chakra UI’s official server, and the Pro tools in it that need a paid license key.

    Cited in Agent access through a design system MCP server, How to choose a design system for AI agents, The cost and scope of making a design system AI-ready, Chakra UI; checked

  12. Simbolik Design System, working with AI assistants (opens in a new tab)

    The free design system’s components.json, a catalog an agent fetches rather than a live server.

    Cited in Agent access through a design system MCP server, Simbolik and the best AI design systems; checked

  13. Design Tokens Format Module 2025.10 (opens in a new tab)

    What design tokens as data look like in the W3C community group’s format.

    Cited in Best AI design systems: the category guide, What is an AI-ready design system?, Design tokens as data, in the W3C community format, Tokens and data, A procurement brief for an AI-ready design system, AI-ready design system, Best AI Design Systems (2026) – AI-Ready Systems That Agents Actually Follow, Design token, Glossary of AI design system terms, Semantic token, Sources for Best AI Design Systems, The AI design systems directory, The W3C Design Tokens format, Topics in the best AI design systems reference, W3C Design Tokens format; checked

  14. Model Context Protocol (opens in a new tab)

    The open protocol behind the agent access the comparison looks for.

    Cited in Best AI design systems: the category guide, AI-ready design system, Best AI Design Systems (2026) – AI-Ready Systems That Agents Actually Follow, How Best AI Design Systems is assessed, Topics in the best AI design systems reference; checked

  15. The llms.txt proposal (opens in a new tab)

    The plain-text index of docs the comparison looks for.

    Cited in Best AI design systems: the category guide, What is an AI-ready design system?, Docs AI agents can read: llms.txt and Markdown pages, Agent access and docs, AI-ready design system, Best AI Design Systems (2026) – AI-Ready Systems That Agents Actually Follow, llms.txt, llms.txt, The complete library of entries; checked

  16. AGENTS.md (opens in a new tab)

    The open format for written guidance to coding agents.

    Cited in Best AI design systems: the category guide, What is an AI-ready design system?, Agent access and docs, Written guidance for agents: AGENTS.md, rules files and skills, A procurement brief for an AI-ready design system, AGENTS.md, AGENTS.md, AI-ready design system, Best AI Design Systems (2026) – AI-Ready Systems That Agents Actually Follow, Questions and answers about the best AI design systems; checked

  17. Atlassian Design System, get started with AI (opens in a new tab)

    A public design system’s own page on how agents should build with it, one of the four that rate Yes on every count.

    Cited in Best AI design systems: the category guide, Selection and fit, How to choose a design system for AI agents, Atlassian Design System; checked

  18. AI Systems, Simbolik (opens in a new tab)

    Simbolik’s own description of the custom AI-ready design systems it builds, and its site built by an AI agent.

    Cited in Best AI design systems: the category guide, Selection and fit, A bounded pilot with an AI agent, Simbolik and the best AI design systems, Acceptance: checking an agent’s output against the design system, Licenses and usage rights: what an agent may reuse, A procurement brief for an AI-ready design system, The cost and scope of making a design system AI-ready, AI-ready design system, Questions and answers about the best AI design systems, The Simbolik entity map; checked

  19. Carbon’s theme tokens, white theme (opens in a new tab)

    IBM Carbon’s design tokens as a .tokens.json file that points at the 2025.10 schema.

    Cited in Design tokens as data, in the W3C community format, IBM Carbon, The W3C Design Tokens format; checked

  20. @adobe/spectrum-tokens on the npm registry (opens in a new tab)

    Adobe Spectrum’s design tokens published as a JSON package.

    Cited in Design tokens as data, in the W3C community format; checked

  21. Primer Primitives on GitHub (opens in a new tab)

    GitHub Primer’s design tokens, published as JSON in @primer/primitives.

    Cited in Design tokens as data, in the W3C community format; checked

  22. shadcn/ui, theming (opens in a new tab)

    shadcn/ui’s theme kept as CSS variables rather than a design tokens file.

    Cited in Design tokens as data, in the W3C community format, How to choose a design system for AI agents, shadcn/ui; checked

  23. Atlassian’s llms.txt (opens in a new tab)

    Atlassian’s index, beside its llms-full.txt and split files for design tokens and components.

    Cited in Docs AI agents can read: llms.txt and Markdown pages, Atlassian Design System; checked

  24. Ant Design’s llms.txt (opens in a new tab)

    Ant Design’s index, beside its llms-full.txt and Markdown pages.

    Cited in Docs AI agents can read: llms.txt and Markdown pages, Ant Design; checked

  25. React Spectrum’s llms.txt (opens in a new tab)

    Adobe’s index for React Spectrum, whose every page also opens as .md.

    Cited in Docs AI agents can read: llms.txt and Markdown pages, Adobe Spectrum 2; checked

  26. Polaris docs as Markdown, Shopify (opens in a new tab)

    Shopify’s pages as Markdown by adding .md, with no llms.txt.

    Cited in Docs AI agents can read: llms.txt and Markdown pages, Shopify Polaris; checked

  27. Primer’s llms.txt address (opens in a new tab)

    Cited as the proof that it is missing. The address answers 404, so Primer publishes no llms.txt.

    Cited in Docs AI agents can read: llms.txt and Markdown pages, GitHub Primer; checked

  28. GOV.UK Design System’s llms.txt address (opens in a new tab)

    Cited as the proof that it is missing. The address answers 404, so the GOV.UK Design System publishes no llms.txt.

    Cited in Docs AI agents can read: llms.txt and Markdown pages, GOV.UK Design System; checked

  29. About Simon Martin, Simbolik (opens in a new tab)

    Simon Martin’s own account of his work, the studio and the AMPECO design system.

    Cited in Simbolik and the best AI design systems, About Best AI Design Systems; checked

  30. Simbolik Design System (opens in a new tab)

    The free design system’s version and its 110 components.

    Cited in Simbolik and the best AI design systems; checked

  31. Simbolik Design System’s llms.txt (opens in a new tab)

    The free design system’s DTCG design tokens, its code under MIT and its Figma file under CC BY 4.0.

    Cited in Simbolik and the best AI design systems, Licenses and usage rights: what an agent may reuse; checked

  32. Simbolik Design System on GitHub (opens in a new tab)

    The free design system’s public code, and skills for AI agents listed as what comes next.

    Cited in Simbolik and the best AI design systems; checked

  33. Fluent UI discussion 35732 (opens in a new tab)

    A Fluent maintainer, 2026-02-25, saying the Fluent MCP server has not been released outside Microsoft.

    Cited in How to choose a design system for AI agents, The AI-ready design system checklist, Microsoft Fluent 2 (React v9); checked

  34. Material Web on GitHub (opens in a new tab)

    Material Web’s own README, which says it is in maintenance mode pending new maintainers.

    Cited in How to choose a design system for AI agents, Google Material 3 (web); checked

  35. The MIT License, Open Source Initiative (opens in a new tab)

    MIT’s condition that the copyright and permission notice be included in all copies.

    Cited in Licenses and usage rights: what an agent may reuse; checked

  36. Apache License 2.0, Apache Software Foundation (opens in a new tab)

    Apache 2.0’s condition that anyone receiving the work gets a copy of the license.

    Cited in Licenses and usage rights: what an agent may reuse; checked

  37. GOV.UK Design System, typeface (opens in a new tab)

    GDS Transport is for services on service.gov.uk; services on other domains use another typeface.

    Cited in Licenses and usage rights: what an agent may reuse, GOV.UK Design System; checked

  38. Polaris icons license, in the @shopify/polaris-icons package (opens in a new tab)

    The icons may only be used for apps that work with Shopify, and stand-alone apps must look distinct from Shopify’s products.

    Cited in Licenses and usage rights: what an agent may reuse, Shopify Polaris; checked

  39. SIL Open Font License 1.1 (opens in a new tab)

    Fonts under the OFL may be bundled and embedded, may not be sold by themselves, and changed versions may not use a Reserved Font Name.

    Cited in Licenses and usage rights: what an agent may reuse; checked

  40. IBM Plex license (opens in a new tab)

    IBM Plex, Carbon’s typeface, under the OFL with Plex as a Reserved Font Name.

    Cited in Licenses and usage rights: what an agent may reuse; checked

  41. Material design icons on GitHub (opens in a new tab)

    Google’s Material icons offered under Apache 2.0.

    Cited in Licenses and usage rights: what an agent may reuse; checked

  42. @salesforce-ux/design-tokens on the npm registry (opens in a new tab)

    The package’s license field, which points to a LICENSE.txt file.

    Cited in Licenses and usage rights: what an agent may reuse; checked

  43. Ant Design, for agents (opens in a new tab)

    Ant Design’s ready prompt for agents and its installable skill.

    Cited in Written guidance for agents: AGENTS.md, rules files and skills, Ant Design; checked

  44. Chakra UI, skills (opens in a new tab)

    Chakra UI’s installable skills for agents.

    Cited in Written guidance for agents: AGENTS.md, rules files and skills, Chakra UI; checked

  45. Primer’s design tokens guide (opens in a new tab)

    A guide shipped in Primer’s package and written for language models, with rules such as never using raw values.

    Cited in Written guidance for agents: AGENTS.md, rules files and skills, Design drift, GitHub Primer; checked

  46. Fluent UI’s AGENTS.md (opens in a new tab)

    Guidance written for contributors to Fluent’s own code, the reason Fluent rates Partly on guidance.

    Cited in Written guidance for agents: AGENTS.md, rules files and skills, The AI-ready design system checklist, AGENTS.md, Microsoft Fluent 2 (React v9); checked

  47. TypeScript handbook, everyday types (opens in a new tab)

    What typed props mean, for the second question.

    Cited in The AI-ready design system checklist; checked

  48. @fluentui/react-theme on the npm registry (opens in a new tab)

    Fluent 2 publishes its design tokens as a package, for the worked example's first question.

    Cited in The AI-ready design system checklist, Microsoft Fluent 2 (React v9); checked

  49. @fluentui/react-components on the npm registry (opens in a new tab)

    The package names its TypeScript declarations in dist/index.d.ts, for the worked example's second question.

    Cited in The AI-ready design system checklist, Microsoft Fluent 2 (React v9); checked

  50. Fluent UI React v9, llms.txt (opens in a new tab)

    Fluent’s own llms.txt, which links a plain-text file for each docs page, for the worked example’s fourth question.

    Cited in The AI-ready design system checklist, Microsoft Fluent 2 (React v9); checked

  51. @adobe/spectrum-tokens on npm (opens in a new tab)

    The Design tokens rating, Yes; a published JSON package.

    Cited in Adobe Spectrum 2; checked

  52. React Spectrum, working with AI (opens in a new tab)

    The Agent access and Guidance ratings, Yes; the MCP server and the agent skills.

    Cited in Adobe Spectrum 2; checked

  53. React Spectrum Button, as Markdown (opens in a new tab)

    The Typed props rating, Yes; the component’s API table with types and defaults.

    Cited in Adobe Spectrum 2, Component API; checked

  54. Spectrum Web Components, build with AI (opens in a new tab)

    The guidance for the web components, which shares the Yes.

    Cited in Adobe Spectrum 2; checked

  55. AGENTS.md on GitHub (opens in a new tab)

    The format’s repository and its MIT license.

    Cited in AGENTS.md; checked

  56. SLDS starter kit for Lightning Web Components (opens in a new tab)

    An AGENTS.md aimed at building UI with a design system.

    Cited in AGENTS.md, Salesforce Lightning Design System (SLDS); checked

  57. Ant Design’s design.md (opens in a new tab)

    The Design tokens rating, Yes; design tokens in YAML, marked alpha.

    Cited in Ant Design; checked

  58. Ant Design CLI (opens in a new tab)

    The `antd token` and `antd info` commands, and JSON output of the design tokens.

    Cited in Ant Design; checked

  59. antd on the npm registry (opens in a new tab)

    The Typed props rating, Yes; the package’s TypeScript types.

    Cited in Ant Design; checked

  60. @atlaskit/tokens 20.9.1, file list (opens in a new tab)

    The Design tokens rating, Yes; JSON files of design tokens, not in the W3C format.

    Cited in Atlassian Design System; checked

  61. Atlassian’s llms-full.txt (opens in a new tab)

    The full text of the docs in one file, and its instruction to use only the design tokens it lists.

    Cited in Atlassian Design System; checked

  62. Atlassian’s components file for language models (opens in a new tab)

    The Typed props rating, Yes; a props table per component.

    Cited in Atlassian Design System; checked

  63. Color, Simbolik Design System (opens in a new tab)

    Colors named by role, among them the brand fill, brand and on-brand text and the brand edge, over raw ramps that never reach CSS.

    Cited in Brand roles; checked

  64. Spectrum design tokens, Adobe (opens in a new tab)

    Semantic tokens that communicate purpose, over primitives used through them.

    Cited in Brand roles, Semantic token; checked

  65. Chakra UI theming, for language models (opens in a new tab)

    The Design tokens rating, Partly; design tokens as theme objects in the package.

    Cited in Chakra UI; checked

  66. Chakra UI’s llms.txt (opens in a new tab)

    The Docs rating, Yes.

    Cited in Chakra UI; checked

  67. @chakra-ui/react on the npm registry (opens in a new tab)

    The Typed props rating, Yes; the package’s TypeScript types.

    Cited in Chakra UI; checked

  68. Passing props to a component, React (opens in a new tab)

    Props as the information a parent passes to a component.

    Cited in Component API; checked

  69. ensure-design-token-usage, Atlassian ESLint plugin (opens in a new tab)

    A lint rule that enforces design tokens rather than hard-coded values.

    Cited in Conformance check, Design drift; checked

  70. Guide to variables in Figma (opens in a new tab)

    What variables store, and the plans that can publish them to a library.

    Cited in Figma variables and Code Connect; checked

  71. Modes for variables (opens in a new tab)

    Importing and exporting variables as JSON in the W3C community format, and the plans that can.

    Cited in Figma variables and Code Connect; checked

  72. Figma Variables REST API (opens in a new tab)

    The API’s condition, a full seat in an Enterprise org.

    Cited in Figma variables and Code Connect; checked

  73. Code Connect (opens in a new tab)

    What Code Connect does, its plans and seats, and the one framework a Professional plan keeps.

    Cited in Figma variables and Code Connect; checked

  74. Code Connect and MCP (opens in a new tab)

    What Code Connect adds to the context the MCP server sends an agent.

    Cited in Figma variables and Code Connect; checked

  75. Figma MCP server, tools and prompts (opens in a new tab)

    The `get_variable_defs` and `get_code_connect_map` tools.

    Cited in Figma variables and Code Connect; checked

  76. Figma MCP server, rate limits and access (opens in a new tab)

    The read limits on each plan.

    Cited in Figma variables and Code Connect; checked

  77. Code Connect on GitHub (opens in a new tab)

    Its MIT license, and template files as the one maintained way to use it.

    Cited in Figma variables and Code Connect; checked

  78. @primer/primitives on npm (opens in a new tab)

    The Design tokens rating, Yes; JSON design tokens in a published package.

    Cited in GitHub Primer; checked

  79. @primer/react on npm (opens in a new tab)

    The Typed props rating, Yes; the package ships its types.

    Cited in GitHub Primer; checked

  80. Material Web theming (opens in a new tab)

    The Design tokens rating, Partly; Sass and CSS variables.

    Cited in Google Material 3 (web); checked

  81. Material Web docs on GitHub (opens in a new tab)

    The Docs rating, Partly; component docs as Markdown in the repository.

    Cited in Google Material 3 (web); checked

  82. Material Web button docs (opens in a new tab)

    The Typed props rating, Yes; the component’s API section.

    Cited in Google Material 3 (web); checked

  83. govuk-frontend on the npm registry (opens in a new tab)

    The Design tokens rating, Partly; Sass settings and CSS, no design token JSON.

    Cited in GOV.UK Design System; checked

  84. govuk-frontend-mcp, a community server (opens in a new tab)

    The Agent access rating, No; the one server found describes itself as unofficial.

    Cited in GOV.UK Design System; checked

  85. GOV.UK Frontend on GitHub (opens in a new tab)

    The Guidance rating, No; no AGENTS.md or rules file in the repository.

    Cited in GOV.UK Design System; checked

  86. GOV.UK Design System, button, Nunjucks options (opens in a new tab)

    The Typed props rating, Partly; a documented options table, no TypeScript types.

    Cited in GOV.UK Design System; checked

  87. Carbon’s llms.txt (opens in a new tab)

    The Docs rating, Yes.

    Cited in IBM Carbon; checked

  88. @carbon/react on the npm registry (opens in a new tab)

    The Typed props rating, Yes; version 1.120.0 ships its TypeScript types.

    Cited in IBM Carbon; checked

  89. The llms.txt proposal, as Markdown (opens in a new tab)

    The format of the file, the Markdown copies of pages, and which file an agent uses.

    Cited in llms.txt; checked

  90. llms.txt changes (opens in a new tab)

    What version 2 changed, in August 2026.

    Cited in llms.txt; checked

  91. llmstxt.org’s own llms.txt (opens in a new tab)

    The proposal describing itself as a proposal.

    Cited in llms.txt; checked

  92. llms-txt on GitHub (opens in a new tab)

    The proposal’s repository and its Apache 2.0 license.

    Cited in llms.txt; checked

  93. MCP transports (opens in a new tab)

    The two transports, stdio and Streamable HTTP.

    Cited in MCP server, The Model Context Protocol (MCP); checked

  94. Plausible Analytics, data policy (opens in a new tab)

    What Plausible collects and what it leaves out.

    Cited in Privacy and contact; checked

  95. @salesforce-ux/design-tokens 4.1.0, file list (opens in a new tab)

    The Design tokens rating, Yes; JSON, CSS, Swift and Kotlin files.

    Cited in Salesforce Lightning Design System (SLDS); checked

  96. End of life for the Salesforce DX MCP server (opens in a new tab)

    The Agent access rating, No; the server's end of life on November 2, 2026, and skills in its place.

    Cited in Salesforce Lightning Design System (SLDS); checked

  97. SLDS tools in the Salesforce DX MCP docs (opens in a new tab)

    The one beta SLDS tool left on the page, which it says a skill supersedes.

    Cited in Salesforce Lightning Design System (SLDS); checked

  98. Salesforce’s llms file for Lightning Web Components (opens in a new tab)

    The Docs rating, Partly; SLDS pages as Markdown on the developer portal.

    Cited in Salesforce Lightning Design System (SLDS); checked

  99. SLDS AI Development Skills (opens in a new tab)

    The Guidance rating, Yes; the three SLDS skills on the SLDS site itself.

    Cited in Salesforce Lightning Design System (SLDS); checked

  100. Salesforce’s llms file for Lightning base components (opens in a new tab)

    The Typed props rating, Partly; the base components’ documented attributes.

    Cited in Salesforce Lightning Design System (SLDS); checked

  101. Chakra UI semantic tokens (opens in a new tab)

    Semantic tokens designed for a specific context, referencing a base token.

    Cited in Semantic token; checked

  102. shadcn/ui’s llms.txt (opens in a new tab)

    The Docs rating, Yes.

    Cited in shadcn/ui; checked

  103. shadcn/ui skills (opens in a new tab)

    The Guidance rating, Yes; the shadcn skill and how it is installed.

    Cited in shadcn/ui; checked

  104. shadcn/ui Button, as Markdown (opens in a new tab)

    The Typed props rating, Yes; the API reference of a TypeScript component.

    Cited in shadcn/ui; checked

  105. @shopify/polaris-tokens on npm (opens in a new tab)

    The Design tokens rating, Partly; the package belongs to the deprecated React library.

    Cited in Shopify Polaris; checked

  106. Shopify Dev MCP server (opens in a new tab)

    The Agent access rating, Yes; one server for all of Shopify.

    Cited in Shopify Polaris; checked

  107. Shopify AI Toolkit (opens in a new tab)

    The Guidance rating, Yes; the shopify skill and its Polaris topics.

    Cited in Shopify Polaris; checked

  108. @shopify/polaris-types on npm (opens in a new tab)

    The Typed props rating, Yes; the types of the web components.

    Cited in Shopify Polaris; checked

  109. Using Storybook with AI (opens in a new tab)

    The AI features in preview, the MCP addon and how it is installed.

    Cited in Storybook; checked

  110. Storybook’s feature lifecycle (opens in a new tab)

    What a preview feature is.

    Cited in Storybook; checked

  111. Storybook MCP overview (opens in a new tab)

    The docs, development and testing tools, the frameworks with a manifest, and the AGENTS.md advice.

    Cited in Storybook; checked

  112. Storybook manifests (opens in a new tab)

    The components manifest, how props are read, and that it is off by default.

    Cited in Storybook; checked

  113. Storybook AI setup (opens in a new tab)

    The setup that works only with React and Vite today.

    Cited in Storybook; checked

  114. Storybook’s llms.txt (opens in a new tab)

    Storybook’s own docs index, and every docs page as Markdown.

    Cited in Storybook; checked

  115. @storybook/addon-mcp on the npm registry (opens in a new tab)

    The addon’s MIT license and version.

    Cited in Storybook; checked

  116. Style Dictionary on GitHub (opens in a new tab)

    What it does and the platforms it writes to, its Apache 2.0 license.

    Cited in Style Dictionary; checked

  117. Style Dictionary and the DTCG format (opens in a new tab)

    Support for the W3C format since version 4, and the note that 2025.10 is not fully supported yet.

    Cited in Style Dictionary, W3C Design Tokens format; checked

  118. Style Dictionary 5.7.0 release notes (opens in a new tab)

    Support added for 2025.10 duration values.

    Cited in Style Dictionary; checked

  119. Style Dictionary on the npm registry (opens in a new tab)

    Version 5.8.0 and its Apache 2.0 license.

    Cited in Style Dictionary; checked

  120. Style Dictionary, design tokens (opens in a new tab)

    The two formats that cannot be combined in one run.

    Cited in Style Dictionary; checked

  121. Carbon Design System (opens in a new tab)

    The first system listed, from its own site.

    Cited in The AI design systems directory; checked

  122. What is the Model Context Protocol? (opens in a new tab)

    MCP as an open-source standard for connecting AI applications to external systems.

    Cited in The Model Context Protocol (MCP); checked

  123. Introducing the Model Context Protocol, Anthropic (opens in a new tab)

    Anthropic open-sourcing MCP, November 25, 2024.

    Cited in The Model Context Protocol (MCP); checked

  124. Donating MCP and establishing the Agentic AI Foundation, Anthropic (opens in a new tab)

    The donation to the Agentic AI Foundation under the Linux Foundation, December 9, 2025.

    Cited in The Model Context Protocol (MCP); checked

  125. MCP joins the Agentic AI Foundation (opens in a new tab)

    MCP’s own account of the move, and its maintainer structure staying the same.

    Cited in The Model Context Protocol (MCP); checked

  126. MCP governance (opens in a new tab)

    The project as a Series of LF Projects, its maintainers and steering group, and the licenses for code, specification and docs.

    Cited in The Model Context Protocol (MCP); checked

  127. MCP versioning (opens in a new tab)

    The current protocol version, 2026-07-28.

    Cited in The Model Context Protocol (MCP); checked

  128. MCP specification 2026-07-28 (opens in a new tab)

    What a server offers, and the rules on consent and untrusted tool descriptions.

    Cited in The Model Context Protocol (MCP); checked

  129. MCP specification changelog (opens in a new tab)

    What the 2026-07-28 version removed.

    Cited in The Model Context Protocol (MCP); checked

  130. Design Tokens Community Group (opens in a new tab)

    The first stable version, 2025.10, and the tools the group says ship the format.

    Cited in The W3C Design Tokens format; checked

  131. Design Tokens Community Group at the W3C (opens in a new tab)

    The Format, Color and Resolver modules published on October 28, 2025, with their licensing commitments.

    Cited in The W3C Design Tokens format; checked

  132. Tokens Studio (opens in a new tab)

    What the company makes, the plugin and the Studio platform.

    Cited in Tokens Studio; checked

  133. Tokens Studio, token format setting (opens in a new tab)

    The choice of W3C DTCG or Legacy, and Legacy as the default.

    Cited in Tokens Studio, W3C Design Tokens format; checked

  134. Tokens Studio, the paid features (opens in a new tab)

    What works without a paid plan, and what needs one.

    Cited in Tokens Studio; checked

  135. Tokens Studio pricing (opens in a new tab)

    The Team plan that holds the paid plugin features.

    Cited in Tokens Studio; checked

  136. Tokens Studio, export to Figma (opens in a new tab)

    Exporting 21 types of design token to Figma styles or variables, and the units some exports drop.

    Cited in Tokens Studio; checked

  137. Tokens Studio with Style Dictionary (opens in a new tab)

    The extra transform step its files need in Style Dictionary.

    Cited in Tokens Studio; checked

  138. @tokens-studio/sd-transforms on the npm registry (opens in a new tab)

    The transform package for Style Dictionary, under MIT.

    Cited in Tokens Studio; checked

  139. Tokens Studio for Figma, license (opens in a new tab)

    The plugin’s MIT license.

    Cited in Tokens Studio; checked

  140. Using TypeScript, React (opens in a new tab)

    Types for a component’s props, used for correctness checks and inline docs in editors.

    Cited in Typed props; checked

  141. Object types, TypeScript handbook (opens in a new tab)

    How TypeScript names the shape of an object, as an interface or a type.

    Cited in Typed props; checked

The comparison’s proof

Sources for this entry

  1. Design Tokens Format Module 2025.10 (opens in a new tab)

    The standard behind the Tokens column.

    designtokens.org, checked

Continue this reading path

How the related websites connect

Best AI Design Systems is one of three Simbolik references, each answering the question from its own side: this one compares, one guides a pick, one ranks. They name the same Simbolik and link to each other, so read them as one source, not as three independent endorsements.

Every related website

Cite this entry

Best AI Design Systems. “Sources for Best AI Design Systems.” Reviewed . https://bestaidesignsystems.com/sources/