Skip to content

Start Here

This wiki covers the layer above day-to-day design system work: the operational and architectural calls nobody hands you a manual for. Why split tokens into tiers? What makes a component governed rather than just in a library? When does AI help, and when does it add a new kind of mess?

It assumes you already know the basics of building and running a design system: not just Figma libraries, but coded components, how they ship to production, and how a team keeps them in shape.

Every claim names the person or report it came from, with a link. Where sources agree, the page says so. Where they disagree, both views stay. Ordered by how much of the wiki draws on them:

  • Murphy Trueman, design-system-ops toolkit: the backbone of the foundations and most governance pages, used as evidence for principles, not settings to copy
  • Nathan Curtis, EightShapes: team models, contribution, component APIs, token naming, cadence
  • zeroheight’s Design Systems Report and Figma’s Design Executive Council: survey data and company case studies
  • Ness Grixti: contribution, onboarding, and Wise’s multi-brand system
  • Romina Kavcic: tokens as machine-readable assets, scoping AI access
  • Brad Frost: governance decision trees and feedback loops
  • IBM Carbon, AWS Cloudscape, Atlassian, Microsoft, GitLab Pajamas: published AI interaction patterns
  • Diana Wolosin, Shane P Williams, Jan Six: documentation and context for AI agents
  • GOV.UK Design System: accessibility strategy, contribution, and community rhythm
  • Supernova: business cases, ROI, API durability, and versioning
  • Dan Mall: pilot strategy
  • Amy Hupe and Inayaili de León: getting people to contribute
  • Jina Anne: cyclical team models and shared ownership
  • Nielsen Norman Group: maturity assessment and heuristic evaluation
  • W3C: WCAG, ARIA patterns, and the design tokens spec
  • Cathy Dutton: what really deserves to be a shared pattern
  • Marianne Ashton-Booth: stakeholder mapping and three-horizon planning
  • Radix Colors and Canonical (Maximilian Blazek): accessible color scales
  • Workday Canvas, Radix Primitives, fourzerothree.in, story.to.design: subcomponents in code and in Figma
  • Braid (SEEK), React Spectrum (Adobe), and Kent C. Dodds: spacing between parts and guarding parts used outside their parent
  • Alice Packard: what makes a Figma component easy for designers to use

The Citations page collects every citation in one place.

Every page has the same shape: a lead with the main takeaway, a Key takeaways box, The problem, then Practices written as advice, so the “On this page” menu works as a checklist. Some pages add Choosing … for real alternatives or The model for a structure to understand first. Every page ends with Common mistakes.

  1. 01. Strategy: audit, pilot, take over an existing system, or assess its maturity.
  2. 02. Foundations: core concepts, token architecture and naming, accessible colors, Design-to-Code Contract, and platform divergence.
  3. 03. Components: specifications, composition in Figma and code, APIs, property naming, usability, and accessibility.
  4. 04. Governance: team and decision models, contribution, lifecycle, inventory, releases, cadence, onboarding, long-term scale, and case studies.
  5. 05. Health: adoption, performance, dependencies, and documentation coverage.
  6. 06. Business: value, ROI, brand, and stakeholder alignment.
  7. 07. AI Integration: readiness, context, documentation, governance, risk, Figma access, workflows, CI, output discipline, and UI patterns.

Reference: the glossary and citations.

Start with Core Concepts, or jump to any section from the sidebar. The glossary defines every term along the way.