Produces a scored comparison of available DESIGN.md design systems against a project's requirements and delivers a recommended selection with adaptation notes. Use when: 'which design system fits this project', 'pick a design reference for me', 'match a design system to this brand', 'compare design system options for this client'.
Design systems are not interchangeable. Stripe's authority differs from Notion's warmth
differs from Linear's precision. Each embeds a philosophy about who the user is, what
they value, and how they should feel while working.
The selector prevents the common mistake of picking a design system because it's popular
rather than because it fits. A fintech dashboard wearing Notion's soft pastels sends
the wrong message. A community platform dressed in Stripe's sharp minimalism feels cold.
Fit is not aesthetic preference -- it is alignment between brand intent and visual
language. This skill makes that alignment measurable.
II. When to Use
Starting a new UI project and choosing visual direction
Evaluating whether a project's current design tokens match its brand
Comparing multiple design system options for a client presentation
Migrating between design systems (need to know what you're moving toward)
Building a custom design system by compositing traits from multiple references
Do NOT use when:
The client has an existing brand guide with defined tokens (use those directly)
Verwandte Skills
You need to implement a specific design system (use the system's docs instead)
You are doing a visual redesign within an existing system (audit the system instead)
The project is a prototype where visual polish doesn't matter yet
III. Workflow
Step 1: Inventory Available DESIGN.md Files
Scan the design system reference library for available options:
Primary location: HTMLCraftStudio/templates/design-systems/
Structure: each system has a DESIGN.md with tokens, philosophy, and usage notes
Currently imported (8 systems):
System
Personality
Primary Use Case
Linear
Precise, focused, developer-tool aesthetic
Task management, dev tools
Stripe
Authoritative, clean, trust-building
Fintech, payments, enterprise
Vercel
Minimal, performant, developer-first
Developer platforms, deployment
Notion
Warm, flexible, content-first
Productivity, wikis, collaboration
Resend
Modern, bold, API-developer aesthetic
Email infrastructure, dev APIs
Supabase
Approachable, open-source-friendly, green
Backend-as-a-service, databases
Cal.com
Open, scheduling-focused, clean
Booking, calendars, scheduling
Mintlify
Documentation-native, readable, structured
Developer docs, API references
50+ additional systems available from VoltAgent/awesome-design-md for import.
Create a project DESIGN.md that references the source system and documents any adaptations
Validate the applied tokens render correctly in a sample component
IV. Best Practices
Score before you feel. Gut reactions to design systems are valid but unreliable. Score first, then see if your intuition agrees. If they diverge, investigate why.
Adaptation is expected. No reference system will be a 100% match. The goal is 80%+ alignment with clear, documented modifications for the rest.
Consider the user, not the builder. A developer building a consumer app should pick for the consumer, not for their own aesthetic preferences.
Check dark mode compatibility. Some design systems have excellent dark mode support; others assume light-only. If dark mode is required, this is a hard filter.
Typography is the hardest to change. Colors are easy to swap. Spacing is mechanical. Typography choices cascade into line-height, measure, hierarchy, and readability. Pick a system whose type approach you can live with.
V. Quality Checklist
Available DESIGN.md files inventoried with current count
Project requirements classified across all 5 dimensions with justifications
Scoring matrix completed for all available systems
Dimension weighting applied if project has clear priorities
Top 3 matches presented with fit reasoning and divergence notes
Selected system's tokens copied to project
CSS custom properties generated from tokens
Project DESIGN.md created with source reference and adaptations
Sample component rendered to validate token application
VI. Common Pitfalls
Popularity bias. Picking Stripe's design system because Stripe is a good company, not because Stripe's visual language fits your product. Popularity is not fit.
Ignoring density requirements. A spacious editorial system applied to a data dashboard wastes 40% of screen real estate. Density is a hard constraint, not a preference.
Skipping adaptation documentation. If you modify 5 tokens from the source system but don't document which ones, the next developer will re-import the original and break your adaptations.
Single-dimension matching. "They use Inter, we use Inter, so they're a match" -- typography is one of five dimensions. A single match doesn't make a system fit.
Forgetting interaction patterns. Two systems can have identical colors and fonts but radically different interaction patterns (hover states, transitions, feedback). Interaction is a dimension, not a detail.
VII. Example
Project: Internal analytics dashboard for a data engineering team.
Requirements classification:
Color: Cool (trust, technical precision)
Typography: Geometric sans with monospace accent (developer audience)
Density: Compact (data-dense, many tables and charts)
Personality: Technical ("we are engineers building for engineers")
Stripe: 12/15 (strong authority, slightly too spacious for dashboards)
Vercel: 11/15 (good type, but interaction patterns lean marketing-site)
Selection: Linear design system with one adaptation -- swap primary purple for
the team's brand blue. All spacing, typography, and interaction tokens used as-is.
Applied: Tokens copied, CSS custom properties generated, project DESIGN.md created
documenting the color swap and referencing Linear DESIGN.md as source.
VIII. Related Skills
voice-before-structure -- Read design philosophy before writing structural artifacts
web-artifacts-builder -- Builds HTML components using applied design system
canvas-design -- Visual art creation using design philosophy principles
brand-guidelines -- Applies organization-specific brand to artifacts
Output
Scoring matrix table comparing all available systems across five dimensions
Top 3 match cards with: score, fit rationale, divergence notes, adaptation notes
Project DESIGN.md file created in the project directory referencing the selected source system and documenting any token adaptations
CSS custom properties block generated from selected system's tokens
Examples
Scenario 1: "Which design system fits my internal analytics dashboard for a data engineering team?" -> Scoring matrix run against 8 systems; Linear scores 14/15 on technical/compact/precise axes; selection delivered with one adaptation noted (swap primary purple for brand blue); project DESIGN.md created
Scenario 2: "I need to pick a design reference for a community-focused open source project" -> Scoring identifies Supabase (approachable, open-source-friendly) as best fit over Stripe (too authoritative); adaptation notes document switching monospace accent to humanist sans for community warmth
Edge Cases
Client already has a brand guide with defined tokens: Do not run the scoring workflow; map their existing tokens to the closest DESIGN.md system for reference, but use the brand guide as the authoritative token source
No available system scores above 9/15: Report the top match with a list of required adaptations; flag that a custom system may be warranted if adaptations exceed 3 dimensions
Dark mode is a hard requirement: Apply dark mode support as a pre-filter before scoring; eliminate any system without documented dark mode tokens
Anti-Patterns
Selecting a system based on the builder's aesthetic preference rather than running the five-dimension scoring matrix
Applying a system without documenting token adaptations — undocumented changes are invisible to the next developer and will be overwritten on re-import
Using a single-dimension match ("they both use Inter") as sufficient justification — typography is one of five dimensions