Introduction
If you work as an experience designer in 2026, you have probably already tested Claude for something: user research, a flow schema, or copy for a prototype. But using Claude like a generic chat is like using Figma only to write notes. The tool can do much more.
Claude Skills are reusable instruction sets that turn Claude from a generic assistant into a specialized collaborator for specific phases of your design process. They work as .md files with predefined workflows: install once, reuse every time, without rewriting your brief from zero in each session.
This guide collects the most useful skills for real experience design work: some public and widespread, others custom-built for your exact workflow.
What changed in April 2026
On April 17, 2026, Anthropic launched Claude Design, the first Anthropic Labs product. It is a visual workspace powered by Claude Opus 4.7 that generates interactive prototypes, slide decks, wireframes, and one-pagers from natural language, uploaded files, or connected repositories.
Claude Design includes native export to Canva, PDF, PPTX, and HTML, and it can read your codebase to apply your design system. That technical ability clearly differentiates it from Figma AI and Canva Magic Studio.
At the same time, Figma integrated Claude Code as an official MCP partner. This makes it possible to move workflows started in Claude Code directly into the Figma canvas for collaborative sharing and iteration.
Claude Code skills sit inside this ecosystem: they do not replace your design judgment, but they automate the operational and repetitive parts of your process.
Skills by project phase
1. Research and discovery
customer-research
This skill is crucial before opening Figma. It analyzes interview transcripts, G2 reviews, Reddit threads, support tickets, and surveys to extract patterns and build personas. You can feed raw, messy material and get structured syntheses, quotes grouped by theme, and ready-to-use JTBD profiles.
Use when: you need to analyze existing feedback, mine competitor reviews, or synthesize user testing output quickly.
ab-testing
Not only for launch. In discovery, it helps you define research hypotheses, success metrics for usability tests, and evaluation criteria before tests start. It also supports statistical significance checks.
2. Strategy and architecture
site-architecture
Maps page hierarchies, navigation structures, and URLs. Useful when presenting information architecture to clients or developers, or when documenting the current structure before redesigning it.
cro (Conversion Rate Optimization)
For designers working on conversion-driven products such as landing pages, checkout flows, or onboarding. It analyzes any page and returns prioritized UX and persuasion-design recommendations.
3. Prototyping and UI
frontend-design (official Anthropic)
With more than 277,000 installs in March 2026, it is the most adopted skill in the ecosystem. It gives Claude a design system and visual philosophy before code generation, producing outputs with intentional typography, non-generic palettes, and recognizable components.
Use when: generating HTML prototypes, React components, or interfaces that should not feel AI-generated.
ui-ux-pro-max
The most complete design skill in the ecosystem. It provides a searchable database: 50+ UI styles, 97 color palettes, 57 typography pairings, 99 UX guidelines, and 25 chart types across 9 tech stacks. Its four-step workflow (requirements analysis, design system generation, output application, QA) is useful for teams managing multiple visual identities.
ckmui-styling
Focused on shadcn/ui components with Radix UI and Tailwind. Ideal when your development team already uses this stack and you want frictionless handoff.
4. Figma integration
This is where the biggest change happened in the past year. Through Figma's official MCP server, Claude Code can read your Figma file, create components, set variables, and organize layers directly from terminal workflows.
Figma-specific Claude Code skills act as a governance layer: they define how AI must behave in that context. The challenge is no longer tool capability, but the lack of explicit rules.
- Token Map + Component Registry to build the design-token map and component registry baseline for the session
- Write to Figma to write variables and components directly in canvas, then run token-by-token QA binding checks
- Design system compliance to enforce token, component, layer-structure, and naming-convention rules
The same tool should behave differently by context: when the designer must stay in the loop, compliance wins; when speed matters for prototyping, velocity can take priority.
5. Content and copy
copywriting
For designers writing interface text, this separates generic UX writing from conversion-oriented copy. It handles headlines, CTAs, microcopy, and value propositions while adapting tone to your product.
copy-editing
Use when copy already exists and needs refinement. Helpful before user testing to reduce wording bias in test results.
stop-slop
A meta but essential skill. It removes typical AI writing patterns from interface copy, specs, and documentation. Operational rules include active voice, specific phrasing, and no empty emphasis.
6. Presentations and deliverables
ckmslides
Generates strategic HTML presentations with Chart.js, integrated design tokens, and copywriting formulas. Faster and more controllable than AI-assisted slide software for stakeholder decks.
sales-enablement
When your design work must persuade clients, boards, or product teams, this skill structures one-pagers, pitch decks, and presentation assets with persuasive logic.
7. Validation and testing
analytics
Sets up measurement for design impact, covering GA4, Mixpanel, GTM, event tracking, and UTM strategy. It helps close the loop between design choices and real data.
ab-testing
In testing, it helps structure A/B experiments for design variants, estimate required sample size, and interpret outcomes with correct statistical significance.
How to build custom skills
The core problem Claude Skills solve is simple: every new session usually forces you to rebuild context from scratch. A well-structured skill applies that context every time.
The skill-creator skill guides the full process: schema definition, instruction writing, and test framework setup to verify behavior.
For experience designers, the most useful custom skills are usually:
- Automated design review: checklist before user tests (accessibility, copy consistency, missing states)
- UX writing standard: approved tone, product vocabulary, and microcopy patterns
- Handoff documentation: structured development annotations with fields your engineering team explicitly requests
The workflow that works
The turning point is not using AI to do more tasks. It is encoding your process inside Claude. When workflow phases are formalized as skills, Claude no longer runs faster in chaos: it brings order to chaos.
Recommended sequence for a project:
- customer-research to analyze discovery material
- site-architecture to map information structure
- frontend-design + ui-ux-pro-max for high-quality prototypes
- Figma MCP + compliance skills to move work into the design file
- copy-editing + stop-slop to validate content quality
- ab-testing + analytics to close the loop with data
Conclusion
Skills are not a way to do more in less time. They are a way to do the right things in the right way, without restarting from zero every time.
Updated May 2026. Claude Design is currently available in research preview for Claude Pro, Max, Team, and Enterprise subscribers. You can subscribe from the official page: claude.ai/upgrade.