GUIDES
21st.dev Fixes AI Slop π
AI-built websites all start to look the same. 21st.dev is a free component library β built by real designers β that plugs straight into Claude. Setup takes two.
What It Actually Is
21st.dev is a giant open-source registry of UI components β React + Tailwind β built by professional designers and engineers. Heroes, navigation bars, dashboards, pricing pages, login pages, footers. The component code is MIT-licensed and free for personal and commercial use. No restrictions.
On top of that thereβs Magic MCP β an AI agent that lets Claude generate components from natural language. Tagline: βv0 in your IDE.β One slash command and Claude pulls the right component into your project.
My Favorite Categories
Hero Sections
Browse, find one that matches your brand vibe, copy. Best return on time of any category.
Nav Bars
A great nav fixes 30% of the AI-slop look. Sticky transparent navs are the move.
Dashboards
Sidebar layouts with stat cards. Use these for SaaS apps and internal tools.
Pricing Sections
3-tier comparisons that actually convert. Saves you a week of design iteration.
Testimonial Sections
Especially the ones with photos + tags. Your social proof page in 5 minutes.
Footers
Most AI-built footers are forgettable. A real one signals you actually shipped a product.
Connect To Claude in 2 Minutes
Sign up at 21st.dev β open the Magic Console β copy your API key. The free tier gives you 5 component generations per month. Pro is $20/month for 50.
Replace YOUR_KEY_HERE with the key you just copied. The CLI handles the Claude Code config automatically. Restart Claude Code when itβs done.
In any Claude Code project, type /ui followed by what you want. Magic MCP generates the component, writes the files in the right places, and matches your existing code style.
3 /ui Prompts to Start With
The real unlock of AI agents is not full autonomy, but human-in-the-loop systems design. Structure your agent inputs with explicit constraints, negative prompts, and automated test checkpoints. When building tools, keep token consumption lean by caching system prompts and isolating tasks into specialized sub-agents.