Materializr
Open-source parametric 3D CAD for makers — constraint sketches, solid modeling, threads, SVG & text engraving, STL/STEP export. Android, iPad, Linux, Mac, Windows. Vibe coded.
翻译准备中
个人 GitHub 账号下公开仓库的所有者可以登录并验证所有权,自行管理介绍,无需管理员批准。
登录并生成验证文件,在 30 分钟内将其提交到此仓库的默认分支。其他分支或分叉仓库中的文件无效。然后在下方检查文件,即可无需管理员批准直接管理项目。请保留文件以维持管理权限。
如需登录,将前往登录页面。登录后请回到此处,再次点击所选的验证按钮。
在项目文件夹中运行 /nomorevibe verify,即可验证所有权并自行管理介绍。
一分钟完成验证Connect your coding agent to over 50,000 real iOS screens, 1,900 user journeys and 1,200 website design systems (MCP), plus the no-ui-slop skill.
翻译准备中
取自仓库 README
正在加载完整 README。 以下为已保存的摘录。
UXKIN: real UI references for AI coding agents
Search real iOS app screens, complete user journeys and website design systems
from Claude, Claude Code, Cursor, Codex, GitHub Copilot and any MCP client,
so the UI your agent builds looks like a real product instead of AI defaults.
63,589 iOS screens from 303 apps ·
1,991 user journeys ·
1,288 website design systems , each with a DESIGN.md查看完整 README ↗https://github.com/user-attachments/assets/bc320cda-4a43-4baa-bdb9-cfdaccef6e49
Ask a coding agent for a settings screen, a paywall or a landing page and it fills every gap with the most common choices: the same gradients, card grids and vague copy as everyone else. UXKIN gives it something better to work from: how real, shipped products handle the same screen or flow, and the actual colors, type and components of real websites.
| What | For | Cost |
|---|---|---|
no-ui-slop skill | Rules that keep your agent from shipping generic, AI-looking UI | Free, no account |
build-ui skill | A step-by-step workflow for new screens, dashboards, polish, simplifying and reviews that ends with a checked result | Free, no account |
ios-screens skill | iPhone screens that feel native: navigation, sheets, Dynamic Type, dark mode, permissions | Free, no account |
design-md skill | Writes a DESIGN.md for your project from what it really uses, so every new screen matches | Free, no account |
ui-from-screenshot skill | Turns a screenshot or mockup into matching UI with your own components, then compares side by side | Free, no account |
ui-states skill | Builds the states agents forget: empty, loading, error, offline, long content | Free, no account |
uxkin-research skill | How your agent finds and compares real apps, flows and design systems before it designs | UXKIN plan to search (the skill itself is MIT) |
| UXKIN MCP server | Your agent searches real screens, journeys and design systems while it builds | UXKIN plan |
| UI Check GitHub Action | Flags dead buttons, missing states, one-off colors and accessibility basics in every pull request | Free |
| AI Slop Detector | See how AI-generated your website looks, and what to change | Free |
| UXKIN for Figma | Search UXKIN, place screens and whole flows on the canvas, import a site's colors and type | Coming soon to Figma Community |
npx skills add uxkin/agent
That installs all seven, or copy them into your agent's skills folder:
skills/no-ui-slop: how to build UI that
fits the product. Works on its own, with no account; with the UXKIN
server connected, it also tells your agent when to look up a real
reference.skills/uxkin-research: how to research
before designing: search the UXKIN library in plain words, open whole
journeys, compare several apps and report what they have in common.build-ui: a plan for each kind of UI
job (new screen, dashboard, polish, simplify, review), finished by
rendering at phone and desktop sizes, fixing and checking again.ios-screens: native iPhone UI in
SwiftUI, UIKit, React Native, Expo or Flutter, checked on small and large
iPhones, light and dark.design-md: reads your project and
writes (or updates) its DESIGN.md: colors with roles, type scale,
spacing, components and rules for agents.ui-from-screenshot: builds
a screenshot or mockup with your own components, then compares the result
with the image and fixes the differences.ui-states: finds and builds every
state of a screen (first use, empty, loading, error, offline, success)
and checks each one.All of them work on their own; with the UXKIN server connected they can
also look up real screens, flows and design systems. The plugins below
include no-ui-slop and uxkin-research; the other five install with
the command above (they live in .agents/skills).
The server is hosted at https://uxkin.com/mcp (Streamable HTTP). Most
apps sign in with your UXKIN account: no token to copy.
| App | How |
|---|---|
| Claude (web, desktop, mobile) | Open UXKIN in Claude's directory, click Connect, log in and click Allow |
| Claude Code | claude mcp add --scope user --transport http uxkin https://uxkin.com/mcp, then /mcp → uxkin → Authenticate |
| Cursor | One-click install on uxkin.com/cursor-ui-design, then Login next to uxkin |
| Codex | codex mcp add uxkin --url https://uxkin.com/mcp, then codex mcp login uxkin (or install the Codex plugin) |
| GitHub Copilot (VS Code) | Command Palette → MCP: Add Server → HTTP → https://uxkin.com/mcp, then sign in when asked (guide) |
| Lovable, Bolt, Devin Desktop | Add a custom MCP server with the URL above (Lovable, Bolt, Devin Desktop / Windsurf) |
| Any other MCP client | Add https://uxkin.com/mcp and sign in, or follow the setup at |
Searching needs a UXKIN plan (Monthly or Lifetime). You can disconnect any app from your account.
Once connected, ask your agent things like:
| Tool | Returns |
|---|---|
find_ui_references | Real iOS app screens, user journeys (multi-screen flows like onboarding or cancelling a subscription) and websites |
find_ui_materials | Design systems from real websites: colors with their roles, typography, components, and the full DESIGN.md for the top match |
get_journey | Every screen of one journey, in order |
list_collections / get_collection | The references you saved into collections on uxkin.com |
All tools are read-only.
It also offers ready-made prompts your app can show as commands (in
Claude Code: /mcp__uxkin__research_flow and so on):
| Prompt | What it asks your agent to do |
|---|---|
research_flow | See how real apps handle a flow (onboarding, checkout, paywall…) and recommend one |
find_design_system | Find real website design systems that fit your product and mood |
review_my_ui | Compare a screen in your project with real apps and suggest fixes |
build_from_collection | Start from references you saved on uxkin.com |
no-ui-slop and uxkin-research skills (skills). The five other skills are installed separately with npx skills add uxkin/agent..claude-plugin),
Cursor (.cursor-plugin) and Codex
(.codex-plugin), the Kiro power (POWER.md),
the Google Antigravity plugin
(plugin.json) and the Gemini CLI extension
(gemini-extension.json): the skill and the MCP
server in one install.server.json), published to the
official MCP Registry.DESIGN.md): UXKIN exports a
DESIGN.md for every website in its library, and UXKIN itself is built from
one. Hand it to your agent to see the format in action.DISTRIBUTION.md): every
registry, app store and directory, and where each one stands.One install adds the skill and the UXKIN server. The skill works right
away; to search the library, connect the server (Claude opens a UXKIN page
where you log in and click Allow). In Claude Code, type /mcp, pick
uxkin and choose Authenticate.
In Claude Code you can also add it from GitHub directly:
claude plugin marketplace add uxkin/agent
claude plugin install uxkin@uxkin
After installing, click Login next to uxkin in Cursor's MCP settings, log in to UXKIN and click Allow.
codex plugin marketplace add uxkin/agent
codex plugin add uxkin@uxkin
codex mcp login uxkin
The last command opens a UXKIN page where you log in and click Allow.
git clone https://github.com/uxkin/agent uxkin
agy plugin install ./uxkin
Then type /mcp, pick uxkin and choose Authenticate: a UXKIN page
opens where you log in and click Allow. To add only the server, put
mcp_config.json in ~/.gemini/config/mcp_config.json.
In Kiro, open the Powers panel → Add Custom Power → Import power
from GitHub, and enter https://github.com/uxkin/agent. When Kiro asks
to authenticate uxkin, click Authenticate, log in to UXKIN and click
Allow.
gemini extensions install https://github.com/uxkin/agent
Then start Gemini CLI and type /mcp auth uxkin: a UXKIN page opens where
you log in and click Allow.
The skill runs locally and sends nothing. The MCP server receives only the search text your agent sends (for example "onboarding for a meditation app"), the result limit and your sign-in, and returns matching references from uxkin.com. It doesn't read your code, files or conversation. See the privacy policy.
Fixes to the skill, setup steps and examples are welcome: see CONTRIBUTING.md. To report a security problem, see SECURITY.md.
The skill and the files in this repository are MIT licensed. The UXKIN library itself (its screens, journeys and design systems) is a hosted service, not part of this repository: see the terms. The MIT license covers the code and text here; it doesn't grant use of the UXKIN name or logo.
Listed in the Claude Market MCP directory
examplesAGENTS.mdCLAUDE.md