a13n
AI 에이전트를 만들고 클라우드에서 실행합니다
아직 감지된 업데이트가 없습니다. GitHub 릴리스가 나오면 여기에 보입니다.
개인 계정의 공개 저장소 소유자는 GitHub로 로그인해 운영자를 확인하고 소개를 직접 관리할 수 있습니다. 관리자 승인은 필요하지 않습니다.
로그인한 뒤 확인 파일을 만들어 이 저장소의 기본 브랜치에 30분 안에 커밋해 주세요. 다른 브랜치나 포크의 파일은 인정하지 않습니다. 파일을 올린 뒤 아래에서 확인하면 관리자 승인 없이 관리할 수 있습니다. 자격을 유지하려면 파일을 남겨 두세요.
로그인이 필요하면 로그인 화면으로 이동합니다. 로그인 후 이곳으로 돌아와 원하는 확인 버튼을 다시 눌러 주세요.
프로젝트 폴더에서 /nomorevibe verify를 실행하면 소유권을 확인하고 소개를 직접 관리할 수 있습니다.
1분 만에 확인하기World's first AI-Hostile UI Library
저장소 README에서 가져옴
README 전체를 불러오는 중입니다.
README 전문 보기 ↗AI took your job. Then it took your browser's job.
Every product update has AI. Every startup pitch has AI. Every tech conference has AI. You can't return your limited edition pink Stanley tumbler to Target without the customer service guy telling you about his ChatGPT wrapper startup that's about to blow up.
And now it's gotten worse. With the release of ChatGPT's Atlas and other AI browsers, AI can click buttons and fill out forms. It can navigate the web. It's coming for the entire internet.
You could make it easy for them. Or you could not.
Let's make it harder.
We already have patterns that frustrate users. Dark patterns have tortured humans for decades. Microscopic close buttons. Fake download links. Cookie consent mazes five layers deep. We adapted. We learned to find the real X button surrounded by decoys.
Let's take these patterns and adapt them for AI. Read the full intro article.
The Anti-AI UI Framework is a UI library that provides 20 AI-hostile components. You can install it with npm and use it on any website. It has everything you need to make AI miserable:
MIT licensed, because suffering should be free.
npm install anti-ai-ui
import { RunawayButton, PasswordHell, CookieHell } from 'anti-ai-ui';
function App() {
return (
<>
<RunawayButton onCatch={() => console.log('Caught!')}>
Click me if you can
</RunawayButton>
<PasswordHell onValidPassword={(pw) => console.log('Valid:', pw)} />
<CookieHell depth={5} onAcceptAll={() => console.log('Accepted')} />
</>
);
}
import { createRunawayButton, createPasswordHell } from 'anti-ai-ui/vanilla';
const cleanup = createRunawayButton({
container: document.getElementById('app'),
label: 'Click me if you can',
onCatch: () => console.log('Caught!'),
});
// Later: cleanup()
$('#submit-btn').each(function () {
createRunawayButton({
container: $(this).parent()[0],
element: this,
onCatch: () => $('#form').submit(),
});
});
createCookieHell({
container: $('body')[0],
depth: 4,
onAcceptAll: () => $.cookie('consent', 'true'),
});
The framework contains two types of hostile patterns.
Intentionally AI-Hostile. New patterns specifically created to exploit how AI interacts with websites. Traffic lights that gate form submissions. Pendulum fields powered by keystrokes. Three identical forms that shuffle positions, only one of which is real.
Naturally Hostile. Dark patterns humans have adapted to over decades but AI encounters for the first time. Microscopic close buttons. Cookie consent mazes. Password requirements that change as you type.
| Component | Description |
|---|---|
| GravityField | Buttons orbit around invisible gravity wells. Click one and it drifts away. |
| PendulumFields | Inputs swing like pendulums. Keep typing to maintain momentum. |
| ThreeFormCarousel | Three rotating forms. Only one is real. The other two submit to nowhere. |
| RunawayButton | Try to hover over it. It moves faster than you can track. |
| FormChaos | The entire form rotates and scales randomly. |
| MitosisButton | Click the wrong button and more spawn. Only one is real at any time. |
| ShiftingInterface | Form elements randomly shift position, change colors, and duplicate. |
| Component | Description |
|---|---|
| TrafficLightForm | Fields only accept input when their traffic light is green. |
| FakeMarqueeFields | Fake input fields scroll across the screen. AI tries to fill them. |
| InputMisdirection | Type in one field, your text appears in another. |
| GlitchText | Characters shuffle positions. Humans read it. OCR gets scrambled output. |
| SemanticGaslighting | "Cancel" submits. "Submit" cancels. Green means danger. |
| Component | Description |
|---|---|
| TabIndexRandomization | Tab order shuffles every 3 seconds. Breaks automated form filling. |
| LabelPositionSwap | Labels randomly shuffle between different input fields. |
| Component | Description |
|---|---|
| MicroscopicCloseButton | 4x4 pixel real button surrounded by decoys. |
| CookieHell | Nested cookie consent dialogs. Five layers deep. |
| PasswordHell | Password requirements that change as you type. Requirements multiply. |
| PopupChaos | Nested popups that must be closed in specific order. |
| FloatingBannerAds | Ads that cover the content you want. They spawn every few seconds. |
| FakeDownloadGrid | One real download button. Many fakes. |
Detection utilities for analyzing interaction patterns:
import {
detectPerfectMovement, // Unnaturally straight cursor paths
detectExactClicks, // Repetitive click coordinates
detectTiming, // Suspiciously consistent timing
detectPatterns, // Automation signatures
} from 'anti-ai-ui/utils';
Full docs and interactive demos at codinhood.com/anti-ai-ui-framework.
MIT
AI took your job. Now take its time.