Index of every page on this site: llms.txt
All skills
hktitan avatar

/khe-ui

@945e23c
by Harshit Khemanihktitan/skills47 stars
2
Run it yourself

Interface craft with numbers: motion, typography, color and surface, layout, components, accessibility, mobile feel, AI tells, review, and the pass before shipping. Builds or judges a screen against exact values and ends a review with a verdict. Use when someone says "review this component", "is this well built", "it works but feels unfinished", "make it feel right", "this looks AI-generated", "which easing", "the page jumps", "is this accessible", "feels like a website on my phone", "are we ready to ship", or is building a screen, a form, a card or a page, or styling one built on Vercel's UI SDK. For an SVG use `khe-svg`; for sound use `khe-sound`; for a launch video use `khe-launch-film`; for a `.design` file or DESIGN.md use `khe-design-tokens`.

Use this Skill: https://skilld.dev/gh/hktitan/skills/khe-ui

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

README.md

≈1.3k tokens on demand. Your agent reads this file only when SKILL.md points to it.

khe-ui

Interface craft with numbers, and a review that ends in a verdict.

npx skills add HKTITAN/skills --skill khe-ui

What it does

  • Builds or judges a screen against exact values: a duration, a curve, a contrast ratio, a pixel size.
  • Covers eleven areas, one file each, opened only when the job needs it: motion, components, accessibility, mobile, typography, color and surface, layout, taste, AI tells, review, and the pass before shipping.
  • Ends a review with Blocked, Approve with changes or Ship it, and says what it could not verify.
  • Deletes before it adds. Most "feels off" is too much of something.
  • Ships its own gate, gold examples and teaching pairs, so "done" is a pasted line and a list of what was seen.

How it runs

khe-ui's flow: khe-design-tokens hands in the contract and ui-sdk the rendered parts; khe-ui opens one lane, runs its gate, looks at the screen and ends on a verdict with what it did not check; a fault found by looking goes back to the build; motion inside an SVG goes to khe-svg and a sound to khe-sound

Used

/khe-ui review InviteDialog.css InviteDialog.tsx

The gold review in examples/review.md: the gate's lines on the two files, then its verdict.

scale-from-zero: `transform: scale(0)` in `.invite` (InviteDialog.css:6)
transition-all: `transition: all 200ms cubic-bezier(0.23, 1, 0.32, 1)` in `.invite` (InviteDialog.css:7)
outline-none: `.invite__email { outline: none }` has no `:focus-visible` outline or ring to replace it (InviteDialog.css:17)
unnamed-icon-button: <button> holds only an icon and has no `aria-label` (InviteDialog.tsx:5)

files scanned: 2
never-intentional: 4 (scale-from-zero 1, transition-all 1, outline-none 1, unnamed-icon-button 1)
candidates: 0
passing is necessary, not sufficient
ui-check failed

Verdict: Blocked (5 findings).
Not checked: the dark theme, 390px on a real phone, VoiceOver, the entrance at 25% speed, where focus lands when the dialog closes, and contrast (the colours are tokens defined outside these two files).

The fifth finding, font-size: 14px on the input, is one the gate did not see; the checklist sheet caught it. Each finding is a Before, After and Why row in the file.

When to use it

Use it when a screen works and something is still wrong, when you want a component or a pull request judged before it merges, or when you are building a screen and want it right the first time. It is one skill on purpose: you say what is wrong in your own words and it opens the one area that owns it.

Invoke with

/khe-ui [review | polish | build | tells | a11y | mobile | ship] [target]

Or describe the problem in your own words.

Before and after

Before: "Your modal animation could be smoother. The transition is a bit long and the scale starts too small."

After: transition: all 0.6s ease-in becomes transition: transform 250ms cubic-bezier(0.23, 1, 0.32, 1), opacity 250ms, scale(0) becomes scale(0.95), each with a one-line reason, under a verdict of "Approve with changes".

Example prompts

  • Review this settings page before I merge it. Did I make anything worse?
  • The dashboard works but it feels unfinished and I can't say why.
  • This card breaks with long names. What else breaks?

Verify

node scripts/ui-check.mjs src/components --gate

The gate reads CSS, HTML, JSX and TSX as text and names every fault it can prove as one line, such as ease-in: or unnamed-icon-button:, with the file and line. It ends ui-check passed only when every never-intentional line is at 0.

  • What it proves: the source is free of the faults a script can name: transition: all, ease-in, scale(0), animated layout properties, a hover that moves the element outside the hover gate, motion with no reduced-motion branch, an input under 16px, an icon button with no name, a removed focus ring, a z-index above 400, locked zoom, and a review with no verdict.
  • What it does not prove: that the screen feels right, that every state exists, that a keyboard reaches it, that the contrast passes, or that it works on a phone. Those need a look. Passing is necessary, not sufficient.

node scripts/ui-check.mjs --self-test proves the gate against the examples and teaching pairs in this folder. --help lists every line.

What is inside

  • SKILL.md: the one idea, the workflow, the modes, the hard rules, the numbers that come up on every screen.
  • references/: eleven lane files; ui-sdk, where each value goes on a screen built with Vercel's UI SDK; verify, the last pass and the gate's lines; fresh-eye, the brief for a second reviewer; teaching, the index of the pairs; plus gotchas.md and pov.md, which are yours to edit.
  • examples/: five gold examples that set the bar: a button with every state, a popover entrance and exit, a form field with validation timing, a list with loading, empty and error branches, and a finished review.
  • teaching/: fourteen bad and good pairs, one fault each. They are also the gate's tests.
  • scripts/ui-check.mjs: the gate. Node 20 or later, no dependencies.
  • LICENSE: MIT.

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

1 warning4d3 checks · Risk SAFE
  • Gen Agent Trust Hub4d

    The khe-ui skill is a robust design engineering toolkit used by AI agents to build and audit high-quality user interfaces. It ships with a local JavaScript utility (ui-check.mjs) for static code analysis and possesses no external dependencies or network capabilities. While safe for development workflows, users should note that the skill's primary purpose is to process untrusted source code, which constitutes a surface for indirect prompt injection.

  • Socket4d

    No alerts

  • Snyk4d

    Risk: MEDIUM · 1 issue

Signed by skilld at 945e23c. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated last week
metadata
{
  "author": "Harshit Khemani",
  "version": "1.0.0"
}

README badge

README badge for hktitan/skills/khe-ui