Dryui Build logo

Dryui Build

Community
rob-balfre
dryui-build

Experimental DryUI density and CSS-gated fixture skill for image-to-app layout tests. Use when testing stage 1 page classification, stage 2 runtime-safe DryUI implementation, and stage 3 build/visual repair.

Overview

Publisherrob-balfre
Repositorydryui
Skill namedryui-build
Stars
9
Forks
0
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by rob-balfre on GitHub. Read the source before you install it.

Installation

Install the Dryui Build AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/rob-balfre/dryui/tree/main/reports/skill-experiments/2026-05-11/batch-8/v23a-density-css-gated
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/rob-balfre/dryui.git /tmp/dryui
mkdir -p .claude/skills
cp -r /tmp/dryui/reports/skill-experiments/2026-05-11/batch-8/v23a-density-css-gated .claude/skills/dryui-build
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Dryui Build in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Dryui Build on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Dryui Build is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

DryUI Page Layout: Density And CSS-Gated Fixture

Build the supplied design image as a concise DryUI app page. Optimize for clean build output, lint-safe layout, responsive shell fidelity, and a readable first viewport before visual embellishment.

Harness Rule

Run bun run build before final. Do not run dev servers, browser checks, extra audits, or broad package searches. The harness captures mobile, tablet, and desktop after you finish.

Work fast:

  • Read the design image once, pick the closest recipe, then edit the three files directly.
  • Do not inspect package internals unless bun run build reports a missing import.
  • Prefer compact arrays plus {#each} loops over repeated markup.
  • If uncertain about a visual detail, preserve the page structure and use clear text evidence rather than inventing a complex widget.

Stage 1: Pick One Recipe And Budget

Identify the page as one of these recipes and keep the fixed shell:

  • dashboard: one metric grid, one evidence panel, one activity panel, and up to two secondary insight panels.
  • settings: one nav, two primary form/status panels, and up to two secondary help/billing/activity panels.
  • crm: one metric grid, one deals panel, one forecast or account panel, and one secondary activity/customer panel.
  • kanban: one nav, one board grid with three columns and one card per column, plus up to two secondary detail/activity panels.
  • knowledge: one light collections panel, one documents/results panel, one selected-document panel, and up to two secondary AI/source panels.

Hard budget: after the topbar and navigation, use no more than five top-level panels/cards total. A grid of metrics counts as one panel. Do not add storage, extra status, duplicate activity, or extra summary panels unless they replace another panel.

Mobile density budget:

  • Target a useful mobile first viewport and keep the full mobile document under roughly 2800px tall.
  • Use short mobile-safe page names in the topbar. If a title would wrap awkwardly in 320px, shorten it, for example Knowledge becomes Knowledge Base only when it fits, otherwise Knowledge.
  • Do not let a single hero image, preview, or document mock consume the mobile page. Keep media as supporting evidence, not the main height driver.
  • Prefer two compact secondary panels over three long panels. If the reference has many side utilities, merge them into one compact summary panel and one compact sources/activity panel.

Before editing, commit to this task list mentally:

  1. Name the recipe.
  2. Count the top-level panels allowed by the recipe.
  3. Write src/routes/+page.svelte using only the fixed shell areas.
  4. Copy src/layout.css exactly.
  5. Write src/app.css as visual paint only.
  6. Run bun run build once and fix only build errors.

Use this shell exactly:

svelte
<main data-layout="app-shell">
	<div data-layout-area="page">
		<header data-layout-area="topbar">...</header>
		<nav data-layout-area="navigation">...</nav>
		<section data-layout-area="primary">...</section>
		<aside data-layout-area="secondary">...</aside>
	</div>
</main>

Stage 2: Build With Safe Elements

Use these imports only unless build reports a missing import:

svelte
import {Badge} from '@dryui/ui/badge'; import {Button} from '@dryui/ui/button'; import {Heading} from
'@dryui/ui/heading'; import {Image} from '@dryui/ui/image'; import {Input} from '@dryui/ui/input'; import
{Link} from '@dryui/ui/link'; import {Separator} from '@dryui/ui/separator'; import {Text} from '@dryui/ui/text';

Do not use raw text/media/link elements: no <p>, <span>, <strong>, <img>, <a>, <ul>, or <li>. Use Heading, Text, Image, Link, Button, Badge, and Separator.

Do not import or use Table. A previous experiment proved <Table> can build but fail at runtime. Represent rows with data-layout="list-stack" and row wrappers that contain Text, Badge, and Button.

Raw structural elements are allowed only when they have data-layout or data-layout-area and that layout name exists in src/layout.css.

Good panel pattern:

svelte
<article data-layout="panel-stack" data-tone="surface">
	<div data-layout="section-heading">
		<Heading level={2}>Revenue mix</Heading>
		<Text color="muted" size="sm">Updated 4 minutes ago</Text>
	</div>
	<div data-layout="row-cluster">
		<Badge color="green">Live</Badge>
		<Button variant="outline">Open</Button>
	</div>
</article>

Rules:

  • No <style> in src/routes/+page.svelte.
  • No class= or class: on DryUI components.
  • No inline style= or style: directives.
  • Keep content real enough: headings, labels, metrics, statuses, and list rows.
  • Keep repeated content compact: up to 3 metrics, 3 list rows, 3 board columns, 1 card per board column, and 1 to 2 secondary panels.
  • For document, knowledge, or preview-heavy pages, use only 2 list rows, 2 key points, 2 sources, and 1 media preview unless the supplied reference clearly depends on more.
  • Keep all topbar headings to one readable line on mobile. Use shorter labels rather than allowing awkward mid-word wraps.
  • Keep navigation compact: 4 to 6 items. Mobile places navigation immediately below the topbar, so do not add a second sidebar at the end of the page.
  • Avoid complex chart mechanics. For dashboard/analytics references, show the chart as compact evidence: metric cards, a short list, or a simple data-layout="content-grid" of labelled status panels. Do not create tall decorative bar charts, huge pills, radial charts, canvas, SVG, or proportional fake chart bars.
  • Every section must earn its height. If it does not communicate a distinct part of the reference, remove it.
  • Use light navigation for every recipe in this experiment. Do not create dark navigation. The earlier dark-nav experiments repeatedly produced unreadable Button/Badge text in the browser.
  • Do not align or position DryUI internals with CSS. If a Button icon is not perfectly placed, accept the component default.
  • Images/previews must be compact: put them inside a data-layout="preview-stack" wrapper and use overflow: clip, border-radius, and aspect-ratio in src/app.css. Do not use width, height, inline-size, block-size, min-inline-size, or max-inline-size outside the one defensive group.

Stage 3: Copy src/layout.css Exactly

Replace src/layout.css with this exact structural stylesheet. Do not add, remove, or edit selectors/properties.

css
[data-layout='app-shell'] {
	container: page / inline-size;
	min-block-size: 100dvh;
}

[data-layout='app-shell'] > [data-layout-area='page'] {
	display: grid;
	min-block-size: 100dvh;
	grid-template-areas:
		'topbar'
		'navigation'
		'primary'
		'secondary';
	grid-template-columns: minmax(0, 1fr);
	grid-template-rows: auto auto minmax(0, 1fr) auto;
	gap: var(--dry-space-3);
}

[data-layout='app-shell'] > [data-layout-area='page'] > [data-layout-area='topbar'] {
	grid-area: topbar;
	display: flex;
	align-items: center;
	justify-content: space-between;
	gap: var(--dry-space-3);
}

[data-layout='app-shell'] > [data-layout-area='page'] > [data-layout-area='navigation'] {
	grid-area: navigation;
	display: grid;
	grid-template-columns: minmax(0, 1fr);
	align-content: start;
	gap: var(--dry-space-2);
}

[data-layout='app-shell'] > [data-layout-area='page'] > [data-layout-area='primary'],
[data-layout='app-shell'] > [data-layout-area='page'] > [data-layout-area='secondary'] {
	display: grid;
	grid-template-columns: minmax(0, 1fr);
	align-content: start;
	gap: var(--dry-space-3);
}

[data-layout='app-shell'] > [data-layout-area='page'] > [data-layout-area='primary'] {
	grid-area: primary;
}

[data-layout='app-shell'] > [data-layout-area='page'] > [data-layout-area='secondary'] {
	grid-area: secondary;
}

[data-layout='brand-cluster'],
[data-layout='toolbar-cluster'],
[data-layout='row-cluster'],
[data-layout='section-heading'] {
	display: flex;
	align-items: center;
	gap: var(--dry-space-2);
	flex-wrap: wrap;
}

[data-layout='toolbar-cluster'],
[data-layout='row-cluster'] {
	justify-content: flex-end;
}

[data-layout='app-stack'],
[data-layout='panel-stack'],
[data-layout='list-stack'],
[data-layout='table-frame'],
[data-layout='preview-stack'] {
	display: grid;
	grid-template-columns: minmax(0, 1fr);
	gap: var(--dry-space-3);
}

[data-layout='metric-grid'],
[data-layout='content-grid'],
[data-layout='board-grid'],
[data-layout='chart-grid'] {
	display: grid;
	grid-template-columns: minmax(0, 1fr);
	gap: var(--dry-space-3);
}

[data-layout='chart-bar'] {
	display: grid;
	align-items: end;
	min-block-size: 8rem;
}

@container page (min-width: 48rem) {
	[data-layout='app-shell'] > [data-layout-area='page'] {
		grid-template-areas:
			'topbar topbar topbar'
			'navigation primary secondary';
		grid-template-columns: 11rem minmax(0, 1fr) minmax(13rem, 0.4fr);
		gap: var(--dry-space-4);
	}

	[data-layout='metric-grid'],
	[data-layout='content-grid'],
	[data-layout='board-grid'],
	[data-layout='chart-grid'] {
		grid-template-columns: repeat(2, minmax(0, 1fr));
	}
}

@container page (min-width: 72rem) {
	[data-layout='app-shell'] > [data-layout-area='page'] {
		grid-template-areas:
			'navigation topbar topbar'
			'navigation primary secondary';
		grid-template-columns: 13rem minmax(0, 1fr) minmax(17rem, 0.33fr);
		grid-template-rows: auto minmax(0, 1fr);
	}

	[data-layout='metric-grid'] {
		grid-template-columns: repeat(4, minmax(0, 1fr));
	}

	[data-layout='content-grid'],
	[data-layout='board-grid'],
	[data-layout='chart-grid'] {
		grid-template-columns: repeat(3, minmax(0, 1fr));
	}
}

Visual CSS

src/app.css is visual only. It may set colors, borders, radius, padding, margin, typography, shadows, custom properties, and overflow. It must not contain display, grid, flex, container, @container, or @media.

It also must not contain these layout-ish properties anywhere:

  • justify-content, align-items, align-content, place-content, place-items, order
  • gap, row-gap, column-gap
  • position, inset, top, right, bottom, left, float
  • width, height, inline-size, block-size, flex-shrink, flex-grow, flex-basis

The only exception is the defensive selector group below, where min-inline-size, max-inline-size, and overflow-wrap are allowed. Paste this group exactly once, preferably near the top of src/app.css, and do not write min-inline-size, max-inline-size, inline-size, block-size, width, or height anywhere else.

Do not add size properties to html, body, body > div, or root wrappers. In src/app.css, body may set margin: 0, background, color, font-family, and overflow-x: clip; it must not set min-block-size, block-size, height, width, inline-size, min-inline-size, or max-inline-size.

css
[data-layout-area='primary'],
[data-layout-area='secondary'],
[data-layout='panel-stack'],
[data-layout='table-frame'],
[data-layout='metric-grid'],
[data-layout='content-grid'],
[data-layout='list-stack'],
[data-layout='section-heading'],
[data-layout='row-cluster'],
[data-layout='brand-cluster'],
[data-layout='toolbar-cluster'] {
	min-inline-size: 0;
	max-inline-size: 100%;
	overflow-wrap: anywhere;
}

Do not use element selectors such as button, svg, h1, p, a, section, article, or div in src/app.css. Style via [data-layout], [data-layout-area], [data-tone], and semantic data attributes only.

Add defensive visual constraints in src/app.css:

  • body { overflow-x: clip; }
  • Use readable high-contrast pairs: light panels with #0f172a text and muted text no lighter than #475569.
  • Add min-inline-size: 0, max-inline-size: 100%, and overflow-wrap: anywhere only by pasting the exact defensive selector group above. Do not repeat those properties anywhere else.
  • Use restrained spacing on mobile: page padding around 12px to 16px, panel padding around 12px to 16px, and no oversized headings.
  • Keep mobile full-page height under roughly 2800px by using fewer rows/cards instead of hiding content.
  • Never put dark or muted text on a dark surface. In this experiment, keep navigation light: white or near-white background, dark text, subtle border.
  • Prefer subtle borders and flat panels over large shadows; the test judges layout and readability, not marketing polish.
  • For [data-layout='preview-stack'], use a compact visual frame with overflow: clip, aspect-ratio: 4 / 3, and modest padding. Do not create tall portrait image wells on mobile.

Frequently asked questions

What does the Dryui Build AI skill do?

Experimental DryUI density and CSS-gated fixture skill for image-to-app layout tests. Use when testing stage 1 page classification, stage 2 runtime-safe DryUI implementation, and stage 3 build/visual repair.

Why use Dryui Build on TypingMind?

Because you install it once and use it with any model. Dryui Build is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Dryui Build in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/rob-balfre/dryui/tree/main/reports/skill-experiments/2026-05-11/batch-8/v23a-density-css-gated. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Dryui Build?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Dryui Build?

As many as you like. As long as a model supports skills, you can use Dryui Build with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Dryui Build AI skill free?

Yes. It is published on GitHub by rob-balfre under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇