Dryui Init logo

Dryui Init

Community
rob-balfre
dryui-init

Bootstrap a SvelteKit + DryUI project using skill-led inspection and explicit file/package edits. Works for new projects, existing SvelteKit apps, and brownfield repos.

Overview

Publisherrob-balfre
Repositorydryui
Skill namedryui-init
Stars
9
Forks
0
Bundled files
13
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.

  • 13 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

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

Installation

Install the Dryui Init 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/skills/dryui-init
  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/skills/dryui-init .claude/skills/dryui-init
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Dryui Init 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 Init 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 Init 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:init

Bootstrap DryUI in the current project, or scaffold a new SvelteKit app when the user asks for a new project.

Install Or Refresh This Skill

The durable install path for DryUI skills is:

bash
npx skills add rob-balfre/dryui

Skill Discipline

This is a short, linear flow. Do not create per-step TaskCreate items — the bookkeeping costs more than the work it tracks. One intent line up front and one summary line at the end is enough.

First tool call is always the script. Do not run an explicit pre-triage ls/[ -f X ] call — the script self-triages and routes via exit code:

bash
bash <skill-base-dir>/scripts/bare-skeleton.sh "$PWD"
  • Exit 0 → done. Print one summary line and stop.
  • Exit 2 with route=existing-sveltekit → follow Apply Setup below.
  • Exit 2 with route=ambiguous → ask the user one question (scaffold over / integrate / abort), then act on the answer.
  • Any other non-zero → read the message, fix the specific issue, do not re-run the whole thing manually.

Do not Read any file you just copied from templates/ to verify it — those files are validated by bun run check at the end of the script. Do not run bunx sv@latest create into a temp dir. Do not overwrite package.json with Write — the script uses jq to merge so user-set keys (especially overrides) are preserved.

What The Script Does

scripts/bare-skeleton.sh is the single entry point. It self-triages and either bootstraps a bare skeleton or exits 2 with a routing hint. Steps in order:

  1. Triage — exits 2 with route=existing-sveltekit if svelte.config.* is present, or route=ambiguous if src/ exists without a config.
  2. Removes index.ts / index.js from the bun init skeleton.
  3. Copies every file in templates/ into the project root, including local AGENTS.md / CLAUDE.md instructions that force UI agents through dryui-build.
  4. Merges scripts, lucide-svelte into dependencies, devDependencies, and type: "module" into package.json via jq. Preserves existing overrides and any other dependency keys.
  5. Appends SvelteKit/Vite ignores to .gitignore (idempotent — checks for /.svelte-kit first).
  6. Detects a local dryui workspace at $DRYUI_LOCAL, ../dryui, ~/dryui, ~/src/dryui, or ~/code/dryui. If found, runs bun link in each packages/{ui,lint,primitives,feedback,feedback-server} and writes overrides with link:@dryui/<pkg> so the consumer pulls the local workspace. If not found, falls back to bun add @dryui/ui + bun add -d @dryui/lint @dryui/feedback @dryui/feedback-server against npm.
  7. Runs scripts/setup-feedback-agents.sh "$PWD" to copy project-local DryUI skills, detect installed feedback agents, merge project-local MCP config files where possible, and write dryui.config.json.
  8. Runs bun run check to validate the contract end-to-end.

The smoke-test +page.svelte is intentionally minimal: a single <Heading level={1}> inside <main data-layout="home"> to prove @dryui/ui resolves and to satisfy dryui/no-raw-element (every raw <main> needs a data-layout hook). templates/src/layout.css declares the matching starter [data-layout='home'] rule and owns the container: page / inline-size query context. It carries no design — DryUI does not ship a default look, the user's first prompt fills the page in. @dryui/feedback is a dev-only dep; the live-feedback widget is opt-in (see "Live Feedback (Opt-In)" below).

For non-bun package managers (npm, pnpm, yarn), open scripts/bare-skeleton.sh and translate the install commands. The contract and templates are package-manager-agnostic.

Inspect First (Existing SvelteKit Or Ambiguous Branches)

Read the project shape before changing files:

  1. Check for package.json, svelte.config.*, vite.config.*, src/app.html, src/routes/+layout.svelte, src/app.css, and src/layout.css.
  2. Identify the package manager from lockfiles. Prefer the existing one; use bun only when there is no signal.
  3. Check whether @dryui/ui and @dryui/lint are already installed.
  4. Check whether root layout imports DryUI themes, app CSS, and layout CSS in the right cascade order.
  5. Check whether dryuiLint() and dryuiLayoutCss() are wired.

Svelte MCP Handoff

When the harness has the official Svelte MCP available, the server is normally named svelte. Local setup uses npx -y @sveltejs/mcp; remote setup uses https://mcp.svelte.dev/mcp. Use the MCP tools by name: list-sections first, get-documentation for every relevant Svelte/SvelteKit section, and svelte-autofixer on any changed .svelte, .svelte.ts, or .svelte.js file before final validation. playground-link is only for throwaway snippets, not code written into the user's project.

Golden Consumer Setup Contract

This section is the source of truth for a DryUI consumer setup. Fresh E2E scaffolds may write deterministic files and local tarball overrides, but they should satisfy this contract instead of carrying a separate setup recipe.

A valid DryUI consumer setup has:

  • @dryui/ui and lucide-svelte as runtime dependencies. @dryui/lint, @dryui/feedback, @dryui/feedback-server, and @types/node as dev dependencies. lucide-svelte is a runtime dep because it satisfies the @dryui/feedback peer dep and apps will routinely import icons from it; the vite SSR noExternal workaround in templates/vite.config.ts only kicks in when something imports it. The @dryui/feedback-server devDep ships the dryui-feedback bin into local node_modules/.bin so bunx dryui-feedback resolves without a registry round-trip. @dryui/feedback is dev-only because the live-feedback widget is opt-in (see "Live Feedback (Opt-In)" below). @types/node is a devDep because SvelteKit's generated .svelte-kit/tsconfig.json references Node types and bun run check fails without it. E2E may also pin local workspace tarballs for @dryui/primitives (transitive of feedback) so no published package leaks into the run.
  • dryuiLint({ strict: true }) as the first Svelte preprocessor, preserving any existing preprocessors after it.
  • dryuiLayoutCss() before sveltekit() in Vite plugins.
  • src/app.html ships bare <html lang="en"> — no class="theme-auto", no data-theme. Light tokens apply by default. Apps opt into dark or system mode by adding class="theme-auto" and/or data-theme="…" themselves.
  • src/routes/+layout.svelte importing @dryui/ui/themes/default.css, @dryui/ui/themes/dark.css, ../app.css, and ../layout.css (last), then rendering {@render children()}. No <Feedback> mount by default — opt in per "Live Feedback (Opt-In)".
  • src/app.css present with body { margin: 0; background: var(--dry-color-bg-base); color: var(--dry-color-text-strong); font-family: var(--dry-font-sans); overflow-x: clip; }. It owns visual page paint and the required body font only; no container or grid/flex layout belongs here.
  • src/layout.css present with a matching starter [data-layout='home'] rule that owns container: page / inline-size, display: grid, and the minimal structural track for the smoke-test page. Page/section grid and flex layout lands here, scoped under [data-layout="<name>"], with @container page (...) for responsive shifts.
  • AGENTS.md and CLAUDE.md present from templates/. They instruct Codex/Claude to load skills/dryui-build/SKILL.md before UI work, treat @dryui/ui, @dryui/lint, src/layout.css, and DryUI theme imports as hard DryUI signals, check component metadata before inventing local primitives, and use screenshots for visual validation.
  • dryui.config.json present with feedback.defaultAgent, feedback.detectedAgents, the dryui-feedback MCP command, and manualAgentConfig paths for agents whose config is user-level. The feedback server reads feedback.defaultAgent and feedback.terminalApp at startup, so a detected agent can auto-launch when feedback is submitted.
  • Project-local DryUI skills copied to skills/, .agents/skills/, .claude/skills/, and .codex/skills/ when the sibling skill bundle is available. These are consumer install targets, not canonical source; they give dispatched feedback agents a local dryui-feedback skill path even if the user's global agent install is missing.

Apply Setup

For an existing SvelteKit app:

  1. Install runtime + dev packages:

    bash
    bun add @dryui/ui lucide-svelte
    bun add -d @dryui/lint @dryui/feedback @dryui/feedback-server

    lucide-svelte is a runtime dep: it satisfies the @dryui/feedback peer and is available for the app's own icon usage. @dryui/feedback-server is a devDep so the dryui-feedback bin lands in local node_modules/.bin; bunx dryui-feedback then resolves locally instead of hitting the registry. @dryui/feedback is also a devDep — the widget is opt-in, see "Live Feedback (Opt-In)" below. Translate to npm, pnpm, or yarn when the repo already uses one of those. For local-workspace consumers, mirror the package list in package.json overrides with link:@dryui/<pkg> (including @dryui/primitives, transitive of feedback) before installing — bun overrides only apply when the package is also a dep.

  2. In svelte.config.*, add dryuiLint({ strict: true }) as the first preprocessor while preserving existing preprocessors.

  3. In vite.config.*, add dryuiLayoutCss() before sveltekit().

  4. In src/app.html, leave <html> bare. Don't add class="theme-auto" or data-theme unless the app explicitly wants system-aware or forced dark mode — see dryui-build for the opt-in recipes.

  5. In src/routes/+layout.svelte, import in this order: DryUI theme CSS, app CSS, then ../layout.css last. Render {@render children()} and stop — no widget mounts by default.

  6. Create or update src/app.css so body owns the page surface, text color, and app font: margin: 0; background: var(--dry-color-bg-base); color: var(--dry-color-text-strong); font-family: var(--dry-font-sans); overflow-x: clip;.

  7. Create src/layout.css if missing. Keep it minimal, but make sure every starter data-layout hook has a matching rule. The starter page uses [data-layout='home'] { container: page / inline-size; display: grid; grid-template-columns: minmax(0, 1fr); min-block-size: 100dvh; gap: var(--dry-space-4); align-content: start; }. Future page/section grid blocks land here as routes need them.

  8. Run the feedback agent setup helper:

    bash
    bash <skill-base-dir>/scripts/setup-feedback-agents.sh "$PWD"

    It is best-effort and idempotent. It detects installed agents (codex, claude, gemini, opencode, copilot, VS Code Copilot, cursor, windsurf, zed), picks the first detected agent as feedback.defaultAgent when no default is already configured (DRYUI_DISPATCH_AGENT forces the value for that run), merges project-local MCP config files it can safely own (.mcp.json, .vscode/mcp.json, .cursor/mcp.json, opencode.json), and writes dryui.config.json with manual config paths for user-level agents.

UI Pipeline After Setup

Use this order for the first real interface:

  1. Capture the user's brief in one line: what you are building, and for whom.
  2. Use the dryui-build skill, component metadata, docs pages, and existing repo usage to confirm components, recipes, contracts, accessibility, and tokens.
  3. For Svelte/SvelteKit uncertainty, use the Svelte MCP sequence: list-sections, get-documentation, then svelte-autofixer on changed Svelte files.
  4. Build with DryUI + Svelte 5 runes, grid layout, and --dry-* tokens.
  5. Run the project's check/build/test command to validate contracts, a11y, tokens, and CSS discipline.
  6. For visual work, inspect mobile, tablet, and desktop screenshots. Route status and text matches are smoke signals only, not acceptance.

Live Feedback (Opt-In)

DryUI does not mount the live-feedback widget by default. The @dryui/feedback dev dep is installed and ready, but nothing imports it until the user opts in. To enable for a session, add to src/routes/+layout.svelte:

svelte
<script lang="ts">
	import { Feedback } from '@dryui/feedback';
</script>

{@render children()}

<Feedback serverUrl="http://localhost:4748" />

The widget is dormant until toggled (Cmd+M / Ctrl+M) and respects DRY_FEEDBACK_DISABLED=1 for CI. Then run bunx dryui-feedback --no-open (or bunx dryui-feedback for the local dashboard) when the user kicks off a feedback session. The server reads dryui.config.json; users can edit feedback.defaultAgent manually or ask their agent to add the MCP entry listed under feedback.manualAgentConfig. Remove the import + mount before shipping production.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Dryui Init AI skill do?

Bootstrap a SvelteKit + DryUI project using skill-led inspection and explicit file/package edits. Works for new projects, existing SvelteKit apps, and brownfield repos.

Why use Dryui Init on TypingMind?

Because you install it once and use it with any model. Dryui Init 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 Init in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/rob-balfre/dryui/tree/main/skills/dryui-init. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Dryui Init?

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 Init?

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

Is the Dryui Init 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 👇