product-design

lobehub/lobehub · Agent Skill

Creates interactive HTML prototypes using the real LobeHub UI stack, focusing on interaction rather than code style.

Jump to install ↓

Overview

PublisherlobehubOrganization
Repositorylobehub/lobehub ↗
Skill folderdesign-prototype
Stars82,305
Bundled files1
Licensenot specified — check with the author
Last updatedSeptember 6, 2026
LinksSkill source ↗ · Repository ↗

Mozgi check

What we found in the skill files. Details come with a subscription.

Security
medium risk · 1 point needs attention
Compatibility
instructions only — works with most agents
Similar skills
139 similar skills found
Also with subscription

Installation

A skill is a folder with a SKILL.md file. Pick your agent and run the commands in a terminal (macOS, Linux or Git Bash on Windows) from your project root.

Copy the prompt into Claude, Codex or another assistant: it will review the skill and install it for you.

Install the “product-design” skill from mozgi.io (https://mozgi.io/skills/product-design/) for me.

1. Download the archive: https://mozgi.io/skills/product-design/download
   If it fails, take the skill folder from the source: https://github.com/lobehub/lobehub/tree/canary/.agents/skills/design-prototype
2. Read SKILL.md and the other files. Check the skill is safe: no hidden commands, no sending data elsewhere, no scripts doing something other than described. If anything looks suspicious, stop and tell me.
3. Put the “product-design” folder into your agent's skills folder: .claude/skills/ for Claude Code, .agents/skills/ for Codex, .cursor/skills/ for Cursor, .gemini/skills/ for Gemini CLI. Inside the project if we are in one, otherwise in the home folder (~/.claude/skills/ etc.).
4. Make sure <skills folder>/product-design/SKILL.md exists.
5. Briefly tell me what this skill does and offer to try it on my task.

If you cannot do a step yourself, tell me what to click or type.

Review the source files

1 file Download Zip

Read SKILL.md and any companion files before deciding whether to install. Click a file to preview it.

SKILL.md contents

Original author text · read-only

Design Prototype

One HTML file, double-click to open, fully interactive, rendered by the **actual

design system** (@lobehub/ui 5.x + antd 6 + antd-style tokens, exact versions from

this repo's node_modules) — while the source stays production-style React

(import { Block } from '@lobehub/ui', createStyles(({ css, token }) => …), hooks,

memo). Promotion to production is mostly "split into files", not "rewrite".

Prototype code may be quick-and-dirty (one file, no i18n, inline data) — but the

interaction must be complete: states, transitions, and affordances are the point

of a prototype, not its code style.

How it works

one-time (~2s, cached):   scripts/build-runtime.sh  →  lobe-prototype-runtime.js (IIFE global)
                                                       + vendored babel.min.js
per prototype (no build): single HTML  =  <script runtime> + babel-standalone
                          + <script type="text/babel"> with production-style React
  • The runtime bundles react, react-dom/client, @lobehub/ui, @lobehub/ui/base-ui,

antd (curated subset), antd-style, lucide-react from the repo's own

node_modules — so tokens/components/versions match production, and react/emotion/

theme-context are singletons by construction.

  • A 3-line window.require shim + babel's transform-modules-commonjs lets the

prototype keep real import statements.

  • ThemeProvider themeMode="auto" → light/dark follow the OS.

> ⚠️ Don't re-attempt the pure-CDN route (esm.sh import maps). It was tried and

> rejected: esm.sh's CJS named-export analysis fails serially across antd's dep chain

> (@ant-design/fast-color → @ant-design/colors → @rc-component/qrcode → …), and

> ?bundle leaks an unversioned react-is 404. Whack-a-mole, not a foundation.

Quickstart

1. Build the runtime into the prototype's directory (any dir works; /tmp is fine):

```bash

bash .agents/skills/design-prototype/scripts/build-runtime.sh /tmp/my-proto

```

Skip if lobe-prototype-runtime.js + babel.min.js are already there and the

design-system versions haven't bumped.

2. Copy references/template.html ↗ into the same dir,

rename, and replace the sample App with the real surface. Keep the runtime shim

block untouched.

3. Open it (open /tmp/my-proto/xxx.html) — no server, no build.

Verify before delivering

Headless-check it renders with zero console errors (the repo's e2e Playwright works):

// node /tmp/check.mjs   — adjust paths
import { chromium } from '<repo>/e2e/node_modules/playwright/index.mjs';
const b = await chromium.launch();
const p = await b.newPage();
const errs = [];
p.on('pageerror', (e) => errs.push(String(e)));
await p.goto('file:///tmp/my-proto/xxx.html');
await p.waitForSelector('#root *', { timeout: 30000 });
await p.screenshot({ path: '/tmp/proto.png' }); // Read the screenshot yourself
console.log('errors:', errs);
await b.close();

Then Read the screenshot — a prototype is a visual deliverable; don't ship it

sight-unseen. Check light + dark (emulate prefers-color-scheme) and a narrow

viewport if the surface has a mobile story.

What's in the runtime (extend freely)

See assets/entry.mjs ↗ — the single source of truth. Currently:

  • @lobehub/ui: ActionIcon, Alert, Avatar, Block, Button, Center, Collapse,

DraggablePanel, Drawer, DropdownMenu, Empty, Flexbox, Highlighter, Hotkey, Icon,

Image, Input, InputNumber, Markdown, Modal, NeuralNetworkLoading, Popover,

ScrollShadow, SearchBar, Segmented, Select, Skeleton, SortableList, Tabs, Tag,

Text, TextArea, ThemeProvider, Tooltip

  • @lobehub/ui/base-ui: full namespace (Select, Modal, DropdownMenu, Switch, Toast,

FloatingSheet, …)

  • antd: App, Badge, Checkbox, Divider, Dropdown, Progress, Radio, Slider, Space,

Steps, Table

  • antd-style, lucide-react, react, react-dom/client: full namespaces

Missing a component → add to entry.mjs, rerun the build script (\~2s). Sizes:

curated ≈ 16MB (fine from disk); a full export * from '@lobehub/ui' works too

(≈ 27MB) if you'd rather never curate.

Pitfalls

  • Switch lives in @lobehub/ui/base-ui, not the root package (as do the other

base-ui primitives). Importing it from the root fails the runtime build.

  • Use createStyles (runtime), not createStaticStyles — static extraction

needs a build step. This is the one sanctioned deviation from production style;

note it when handing the prototype to an implementer.

  • babel must stay the raw UMD file (unpkg/jsdelivr). esm.sh rewrites it to ESM

and a classic <script src> chokes on export.

  • The require shim throws with the module name when an import isn't in the runtime —

that's the "add to entry.mjs and rebuild" signal, not a template bug.

  • Text type accepts only secondary|success|warning|danger|info; Tag color has

no primary (DESIGN.md "Applying tokens in components").

Design bar (same as production surfaces)

A prototype is a surface — the ux checklists apply to what it

_demonstrates_:

  • Show at least one non-happy-path state (empty / loading / error / in-progress),

ideally behind a state-toggle strip like the template's. A happy-path-only

prototype under-specifies the design and silently blesses missing states.

  • Reuse the app's surface contracts: side panels are DraggablePanel

(collapse + drag-resize come free), loading is skeleton/NeuralNetworkLoading

(never antd Spin), modals via createModal-style flows.

  • Don't paint affordances you don't wire (cursor: zoom-in with no zoom, keycap

chips with no keys) — in an _interactive_ prototype a dead affordance is a spec bug.

  • Walk the ux Quick review against the prototype before delivering; annotate

anything deliberately out of scope in an HTML comment so the implementer knows

it's a cut, not a decision.

Frequently asked questions

What does the product-design skill do?

Creates interactive HTML prototypes using the real LobeHub UI stack, focusing on interaction rather than code style.

How do I install product-design?

Open the “Installation” block on this page and pick your agent — Claude Code, Codex, Cursor or Gemini CLI — for ready-made commands that copy the skill folder to the right place. Or copy the prompt from the “Via assistant” tab and let the assistant review and install the skill.

Which AI models work with product-design?

Skills are plain Markdown instructions (the Agent Skills format), so product-design works with any agent that reads SKILL.md: Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode and more. The model loads the instructions only when a task matches the skill's description.

Is product-design free?

The skill is published openly by lobehub. The skill itself costs nothing — you only pay your AI provider for the tokens you use. No license is specified in the repository — check the terms with the author before commercial use.

Is it safe to install skills?

A skill is a set of instructions your agent will follow, so read SKILL.md and its companion files before installing. Scripts in the skill folder only run if your agent runs them — keep an eye on what it does and do not grant unnecessary permissions.

How is a skill different from an MCP server or a plugin?

An MCP server or plugin gives the model new tools — code that runs and returns a result. A skill gives it knowledge and process: how to approach a task, which steps to follow and what good output looks like. It needs no server or API key.

Related skills

All skills →