agentsclimarketplace

Agentic webdesign zephyr

Skill S3YED/appie-kit/skills/agentic/agentic-webdesign-zephyr

Build Your Own AI Employee. The complete starter kit for OpenClaw + Hermes Agent. 155 deduplicated skills, drag-and-drop workspace, case studies, install scripts.

Install
npx -y skills add S3YED/appie-kit --skill agentic-webdesign-zephyr

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 6 stars6 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use Zephyr Framework for zero-JS interactive web components with built-in MCP server, A2UI catalog, and Agent API. AI agents control live UI through typed tools, not generated markup.

SKILL.md

4.4 KB, as published. Nobody here has run it

Zephyr Framework — Zero-JS Agentic Web Components

What it is

"The first UI framework built for AI agents." 14 interactive web components (modal, tabs, accordion, toast, combobox, carousel, select, dropdown, datepicker, infinite scroll, sortable, file upload, virtual list, etc.) with pure CSS interactions. Zero runtime JavaScript after initial ~7KB page load. MCP server gives AI agents typed tools to control live components — they don't generate HTML, they interact with existing components.

Quick start

# Scaffold new project
npx create-zephyr-framework my-app
cd my-app && npm start

# Or via CDN (zero build)
# Add two script tags, use components directly

Core Packages

PackageDescription
zephyr-frameworkCore: 14 web components, CSS, agent API, A2UI catalog
zephyr-framework-mcpMCP server: bridges Claude Desktop/Cursor/Windsurf to live components
create-zephyr-frameworkCLI scaffolder with starter page

MCP Tools (6 typed tools)

// Agents call these on live components
zephyr_get_state()      → JSON snapshot of all component states
zephyr_get_schema()     → Schema for all 14 components
zephyr_get_prompt()     → LLM context for prompting
zephyr_act(selector, action, args?)  → Execute action: open modal, switch tab, etc.
zephyr_describe(selector) → Deep inspect one component
zephyr_render(html)     → Render HTML into page

Agent doesn't see raw HTML — it sees typed tools with structured I/O:

zephyr_act('#modal-demo', 'open') → { success: true }
zephyr_get_state() → [{ id: "modal-demo", state: "open", actions: ["close", "toggle"] }, ...]

Agent API (Browser)

// Direct browser access for JS-based agents
Zephyr.agent.getState()     // Snapshot all components
Zephyr.agent.act('#id', 'action')  // Act on component
Zephyr.agent.describe('#id')       // Deep inspect
Zephyr.agent.render(html)          // Render HTML
Zephyr.agent.compose(components)   // Compose layout
Zephyr.agent.observe(callback)     // Watch state changes
Zephyr.agent.record() / replay()   // Record/replay interactions
Zephyr.agent.headless()            // Toggle fast mode

A2UI Integration

Zephyr publishes zephyr-a2ui-catalog.json — Google's Agent-to-UI protocol catalog. Any A2UI-compatible agent (Gemini, CopilotKit) discovers Zephyr components without integration code.

Drop-in Chat Widget

<zephyr-agent-chat data-api-key="sk-..."></zephyr-agent-chat>

Visitors type natural language, agent controls UI. Works with Anthropic and OpenAI. Proxy mode for production.

CSS-Only Interactions

  • Modal: CSS Grid + :has() selector for open/close — no JS
  • Tabs: View Transitions API for smooth slide animations
  • Accordion: CSS :has() + grid-template-rows transition
  • Carousel: CSS scroll-snap + :has() for prev/next
  • Toast: CSS animations with auto-dismiss via animation-delay
  • Dropdown/Combobox: :focus-within + keyboard nav

All GPU-accelerated, 60fps, zero main-thread overhead during interaction.

Comparison

ZephyrMCP AppsAG-UIGenerative UI
FrameworkNone (Web Components)ReactReactReact
Build stepNoYesYesYes
Agent generates markupNoYesNoYes
Pre-built components14 + dashboardNoneNoneNone
MCP integrationYesNativeNoNo
A2UI catalogYesNoNoNo

MCP Client Setup

{
  "mcpServers": {
    "zephyr": {
      "command": "npx",
      "args": ["zephyr-framework-mcp"],
      "env": {
        "ZEPHYR_ROOT": "/path/to/project",
        "ZEPHYR_MCP_PORT": "3456"
      }
    }
  }
}

Pitfalls

  • Components require the framework JS to register custom elements (one-time, ~7KB)
  • MCP server needs a running browser/page to connect to (WebSocket bridge on port 3456)
  • Chat widget in production should use proxy mode (don't expose API keys in HTML)
  • Dashboard add-on and runtime add-on are separate packages

Version

MIT License. Active development. npm packages published and maintained.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.