beta

AI Chat — `<mv-ai-chat>`

AI chat interface: streaming replies with a caret, shimmering “thinking” state, safe Markdown rendering (lists, bold, code with a copy button), auto-growing composer with attachments, model picker, suggestions, stop / regenerate, and auto-scroll that yields to the user.

Category	Blocks
Type	Web Component (`<mv-ai-chat>`)
Status	beta
Also installs	button, kbd, shimmer-text, textarea
Keywords	ai, chat, assistant, llm, streaming, markdown, composer, conversation

When to use

A product needs a full assistant conversation with streamed replies, a thinking state and stop / regenerate
Users must attach files, pick a model or start from suggested prompts in one composer
Assistant replies contain Markdown lists and code blocks that users will want to copy

Avoid when

Only a single request is typed and its outcome, constraints and limits must be made explicit → use Intent Composer instead
The need is a quick search or command launcher rather than a conversation → use Command instead

Install

node scripts/add.mjs ai-chat --out ./src/marvelous

AI agent with the Marvelous UI MCP server: install_components({ slugs: ["ai-chat"], target_dir: "<absolute path>/src/marvelous", framework: "react" }).

Files copied (dependencies included): tokens/tokens.css, core/base.css, core/dom.js, core/element.js, core/motion.js, components/ai-chat/ai-chat.js, components/ai-chat/ai-chat.css, components/button/button.css, components/kbd/kbd.css, components/shimmer-text/shimmer-text.css, components/textarea/textarea.css, components/textarea/char-count.js.

Usage

Canonical markup — start from it and customize with attributes, data-* and CSS variables:

<mv-ai-chat id="aic-demo" label="Nova" model="nova-pro" style="height:680px;border:1px solid var(--mv-border);border-radius:var(--mv-radius-xl);overflow:hidden">
  <header class="mv-ai-chat-header">
    <mv-menu placement="bottom-start">
      <button class="mv-button mv-ai-chat-model" data-variant="ghost" data-size="sm" type="button">
        <span data-ai-chat-model-label>Nova Pro</span>
        <svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><path d="m6 9 6 6 6-6"/></svg>
      </button>
      <div data-content style="--mv-menu-width:17rem">
        <div class="mv-menu-label">Model</div>
        <div role="group" data-name="model">
          <button role="menuitemradio" value="nova-pro" data-label="Nova Pro" aria-checked="true"><span>Nova Pro<span class="mv-ai-chat-menu-desc">Deep reasoning</span></span></button>
          <button role="menuitemradio" value="nova-flash" data-label="Nova Flash"><span>Nova Flash<span class="mv-ai-chat-menu-desc">Fast everyday answers</span></span></button>
          <button role="menuitemradio" value="nova-code" data-label="Nova Code"><span>Nova Code<span class="mv-ai-chat-menu-desc">Built for coding</span></span></button>
        </div>
      </div>
    </mv-menu>
    <div class="mv-ai-chat-header-end">
      <button class="mv-button" data-variant="ghost" data-size="sm" type="button" data-ai-chat-reset>
        <svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><path d="M12 20h9"/><path d="M16.4 3.6a2 2 0 0 1 2.9 2.9L7 18.8 3 20l1.2-4z"/></svg>
        <span class="mv-ai-chat-header-hide-sm">New chat</span>
      </button>
    </div>
  </header>

  <div class="mv-ai-chat-empty">
    <span class="mv-ai-chat-empty-mark" aria-hidden="true"><svg viewBox="0 0 24 24" fill="currentColor"><path d="M12 2.5c.4 3.9 1.9 6.6 4.2 8 1.3.8 2.9 1.3 4.8 1.5-3.9.4-6.6 1.9-8 4.2-.8 1.3-1.3 2.9-1.5 4.8-.4-3.9-1.9-6.6-4.2-8-1.3-.8-2.9-1.3-4.8-1.5 3.9-.4 6.6-1.9 8-4.2.8-1.3 1.3-2.9 1.5-4.8z"/></svg></span>
    <h2>Hi Maya, ready to dive in?</h2>
    <p>Ask a question, attach a file or start from a suggestion.</p>
    <div class="mv-ai-chat-suggestions">
      <button class="mv-ai-chat-suggestion" type="button" data-prompt="Write a follow-up email about Ms. Chen’s quote"><strong>Write a follow-up</strong><span>about Ms. Chen’s quote</span></button>
      <button class="mv-ai-chat-suggestion" type="button" data-prompt="Summarize September sales"><strong>Summarize sales</strong><span>for September, in 5 points</span></button>
      <button class="mv-ai-chat-suggestion" type="button" data-prompt="Explain Web Components with an example"><strong>Explain Web Components</strong><span>with a code example</span></button>
    </div>
  </div>

  <form class="mv-ai-chat-composer">
    <div class="mv-textarea-group">
      <ul class="mv-ai-chat-attachments" aria-label="Attachments" hidden></ul>
      <textarea class="mv-textarea" rows="1" placeholder="Message Nova…" aria-label="Message Nova" aria-keyshortcuts="Enter"></textarea>
      <div class="mv-textarea-toolbar">
        <label class="mv-button mv-ai-chat-attach" data-variant="ghost" data-size="icon" title="Attach a file">
          <svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><path d="m21 11-8.6 8.6a5.5 5.5 0 0 1-7.8-7.8l8.6-8.6a3.7 3.7 0 0 1 5.2 5.2l-8.6 8.6a1.8 1.8 0 0 1-2.6-2.6L15 7"/></svg>
          <input type="file" multiple aria-label="Attach files">
        </label>
        <span class="mv-textarea-toolbar-spacer"></span>
        <button class="mv-button" data-size="icon" type="submit" data-ai-chat-send aria-label="Send message">
          <svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2.2" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><path d="M12 19V5M5 12l7-7 7 7"/></svg>
          <svg viewBox="0 0 24 24" fill="currentColor" aria-hidden="true"><rect x="6" y="6" width="12" height="12" rx="2"/></svg>
        </button>
      </div>
    </div>
    <p class="mv-ai-chat-hint"><kbd class="mv-kbd" data-size="sm">Enter</kbd> to send · <kbd class="mv-kbd" data-size="sm">Shift</kbd> + <kbd class="mv-kbd" data-size="sm">Enter</kbd> for a new line · Nova can make mistakes. Check important info.</p>
  </form>

  <script type="application/json">[
    { "role": "user", "content": "How do I build a Web Component that shows a counter?", "files": [{ "name": "counter-brief.pdf", "size": 184320 }] },
    { "role": "assistant", "content": "Here’s a minimal, dependency-free counter:\n\n```js\nclass MvCounter extends HTMLElement {\n  #count = 0;\n  connectedCallback() {\n    this.innerHTML = `<button type=\"button\">0</button>`;\n    this.querySelector(\"button\").addEventListener(\"click\", (e) => {\n      e.currentTarget.textContent = ++this.#count;\n    });\n  }\n}\ncustomElements.define(\"mv-counter\", MvCounter);\n```\n\nA few best practices:\n\n- prefix the tag name (`mv-`) to avoid collisions\n- clean up your listeners in `disconnectedCallback()`\n- expose state through **attributes** to stay compatible with every framework" }
  ]</script>
</mv-ai-chat>

API

Attributes

Name	Type	Default	Description
`label`	string	`Assistant`	Assistant name (accessible labels, announcements).
`model`	string		Current model; synced with a radio menu data-name="model" and passed to the responder.
`user-label`	string	`You`	Accessible name for the user’s messages.
`thinking-label`	string	`Thinking`	Text for the waiting state.
`data-state`	thinking \| streaming		Set while generating.
`data-empty`	boolean		Set when the conversation is empty.

Properties

Name	Type	Description
`responder`	(messages, { signal, model }) => AsyncIterable`<string>` \| Promise`<string>` \| string	Response source. Default: local simulation (canned English replies, no network). Honor signal for stopping.
`messages`	{ role, content, files? }[]	Reads/writes the history. Also settable via a child `<script type="application/json">`.
`busy`	boolean	Read-only: a response is in progress.

Methods

Name	Description
`send(text)`	Sends a message (with pending attachments) and generates the response.
`stop()`	Stops generating (also Escape in the field, or the send button while it reads Stop).
`regenerate()`	Replaces the last response.
`reset()`	Clears the conversation.
`renderMarkdown(source, element)`	Export: lightweight, safe Markdown rendering (textContent only).

Events

Name	Description
`mv-send`	Cancelable. detail: { content, files }.
`mv-response`	Response finished. detail: { content, stopped }.
`mv-stop / mv-reset / mv-error`	Stop, reset, responder error (detail.error).

Content structure

Name	Description
`.mv-ai-chat-header`	Free-form header; [data-ai-chat-model-label] shows the selected model, [data-ai-chat-reset] clears the conversation.
`.mv-ai-chat-empty`	Empty state (hidden after the first message); any [data-prompt] sends its text.
`form.mv-ai-chat-composer`	Composer: textarea (Enter sends, Shift + Enter = new line), input[type=file], ul.mv-ai-chat-attachments, [data-ai-chat-send] button with two icons (send / stop).

CSS classes

Name	Description
`mv-ai-chat-log / -thread / -message / -content / -code / -actions`	Generated parts (message: data-role user \| assistant, data-pending).
`mv-ai-chat-suggestions / -suggestion / -empty-mark`	Suggestion grid and empty-state badge.
`mv-ai-chat-chip / -attach / -hint / -model / -menu-desc`	Attachment, file button, hint, model trigger and model description.

CSS variables

Name	Default	Description
`--mv-ai-chat-max-width`	`46rem`	Reading width of the thread and composer.
`--mv-ai-chat-user-bg`		Background of the user’s bubbles.

Accessibility

Thread with role=log (aria-busy while generating); each message is a named <article> (“You” / assistant name). Streaming isn’t read word by word: a role=status region announces “… is writing a reply”, then the final response (code is summarized). Send button: label toggles between Send / Stop. Focusable code blocks (horizontal scrolling by keyboard), named Copy / Regenerate buttons. Reduced motion: no typing animation or blinking caret.