---
title: Skills
description: Give your AI assistant knowledge of Nexus UI components and patterns
---

Skills give AI coding agents like Cursor, Claude Code, and Windsurf project-aware context about Nexus UI. When installed, your assistant knows how to find, install, compose, and customize components using the correct APIs and patterns.

For example, you can ask your assistant to:

- _"Add a chat input with send and attach buttons."_
- _"Build a streaming chat interface with messages and a prompt input."_
- _"Show me all available Nexus UI components."_

The skill provides your assistant with component APIs, composition patterns, installation commands, and AI SDK integration — so it generates correct code on the first try.

## Install

<Tabs items={["npm", "pnpm", "yarn", "bun"]}>
<Tab value="npm">
```bash
npx skills add victorcodess/nexus-ui
```
</Tab>
<Tab value="pnpm">
```bash
pnpm dlx skills add victorcodess/nexus-ui
```
</Tab>
<Tab value="yarn">
```bash
npx skills add victorcodess/nexus-ui
```
</Tab>
<Tab value="bun">
```bash
bunx skills add victorcodess/nexus-ui
```
</Tab>
</Tabs>

Once installed, your AI assistant automatically loads it when working with Nexus UI components. Learn more at [skills.sh](https://skills.sh).

## What's included

The skill provides your AI assistant with:

### Component APIs

Full reference for all Nexus UI components — sub-components, props, import paths, and composition patterns. The assistant knows the correct way to combine `PromptInput`, `PromptInputTextarea`, `PromptInputActions`, and other primitives.

### Installation & CLI

How to add the `@nexus-ui` registry to `components.json`, install components via the shadcn CLI, and manage dependencies.

### AI SDK integration

Patterns for using Nexus UI with the [Vercel AI SDK](https://sdk.vercel.ai) — connecting `useChat` to `PromptInputTextarea`, handling form submission, and wiring up streaming.

### Styling & customization

How to override styles via `className`, use Tailwind CSS v4 design tokens, and customize components after installation.

## MCP Server

The [Model Context Protocol](https://modelcontextprotocol.io) lets AI tools access Nexus UI components directly from your editor.

### Cursor

Create a `.cursor/mcp.json` file in your project root:

```json title=".cursor/mcp.json"
{
  "mcpServers": {
    "nexus-ui": {
      "description": "Nexus UI registry",
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "shadcn@canary", "mcp"],
      "env": {
        "REGISTRY_URL": "https://nexus-ui.dev/api/registry/registry.json"
      }
    }
  }
}
```

Restart Cursor after adding the configuration.

### Other AI tools

For other tools that support MCP (Claude Code, Windsurf, etc.), use the same server configuration above and follow your tool's MCP setup docs.

### Usage

Once configured, you can interact with the registry directly in chat:

```
Show me all available Nexus UI components
```

```
Add the PromptInput component from Nexus UI
```

## LLM resources

Machine-readable documentation for LLMs:

| Resource | URL |
|----------|-----|
| llms.txt | [nexus-ui.dev/llms.txt](https://nexus-ui.dev/llms.txt) |
| llms-full.txt | [nexus-ui.dev/llms-full.txt](https://nexus-ui.dev/llms-full.txt) |