ask-user
An MCP App that enables AI agents to ask users multiple questions with tab-based navigation, multiple-choice options, multi-select support, and custom text input, all rendered inline in the conversation.
README
Ask User MCP App
An MCP App that enables AI agents to ask users multiple questions with tab-based navigation, multiple-choice options, multi-select support, and custom text input - all rendered inline in the conversation.
Features
- Multi-Question Tabs: Ask multiple questions at once, displayed as navigable tabs
- Keyboard Navigation: Tab/Shift+Tab to navigate tabs, Arrow keys for options, Enter to select/submit
- Next/Submit Button: Each question has a Next button to advance to the next tab (or "Review & Submit" on the last question)
- Multiple Choice Questions: Present 2-4 options per question
- Single & Multi-Select: Support both radio-button style (single) and checkbox style (multi) selection
- Custom "Other" Input: Optional text input for answers not covered by predefined options
- Skip Questions: Non-required questions can be skipped
- Theme Support: Automatically adapts to host's light/dark theme
- Dual Transport: Works with both HTTP (web clients) and stdio (desktop clients)
Installation in Claude.ai (Hosted)
Use the hosted MCP server directly with Claude's connector feature:
- Open Claude.ai and go to Settings → Connectors
- Click Add custom connector and enter:
- Remote MCP server URL:
https://ask-user-mcp-app-seven.vercel.app/mcp
- Remote MCP server URL:
Browse the hosted landing page at ask-user-mcp-app-seven.vercel.app for a quick overview of the tool schema and install steps. 3. Save and start a new conversation
The ask_user tool will now be available to Claude.
Installation
Or clone and build locally:
git clone https://github.com/anthropics/ask-user-mcp-app
cd ask-user-mcp-app
npm install
npm run build
Usage
Local Development (stdio)
Build and run locally with Claude Desktop:
# Clone and build
git clone https://github.com/anthropics/ask-user-mcp-app
cd ask-user-mcp-app
npm install
npm run build
Add to your claude_desktop_config.json (use the absolute path to your clone):
{
"mcpServers": {
"ask-user": {
"command": "node",
"args": ["/absolute/path/to/ask-user-mcp-app/dist/main.js", "--stdio"]
}
}
}
Tool Schema
The ask_user tool accepts a questions array containing one or more questions to display as tabs:
Top-Level Parameters
| Parameter | Type | Required | Description |
|---|---|---|---|
questions |
array | Yes | Array of question objects (minimum 1) |
Question Object Parameters
| Parameter | Type | Required | Description |
|---|---|---|---|
question |
string | Yes | The question to ask (also used as unique identifier) |
header |
string | Yes | Short label displayed as tab name (max 12 chars) |
options |
array | Yes | 2-4 choices, each with label, value, and optional description |
multiSelect |
boolean | No | Allow multiple selections (default: false) |
allowOther |
boolean | No | Include "Other" text input option (default: true) |
required |
boolean | No | Whether this question must be answered (default: false) |
Example Tool Call
{
"name": "ask_user",
"arguments": {
"questions": [
{
"question": "Which frontend framework would you like to use for this project?",
"header": "Framework",
"options": [
{ "label": "React", "value": "react", "description": "Popular library with component-based architecture" },
{ "label": "Vue", "value": "vue", "description": "Progressive framework for simplicity" },
{ "label": "Svelte", "value": "svelte", "description": "Compiler-based, highly optimized" }
],
"required": true
},
{
"question": "How should we handle authentication?",
"header": "Auth Method",
"options": [
{ "label": "OAuth 2.0", "value": "oauth", "description": "Industry standard, supports SSO" },
{ "label": "JWT", "value": "jwt", "description": "Stateless, good for APIs" },
{ "label": "Session-based", "value": "session", "description": "Traditional, server-side state" }
]
},
{
"question": "Which testing frameworks should we include?",
"header": "Testing",
"options": [
{ "label": "Jest", "value": "jest", "description": "Popular testing framework" },
{ "label": "Vitest", "value": "vitest", "description": "Fast, Vite-native testing" },
{ "label": "Playwright", "value": "playwright", "description": "E2E testing" }
],
"multiSelect": true
}
]
}
}
Response Format
When the user submits their answers, the response is formatted as:
Which frontend framework would you like to use for this project? -> React
How should we handle authentication? -> JWT
Which testing frameworks should we include? -> Vitest, Playwright
Keyboard Navigation
| Key | Action |
|---|---|
| Tab / Shift+Tab | Navigate between question tabs |
| Arrow Left / Right | Navigate between question tabs |
| Arrow Up / Down | Navigate between options in current question |
| Enter / Space | Select current option |
| Enter (on Submit tab) | Submit all answers |
Development
Commands
| Command | Description |
|---|---|
npm run build |
Build server and UI |
npm run build:server |
Build only server (TypeScript) |
npm run build:ui |
Build only UI (Vite + React) |
npm run serve |
Start HTTP server |
npm start |
Build and serve |
npm run dev |
Development mode with tsx |
Testing Locally
-
Start the server:
npm start -
Use the basic-host from the ext-apps repo:
cd ext-apps/examples/basic-host SERVERS='["http://localhost:3001/mcp"]' npm start -
Open http://localhost:8080 and test the
ask_usertool
Architecture
┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
│ Host Client │────▶│ MCP Server │────▶│ React UI │
│ (Claude, etc.) │ │ (server.ts) │ │ (mcp-app.tsx) │
└─────────────────┘ └─────────────────┘ └─────────────────┘
│ │ │
│ Tool call with │ Serves bundled │
│ ui/resourceUri │ HTML via resource │
│ │ │
└───────────────────────┴───────────────────────┘
postMessage communication
Key Files
- server.ts: Registers the
ask_usertool andui://resource - main.ts: Entry point supporting HTTP and stdio transports
- mcp-app.tsx: React UI with multi-question state management
- vite.config.ts: Bundles UI into single HTML file via
vite-plugin-singlefile
UI Components
| Component | Purpose |
|---|---|
TabBar |
Pill-style tab navigation with progress indicator |
QuestionPanel |
Renders individual question with refined card styling |
SubmitTab |
Review page showing all answers with staggered animations |
OptionList / OptionButton |
Elegant option cards with smooth transitions |
OtherInput |
Custom text input with auto-focus behavior |
Design System
The UI uses a warm, Claude-native color palette for seamless integration:
- Surfaces: Warm cream backgrounds (
#faf8f5,#f5f1eb) - Accent: Terra cotta (
#c4704b) for selections and CTAs - Text: Warm gray hierarchy with subtle undertones
- Interactions: Smooth 150ms transitions, focus rings, hover states
- Dark mode: Full support with warm dark palette
Navigation Hooks
| Hook | Purpose |
|---|---|
useTabNavigation |
Handles Tab/Shift+Tab and arrow key navigation between tabs |
useOptionNavigation |
Handles arrow key navigation and Enter/Space selection within options |
Dependencies
@modelcontextprotocol/ext-apps- MCP Apps SDK@modelcontextprotocol/sdk- Core MCP SDKreact/react-dom- UI frameworkexpress/cors- HTTP serverzod- Schema validationtailwindcss- Styling
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.