React Context MCP
Enables AI assistants to inspect React component trees, props, state, and source locations in Chrome, facilitating UI debugging and analysis.
README
<div align="center"> <img src="logo.png" alt="React Context MCP" width="600"> <br><br>
</div>
Instant React Component Discovery for AI Assistants
The Problem: You see a "Sign up" button and need to find which React component renders it, what file it's in, and what props it has.
With React Context MCP: Ask your AI assistant and get a complete component tree with accessibility information instantly.
// You ask your AI:
"Show me the component tree for this page"
// Your AI calls:
get_component_map()
// You get:
React Component Tree:
OnboardingPage (src/pages/OnboardingPage.tsx:25:4)
└─ OnboardingScreen {content={...}, onSignUp={fn}, onLogIn={fn}} (src/pages/OnboardingPage.tsx:136:8)
└─ Box {display="flex", flexDirection="column"} (src/design-system/OnboardingScreen.tsx:138:4)
└─ Typography {as="h1", variant="h2Bold"} [role="heading" name="Create Account"] (src/components/Typography.tsx:103:10)
└─ h1 [role="heading" name="Create Account"]
└─ Button {size="large", onClick={fn}} [role="button" name="Sign up"] (src/components/Button.tsx:115:12)
└─ button [role="button" name="Sign up"]
What You Can Do
Ask your AI assistant to:
- Get complete component trees - See all React components with props, source locations, and accessibility structure
- Find any UI element - Trace buttons, inputs, or any element to its React component
- Inspect component details - Get props, state, and owner chains for any component
- Navigate multi-page flows - Analyze components across different screens
react-context-mcp is a Model Context Protocol (MCP) server that connects your AI assistant to React applications running in Chrome, providing instant access to component trees, props, state, and source locations.
How It Works
The Complete Picture: get_component_map
When you ask "Show me the component tree", your AI calls:
get_component_map({ verbose: true })
Returns a markdown tree showing:
- All React components (Button, TextField, OnboardingScreen, etc.)
- Component props in JSX format (
size="large",onClick={fn}) - Accessibility information (role, accessible name for screen readers)
- DOM elements with semantic roles (button, h1, p, img)
- Source locations (file:line:column)
React Component Tree:
App (src/main.tsx:8:4)
└─ OnboardingScreen {onSignUp={fn}, onLogIn={fn}} (src/pages/OnboardingPage.tsx:136:8)
└─ Stack {direction="column", gap="3"} (src/design-system/OnboardingScreen.tsx:216:8)
└─ Text {variant="h1"} [role="heading" name="Send instantly"] (src/components/Text.tsx:222:12)
└─ h1 [role="heading" name="Send instantly"]
└─ Text {variant="body-secondary"} [role="paragraph"] (src/components/Text.tsx:232:12)
└─ p [role="paragraph"]
└─ Button {variant="primary", size="large"} [role="button" name="Sign up"] (src/components/Button.tsx:361:10)
└─ button [role="button" name="Sign up"]
Focused Element Inspection
For specific element details, use the two-step process:
Step 1: Take a snapshot to get element IDs
take_snapshot({ verbose: true })
Returns the accessibility tree with backendDOMNodeId for every element:
{
"role": "button",
"name": "Sign up",
"backendDOMNodeId": 48
}
Step 2: Get React component details
get_react_component_from_backend_node_id(48)
Returns complete component information:
{
"component": {
"name": "Button",
"type": "ForwardRef",
"source": {
"fileName": "src/components/Button.tsx",
"lineNumber": 42,
"columnNumber": 8
},
"props": {
"variant": "primary",
"size": "large",
"onClick": "[Function]",
"children": "Sign up"
},
"owners": [
{
"name": "OnboardingScreen",
"source": "src/screens/OnboardingScreen.tsx:222:12"
}
]
}
}
Requirements
- Node.js v20.19+ or v22.12+ or v23+
- Chrome current stable version
- React application with development build (for source location tracking)
Getting Started
Quick Install
Add to your MCP client configuration:
{
"mcpServers": {
"react-context": {
"command": "npx",
"args": ["-y", "react-context-mcp@latest"]
}
}
}
[!NOTE] Using
@latestensures you always get the most recent version.
MCP Client Setup
<details> <summary><b>Claude Code</b></summary>
Use the Claude Code CLI:
claude mcp add react-context npx react-context-mcp@latest
</details>
<details> <summary><b>Cursor</b></summary>
Go to Cursor Settings → MCP → New MCP Server, then add:
{
"mcpServers": {
"react-context": {
"command": "npx",
"args": ["-y", "react-context-mcp@latest"]
}
}
}
</details>
<details> <summary><b>Cline / Windsurf / Other Clients</b></summary>
Add the configuration above to your MCP settings file. Refer to your client's documentation for the config file location.
</details>
First Prompt
Try this in your MCP client:
Navigate to http://localhost:3000 and show me the component tree
Your AI assistant will open the browser, navigate to the page, and display the complete React component hierarchy with accessibility information.
Source Location Tracking
⚠️ IMPORTANT: To get accurate component source locations (file name, line number), you must configure the Babel plugin in your React project.
Why Is This Required?
React Context MCP extracts source locations from data-inspector-* DOM attributes added by Babel. React 19 removed the _debugSource fiber property, making the Babel plugin approach the only reliable method for source tracking.
Without the plugin:
- ❌ Component source locations will show as
undefined - ✅ Component names, props, and tree structure work normally
With the plugin:
- ✅ Exact file paths (e.g.,
src/components/Button.tsx) - ✅ Precise line and column numbers
- ✅ Complete component hierarchy with sources
Configuration
Vite
npm install --save-dev @react-dev-inspector/babel-plugin
Add to vite.config.ts:
import { defineConfig } from 'vite'
import react from '@vitejs/plugin-react'
export default defineConfig({
plugins: [
react({
babel: {
plugins: [
['@react-dev-inspector/babel-plugin', {
excludes: ['node_modules']
}]
]
}
})
]
})
Next.js / CRA / Manual Babel
See the detailed configuration instructions for Next.js, Create React App, and manual Babel setup in the react-dev-inspector documentation.
MCP Tools
Page Management (5 tools)
list_pages- List all open browser tabsselect_page- Select a tab to work withclose_page- Close a specific tabnew_page- Open a new tab and navigate to URLnavigate_page- Navigate, reload, or go back/forward
React Inspection (3 tools)
get_component_map
Primary tool - Get complete React component tree as markdown
Arguments:
verbose(boolean, optional) - Include all DOM elements (default: true)includeState(boolean, optional) - Include component state (default: false)
Response:
React Component Tree:
App (src/App.tsx:10:4)
└─ Button {variant="primary", size="large"} [role="button" name="Sign up"] (src/Button.tsx:42:5)
└─ button [role="button" name="Sign up"]
Shows:
- React component hierarchy
- Props in JSX format (
prop="value",prop={value}) - ARIA attributes ([role="..." name="..."])
- DOM elements with semantic roles (button, h1, p, img, etc.)
- Source locations (file:line:column)
take_snapshot
Get accessibility tree with element IDs
Arguments:
verbose(boolean, optional) - Include all elements (default: false)
Response:
{
"role": "RootWebArea",
"name": "My App",
"children": [
{
"role": "button",
"name": "Sign up",
"backendDOMNodeId": 48
}
]
}
Use backendDOMNodeId with get_react_component_from_backend_node_id for detailed component inspection.
get_react_component_from_backend_node_id
Get React component details using backendDOMNodeId from snapshot
Arguments:
backendDOMNodeId(number) - From take_snapshot
Response:
{
"success": true,
"component": {
"name": "Button",
"type": "ForwardRef",
"source": {
"fileName": "src/components/Button.tsx",
"lineNumber": 42
},
"props": {"variant": "primary", "children": "Sign up"},
"owners": [
{"name": "OnboardingScreen", "source": {...}},
{"name": "App", "source": {...}}
]
}
}
Benefits:
- ✅ Fastest method for component lookup
- ✅ Returns complete owner chain (parent components)
- ✅ Precise source locations
Important: backendDOMNodeId is only valid within the same browser session.
Command-Line Options
# Auto-navigate on startup
TARGET_URL=http://localhost:3000 react-context-mcp
# Connect to existing Chrome with remote debugging
react-context-mcp --browserUrl http://localhost:9222
# Isolated mode (separate Chrome profile)
react-context-mcp --isolated --headless
# Custom Chrome executable
react-context-mcp --executablePath /path/to/chrome
# Set viewport size
react-context-mcp --viewport 1920x1080
Available flags:
--headless- Run Chrome in headless mode--isolated- Use isolated user data directory--browserUrl <url>- Connect to existing Chrome debugging session--wsEndpoint <url>- WebSocket endpoint for CDP--executablePath <path>- Path to Chrome executable--channel <channel>- Chrome channel (stable, canary, beta, dev)--viewport <WxH>- Viewport size (e.g., 1280x720)
Troubleshooting
Browser Already Running
Use --isolated flag:
react-context-mcp --isolated
Missing Source Locations
- Requires development build
- Add Babel plugin (see Source Location Tracking section)
- Restart dev server after configuration changes
backendDOMNodeId Not Found
- Only valid within the same browser session
- Always use
take_snapshotandget_react_component_from_backend_node_idin the same MCP session
Development & Publishing
Build
npm run build
Test Locally
# Test the built package
npm start
# Or with target URL
TARGET_URL=http://localhost:3000 npm start
Publish to npm
-
Update version in
package.json:{ "version": "0.2.0" } -
Build and publish:
npm run build npm publish -
Verify publication:
npm info react-context-mcp
Push to GitHub
# Commit all changes
git add .
git commit -m "feat: your feature description"
# Tag the version
git tag v0.2.0
# Push with tags
git push origin main --tags
License
Apache-2.0
Links
- npm Package: https://www.npmjs.com/package/react-context-mcp
- GitHub Repository: https://github.com/uxfreak/react-context-mcp
- Issues: https://github.com/uxfreak/react-context-mcp/issues
- Model Context Protocol: https://modelcontextprotocol.io
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。