@shaykec/appxray-mcp-server
v0.2.1
Published
MCP server providing inside-out app intelligence tools for AI coding agents
Maintainers
Readme
@shaykec/appxray-mcp-server
MCP server that gives AI coding agents inside-out access to running applications.
Instant Setup — Paste This to Your AI Agent
Copy the following prompt into Cursor, Claude Code, or any MCP-capable AI agent chat:
Integrate appxray into my project. Install the appxray MCP server and the right SDK for my app's platform. Here's how:
- Add to my MCP config (
.cursor/mcp.jsonor equivalent):{ "mcpServers": { "appxray": { "command": "npx", "args": ["-y", "@shaykec/appxray-mcp-server"], "env": { "APPXRAY_AUTO_CONNECT": "true" } } } }- Detect my app's platform and install the matching SDK:
- Web (Vite/React/Next.js/Vue):
npm install @shaykec/appxray-web— addappxrayVite()to vite.config plugins, orimport "@shaykec/appxray-web/auto"in the entry file- Node.js (Express/Fastify):
npm install @shaykec/appxray-node— addapp.use(appxrayMiddleware())andAppXrayNode.init({ app, appName: "MyApp" })- React Native:
npm install @shaykec/appxray-react-native— callappxray.init()in a useEffect- iOS/macOS: Add Swift Package
https://github.com/shayke-cohen/appxray.git— callAppXray.shared.start()in init- Android:
debugImplementation("com.appxray:sdk-android:0.1.0")— auto-inits in debug- Flutter:
flutter pub add --dev appxray— wrap withAppXrayApp(child: MyApp())- Verify the build succeeds and the app is discoverable.
The agent will auto-detect your platform, install everything, and verify the connection.
What It Does
AI coding agents are blind to what's happening inside your running app. They rely on screenshots and guesswork. appxray connects them directly to your app's internals — component tree, live state, network traffic, storage, navigation, and more.
This package is the MCP server. It discovers appxray-enabled apps on your machine and exposes 14 aggregate tools (55+ operations) via the Model Context Protocol.
Setup
Add to your AI agent's MCP config (Cursor, Claude Desktop, Claude Code, etc.):
{
"mcpServers": {
"appxray": {
"command": "npx",
"args": ["-y", "@shaykec/appxray-mcp-server"],
"env": { "APPXRAY_AUTO_CONNECT": "true" }
}
}
}Or install globally:
npm install -g @shaykec/appxray-mcp-serverTools
| Tool | Operations |
|------|-----------|
| session | discover, connect, disconnect, info, list, switch |
| inspect | tree, state, network, storage, routes, errors, schema, metrics, bindings, logs, accessibility, profile, timeline, renders |
| act | setState, trigger, input, navigate, emit |
| interact | find, tap, doubleTap, longPress, type, swipe, drag, screenshot, waitFor, fillForm, record |
| batch | Execute multiple operations in one call |
| mock | install, remove, clear, list, override, removeOverride, listOverrides |
| config | get, set, list, reset (feature flags, env vars) |
| timetravel | checkpoint, restore, history |
| chaos | start, stop, list |
| advanced | eval, coverage, subscribe, events, storageWrite, storageClear |
| diagnose | One-call health check: errors, network, performance, a11y |
| diff | snapshot, compare, list |
| suggest | Root cause analysis with ranked hypotheses |
| trace | render, stateChange, dataFlow, whyEmpty |
Every tool response includes a live snapshot with screenshot, component tree, console logs, errors, network activity, and state summary.
SDK Required
Your app needs an appxray SDK embedded to be discoverable. Available SDKs:
| Platform | Package |
|----------|---------|
| Web (React/Next.js/Vue) | @shaykec/appxray-web |
| Node.js (Express/Fastify) | @shaykec/appxray-node |
| React Native | @shaykec/appxray-react-native |
| iOS/macOS | Swift Package (source) |
| Android | Kotlin library (source) |
| Flutter | Dart package (source) |
Configuration
| Environment Variable | Default | Description |
|---------------------|---------|-------------|
| APPXRAY_AUTO_CONNECT | false | Auto-connect to the first discovered app |
| APPXRAY_PORT_MIN | 19400 | Port range start for discovery |
| APPXRAY_PORT_MAX | 19499 | Port range end for discovery |
| APPXRAY_AUTH_TOKEN | — | Pre-shared auth token for app connections |
CLI: Test Runner
The package also ships appxray-run, a YAML-based test runner:
npx @shaykec/appxray-mcp-server appxray-run tests/login-flow.yamlRelated Packages
@shaykec/appxray-protocol— Shared protocol types@shaykec/appxray-web— Web SDK@shaykec/appxray-node— Node.js SDK@shaykec/appxray-react-native— React Native SDK
License
MIT
