QApilot MCP for Android
Android app testing inside your coding agent: Claude, Cursor, or Codex drive a real device via MCP.
QApilot MCP is an MCP server that lets Claude Desktop, Cursor, or OpenAI Codex drive a real Android device or emulator. You describe test flows in plain English; the agent builds a structured plan, executes it over Appium/UIAutomator2, streams a live preview, and can accept the passing steps into QApilot as a saved test case. Saved cases replay one-by-one, in batch, or from an Excel sheet, with reports written as output.json, report.yaml, and a Gherkin feature file on pass.
What is QApilot MCP for Android?
QApilot MCP is a Model Context Protocol server that lets AI coding agents such as Claude Desktop, Cursor, and OpenAI Codex test Android apps on a real device or emulator. The user describes test flows in plain English; the agent builds a structured plan, executes the steps through Appium and UIAutomator2, streams a live preview of the device screen, and can push the passing steps to the QApilot platform as a saved, reusable test case.
Key features
- Drives local Android devices and emulators from any MCP-compatible AI client (config blocks provided for Claude Desktop, Cursor, and OpenAI Codex)
- Plain-English test flows become structured test plans executed on the device in real time
- 25+ MCP tools covering signup/login, session, device selection, planning, execution, step acceptance, and reporting
- Live preview URL streams the device screen while steps run
- Accepted happy-path steps save as QApilot test cases replayable by ID, in batch, or from an Excel (.xlsx) sheet
- Reports write output.json and report.yaml (plus a Gherkin .feature file on pass) to a dated local folder
- Per-app XPath/skill caching and a usage tool that reports token and API consumption per session
- Pinned stack: Appium 2.19.0 with the UIAutomator2 4.2.6 driver, Node.js 18+, Java JDK 11+, Android SDK
Who it's for
- Mobile developers using Claude Code, Cursor, or Codex who want the agent to verify Android changes on a device before merge
- QA teams converting recorded agent sessions into saved, replayable regression cases
- Teams whose manual test cases already live in Excel and want an AI agent to execute them
- Automating Android regression flows without writing Appium scripts by hand
When not to use it
It is Android-only — the MCP guide covers no iOS support — and it requires a local Appium/Android SDK stack plus a QApilot account. Teams needing transparent self-serve pricing should note the guide publishes none; the platform is sold through demos.
FAQ
What does QApilot MCP do?
An MCP server that gives AI coding agents (Claude Desktop, Cursor, Codex) hands on Android: they launch your app on a real device or emulator, run plain-English test flows, and generate reports.
Does QApilot MCP need Appium?
Yes. The guide pins Appium 2.19.0 and the UIAutomator2 4.2.6 driver, with Node 18+, Java JDK 11+, and the Android SDK installed; Appium must be running during test sessions.
Are the tests reusable or one-off?
Reusable — after a passed run you accept the happy-path steps as a saved QApilot test case, then replay by test-case ID, in batch, or from an Excel file.
Does QApilot MCP work on iOS?
The MCP launch is Android-only, driving devices or emulators via ADB and Appium; the MCP guide does not cover iOS.
How much does QApilot MCP cost?
Pricing is not published on the MCP guide. The CLI installs free via npm but requires a QApilot account, and the platform is sold demo-led; pricing not verified.
Share this launch
Embed this badge
<a href="https://orangebot.ai/product/qapilot-mcp" target="_blank" rel="noopener noreferrer"> <img src="https://orangebot.ai/api/badge/qapilot-mcp.svg" alt="Featured on OrangeBot" width="200" height="54" /> </a>