QApilot MCP for Android connects AI coding clients to local Android devices, executing plain-English tests without manual Appium code.
What the product does and how it is positioned
QApilot MCP for Android is a Model Context Protocol CLI bridge that enables AI assistants such as Claude Desktop and Cursor to drive mobile application testing. It connects directly to local Appium servers and Android Debug Bridge environments to manipulate physical devices and emulators without requiring users to write manual Appium or UIAutomator scripts.
The tool automates test workflows by translating conversational prompts into structured plans, performing manual actions, streaming real-time execution to a browser preview, and saving verified test paths back to QApilot projects. At the conclusion of testing runs, it outputs structured data, Gherkin specifications, and session metrics.
Source-supported ways to use the product
Quality engineers can describe test interactions in natural language, enabling the AI client to construct structured test plans, execute actions on the connected Android device, and capture screenshots.
Teams can run previously accepted test cases sequentially by ID or bulk-execute full test suites loaded directly from an Excel (.xlsx) spreadsheet against connected emulators.
The documented workflow, where available
Verify prerequisites including Node.js 18+, Java JDK 11+, Android SDK, and start Appium 2.19.0 with the pinned uiautomator2 driver.
Add the qapilot-mobile-mcp command and arguments to the MCP settings configuration of Claude Desktop, Cursor, or compatible AI clients.
Log into the QApilot workspace and launch the target application by specifying its Android package identifier.
Direct the AI assistant to perform actions, inspect real-time progress via the live preview URL, and accept successful happy-path steps into QApilot.
Generate session reports, producing JSON, YAML, and Gherkin files while clearing the execution state for subsequent runs.
The CLI operates as an MCP server communicating over standard input and output streams, exposing dedicated tools for device management, plan construction, action dispatch, and state queries. AI clients leverage tools like mobile_launch_app, mobile_submit_plan, and mobile_execute to manage the Appium session life cycle programmatically.
During active execution, state inspection commands allow clients to query execution status and fetch cached application skills and XPaths. When a run concludes, calling the reporting tool writes test artifacts locally to the Downloads directory and resets the active execution context.
Checks to run with your own material and workflow
What was checked and when
Answers based on the source-checked product record
The server can be configured with Claude Desktop, Cursor, OpenAI Codex, or any AI tool compatible with the Model Context Protocol.
The official documentation requires Appium 2.19.0 installed with the uiautomator2 driver version 4.2.6.
No, only happy-path steps are accepted; failures are excluded and must be corrected before saving steps as a test case.
Reports are saved to the user's Downloads directory and contain output.json, report.yaml, and a scenario.feature Gherkin file for passed sessions.
Yes, saved test cases can be executed directly from an Excel (.xlsx) file using the mobile_run_excel_testcases tool.
