Mobile Easy Use

What agent-easy-use/mobile-easy-use does

The agent-easy-use/mobile-easy-use MCP server connects an AI agent to a running Android or iOS application through a Mobile Easy Use runtime. It is intended for runtime exploration rather than source inspection alone. Depending on the platform and app implementation, an agent can examine native objects, business state, caches, configuration, UI properties, method calls, return values, call stacks, execution threads, timing, memory changes, logs, and screenshots.

The repository also supports enhanced mobile testing. A test can combine UI assertions, screenshots, and native business-state checks. Runtime fields or method results can be changed temporarily to exercise failure paths, empty data, or configuration branches, then restored for later runs.

How it works

The app must first be integrated with the Mobile Easy Use runtime. Android integration adds an AAR to the debug variant, while iOS integration embeds the bridge and runtime in an internal debug configuration. After the app is built and installed, MCP connects to the target process through the device, emulator, or simulator environment.

The agent can use platform-specific Probe Skills to generate and execute an investigation from source code, existing Presets, and a natural-language question. Probe actions can correlate method execution, state changes, and visible UI behavior. Separate Test Skills generate JavaScript test files and run test directories for scenarios that need repeatable checks.

Setup and configuration

Node.js 20 or newer is required. Android development needs ADB and an available device or emulator. iOS development needs macOS, Xcode, and the Xcode command-line tools; physical iOS devices additionally require iproxy and development signing for the XCTest Runner used by input actions.

Install the repository Skills from the target app project root with npx skills add agent-easy-use/mobile-easy-use --skill '*'. Skills and MCP are separate: Skills provide integration, exploration, testing, Presets, and related script workflows, while MCP supplies the connection and execution tools.

Add the MCP package to the client configuration with npx, using the exact compatible version returned by the integration workflow. Set the MCP working directory to the target app project root so it can resolve project Presets. The agent and MCP process also need shared local filesystem access to SDK declarations, probes, and test files. Build and install the integrated debug app before connecting.

Tools and capabilities

The agent-easy-use/mobile-easy-use MCP server exposes connection and SDK-discovery functionality, including tools such as connect and get_sdk_declarations. The surrounding workflows provide platform-specific commands such as to-android-probe, to-ios-probe, to-android-test-script, and to-ios-test-script.

Supported work includes:

• Inspecting native objects, fields, getters, business state, and caches.
• Tracing calls with arguments, return values, stacks, threads, timing, and memory observations.
• Temporarily overriding fields, configuration, or method returns.
• Performing UI actions and checking visibility, properties, screenshots, and control state.
• Generating and running tests that combine UI, image, and internal-state assertions.

Limitations and notes

The available observations depend on the platform and the application implementation. Integration should be limited to internal or debug builds. The MCP version must match the runtime release; new connections check compatibility and can suggest a version adjustment. iOS physical-device input actions have additional signing and proxy requirements. A running, integrated app is required before Probe or runtime operations can collect evidence.