AI App Bridge provides a runtime interface for AI agents to interact with running Android and Flutter mobile applications, enabling autonomous iteration and debugging. It allows AI to inspect UI, operate elements, collect runtime data, and verify results, bridging the gap between AI-generated code and verified, delivered changes. This tool is designed for AI developers and researchers aiming to enhance AI-driven app development and testing workflows.

How It Works

This project acts as a runtime bridge, equipping AI agents with "eyes and hands" into live mobile applications. It facilitates an observe-act-verify-iterate loop by enabling agents to inspect the current screen, analyze native Android View trees, Flutter Widget trees, and WebView DOM structures. Agents can perform UI operations, evaluate JavaScript within WebViews, and capture detailed runtime evidence such as logs, network requests, and state changes, providing real-time feedback for AI decision-making.

Quick Start & Requirements

• Android: Add com.github.mobileAiDev.ai-app-bridge:ai-app-bridge-android:0.2.8 to debugImplementation in app/build.gradle.kts. The runtime SDK auto-initializes in debuggable apps (minSdk 19+). Optional OkHttp capture requires the io.github.mobileaidev.aiappbridge.android Gradle plugin.
• Flutter: Add ai_app_bridge_flutter: ^0.2.3 to pubspec.yaml and initialize AiAppBridge.instance.initialize() once. Register an H5 adapter for WebView DOM support.
• AI Agent Connection: Load the skills/ai-app-bridge-use/SKILL.md guide into your AI agent. Install the MCP server via npm install -g @mobileaidev/ai-app-bridge and configure it in your AI agent's MCP settings.
• Prerequisites: Android debuggable builds, Flutter projects, Node.js for MCP server.

Highlighted Details

• Inspects Android View trees, Flutter Widget trees, and WebView DOM.
• Supports native UI operations, with ADB/UIAutomator fallbacks.
• Enables WebView JavaScript evaluation and debug capabilities.
• Captures logs, network requests, state records, and event buffers.
• Includes a debug Gradle plugin for OkHttp auto-capture.
• Provides a Node CLI/MCP stdio server for AI tool integration.

Maintenance & Community

No specific details on maintainers, community channels (Discord/Slack), or roadmap were found in the provided README.

Licensing & Compatibility

Licensed under the Apache License 2.0. Modified versions must retain original notices and clearly state modifications.

Limitations & Caveats

AI App Bridge is strictly intended for debug builds only. Deployment in production or release builds requires a thorough, deliberate security review due to its runtime access capabilities.