Claude Mobile provides an MCP server for automating mobile (Android, iOS Simulator), desktop (macOS), and browser (CDP) applications through natural language interfaces like Claude. It aims to unify automation across diverse platforms with a token-optimized toolset, benefiting quality engineers and developers by simplifying complex testing and interaction workflows.

How It Works

The project functions as a Meta-Command Protocol (MCP) server, translating LLM-generated natural language commands into executable actions on target platforms. It employs a core set of 8 meta-tools and 3 optional, dynamically loaded modules, significantly reducing token overhead. This architecture allows a consistent interaction model across Android (via ADB), iOS Simulators (via simctl and WebDriverAgent), macOS desktop applications, and web browsers (via Chrome DevTools Protocol).

Quick Start & Requirements

• Primary Install: macOS: brew install claude-in-mobile. Any client: npx add-mcp claude-in-mobile -y.
• Prerequisites: Android requires ADB. iOS Simulator requires Xcode. Desktop automation requires macOS and JDK 17+. Browser automation requires Chrome/Chromium. The claude-in-mobile doctor command verifies all dependencies.
• Setup: WebDriverAgent auto-builds on first use (~2 min).
• Links: Repository: https://github.com/AlexGladkov/claude-in-mobile. Desktop API: docs/SPEC_DESKTOP.md.

Highlighted Details

• Unified API: A consistent set of 8 meta-tools (device, input, screen, ui, app, system, flow_batch, flow_run) operates across all supported platforms.
• Token Optimization: Utilizes 8 meta-tools and 3 optional modules, achieving an ~85% token reduction compared to larger toolsets.
• Advanced Quality Engineering: Features include accessibility auditing, visual regression testing, test recording, multi-device synchronization, autonomous app exploration (App Autopilot), and real-time performance monitoring.
• Security: Hardened against shell injection, URL validation, and path traversal.
• Parallel Execution: Supports running actions on multiple devices simultaneously.

Maintenance & Community

No specific details regarding maintainers, community channels (e.g., Discord, Slack), or roadmap are provided in the README.

Licensing & Compatibility

The project is released under the MIT License, which is permissive for commercial use and integration into closed-source projects.

Limitations & Caveats

iOS automation is currently restricted to simulators, with no support for physical devices. Desktop automation is limited to macOS, although Windows and Linux support are planned.