droidmind

Source on GitHub

Category: Testing & Debugging Tools
Tags: mcp, android, automation, ai-integration

Description

DroidMind is a tool that allows AI assistants to control, debug, and analyze Android devices using the Model Context Protocol (MCP) over ADB in a secure, structured way. It enables natural language device interaction, suitable for agentic coding workflows where assistants can build and debug directly on real devices.

Features

• Device Control: Connect to Android devices over USB or TCP/IP, run shell commands, reboot devices.
• System Analysis: Inspect device properties, view hardware info, analyze system logs.
• File System Access: Browse directories, read/write files, manage files on devices.
• Visual Diagnostics: Capture device screenshots for analysis and debugging.
• App Management: Install, uninstall, start, stop, and clear app data for applications on devices.
• Multi-Device Support: Control and switch between multiple connected devices.
• UI Automation: Automate UI interactions (taps, swipes, text input, key presses).
• App Inspection: View app manifests, shared preferences, and app-specific logs.
• Security Framework: Command validation, risk assessment, input sanitization, protected system paths, comprehensive command logging, suspicious pattern detection, and ADB command security.
• MCP Integration: Works with AI assistants such as Claude, Cursor, Cline, and others via the MCP protocol.
• Resource Endpoints: Extensive set of MCP resources for device listing, property fetching, log retrieval, file system operations, app data access, etc.
• CLI Tools: Tools for device listing, property fetching, log analysis, file operations, app management, device rebooting, screenshot capturing, bug report generation, heap dumps, and more.
• Example AI Queries: Provides sample queries for AI assistants to perform device operations, retrieve logs, analyze device status, manage apps, and automate UI.
• Development Support: Uses UV for dependency management, supports testing, linting, and type checking.

Security Features

• Command Validation: All shell commands are checked against an allowlist.
• Risk Assessment: Commands categorized by risk level (SAFE to CRITICAL).
• Command Sanitization: Prevents command injection.
• Protected Paths: Critical system paths are protected from modification.
• Comprehensive Logging: All commands are logged with risk levels.
• Suspicious Pattern Detection: Blocks potentially dangerous commands.
• ADB Command Security: Special handling and async validation for ADB commands.
• User Warnings: High-risk commands display warnings, critical ones are blocked unless explicitly overridden.

Requirements

• Python 3.13+
• Android device with USB debugging enabled
• ADB installed and in PATH
• UV package manager (recommended)
• For network: ADB over TCP/IP enabled on device

Installation & Usage

• Clone the repo and install dependencies with UV.
• Run the server (droidmind --transport sse) to connect AI assistants via MCP.

License

Apache-2.0

Pricing

No pricing information is provided; DroidMind appears to be open source and free to use under the Apache-2.0 license.