Quash MCP - AI-Powered Mobile Automation

A Model Context Protocol (MCP) server for mobile automation testing with Quash. Control Android devices and run automated tests from any MCP-compatible host using natural language.

Features

• 🤖 AI-Powered Automation: Control your Android device using plain English
• 📱 Device Connection: Works with emulators and physical devices
• ⚙️ Flexible Configuration: Customize AI model, temperature, vision, reasoning, and more
• 🔄 Real-Time Execution: Live progress streaming during task execution
• 🎯 Suite Execution: Run multiple tasks in sequence with retry logic
• 📊 Usage Tracking: Monitor API costs and token usage
• 🔐 Secure: API key authentication via Quash platform

Installation

pip install quash-mcp

All dependencies (including ADB tools and device connectivity) are automatically installed. AI execution happens on the Quash backend, keeping the client lightweight and proprietary logic protected.

Quick Start

1. Get Your API Key

1. Visit quashbugs.com/mcp (or your deployment URL)
2. Sign in with Google
3. Go to Dashboard → API Keys
4. Create a new API key

2. Add to Your MCP Host

Quash MCP works with any MCP-compatible host. Configure it to use python3 -m quash_mcp which works across all Python environments:

Manual Configuration (Recommended)

Add to your MCP host's config file:

Config file locations:

• Claude Desktop (macOS): ~/Library/Application Support/Claude/claude_desktop_config.json
• Claude Desktop (Linux): ~/.config/claude/claude_desktop_config.json
• Claude Desktop (Windows): %APPDATA%\Claude\claude_desktop_config.json
• Claude Code: ~/.claude.json (project-specific under projects.<path>.mcpServers)

{
  "mcpServers": {
    "quash": {
      "command": "python3",
      "args": ["-m", "quash_mcp"]
    }
  }
}

Why python3 -m quash_mcp?

• Works in any Python environment (conda, venv, system)
• No PATH configuration needed
• Uses whichever Python has quash-mcp installed

CLI Configuration (If Supported by Host)

Some MCP hosts might provide a command-line interface to add servers.

Examples:

• Claude Code:

  claude mcp add quash quash-mcp
  

• Gemini CLI:

  gemini mcp add quash quash-mcp
  

Alternative: Direct Command (if in PATH)

If quash-mcp is in your PATH:

{
  "mcpServers": {
    "quash": {
      "command": "quash-mcp"
    }
  }
}

Then restart your MCP host.

3. Start Automating

Ask your AI assistant (via your MCP host):

"Setup Quash and connect to my Android device"
"Configure with my API key: mhg_xxxx..."
"Execute task: Open Settings and enable WiFi"

Available Tools

1. build

Setup and verify all dependencies.

Example:

Can you run the build tool to setup my system for Quash?

2. connect

Connect to an Android device or emulator.

Parameters:

• device_serial (optional): Device serial number

Example:

Connect to my Android device

3. configure

Configure agent execution parameters.

Parameters:

• quash_api_key: Your Quash API key from the web portal
• model: LLM model (e.g., "anthropic/claude-sonnet-4", "openai/gpt-4o")
• temperature: 0-2 (default: 0.2)
• max_steps: Maximum execution steps (default: 15)
• vision: Enable screenshots (default: false)
• reasoning: Enable multi-step planning (default: false)
• reflection: Enable self-improvement (default: false)
• debug: Verbose logging (default: false)

Example:

Configure Quash with my API key mhg_xxx, use Claude Sonnet 4, and enable vision

4. execute

Run an automation task on the device.

Parameters:

• task: Natural language task description

Example:

Execute task: Open Settings and navigate to WiFi settings

5. runsuite

Execute multiple tasks in sequence with retry logic.

Parameters:

• suite_name: Name of the test suite
• tasks: Array of tasks with retry and failure handling options

Example:

Run a test suite with these tasks: [
  {"prompt": "Open Settings", "type": "setup"},
  {"prompt": "Enable WiFi", "type": "test", "retries": 2},
  {"prompt": "Close Settings", "type": "teardown"}
]

6. usage

View API usage statistics and costs.

Example:

Show me my Quash usage statistics

Complete Workflow Example

User: "Setup Quash on my machine"
→ Runs build tool
→ Returns: All dependencies installed ✓

User: "Connect to my Android emulator"
→ Runs connect tool
→ Returns: Connected to emulator-5554 ✓

User: "Configure to use Claude Sonnet 4 with vision and my API key is mhg_xxx..."
→ Runs configure tool
→ Returns: Configuration set ✓

User: "Execute task: Open Instagram and go to my profile"
→ Runs execute tool with live streaming
→ Returns: Task completed ✓

User: "Show me my usage statistics"
→ Runs usage tool
→ Returns: Total cost: $0.15, 10 executions ✓

Requirements

• Python 3.11+ - Required for the MCP server
• Android Device - Emulator or physical device with USB debugging enabled
• Quash API Key - Get from quashbugs.com/mcp

Dependencies automatically installed:

• Android Debug Bridge (ADB) - via adbutils
• Quash Portal APK - via apkutils
• MCP protocol support - via mcp
• HTTP client - via httpx

Architecture

v0.2.0 uses a client-server architecture:

• Client (quash-mcp): Lightweight MCP server handling device connections and API calls
• Server (Quash backend): Proprietary AI execution engine (LLMs, agents, pricing logic)

quash-mcp/
├── quash_mcp/
│   ├── __main__.py          # Module entry point for python -m
│   ├── server.py            # Main MCP server entry point
│   ├── backend_client.py    # API communication with Quash backend
│   ├── state.py             # Session state management
│   └── tools/
│       ├── build.py         # Dependency checker and installer
│       ├── connect.py       # Device connectivity
│       ├── configure.py     # Agent configuration
│       ├── execute.py       # Task execution (calls backend API)
│       ├── runsuite.py      # Suite execution (calls backend API)
│       └── usage.py         # Usage statistics (from backend)
└── pyproject.toml

Troubleshooting

"No devices found"

• Start Android emulator via Android Studio > AVD Manager
• Connect physical device with USB debugging enabled
• For WiFi debugging: adb tcpip 5555 && adb connect <device-ip>:5555

"Portal not ready"

• The connect tool automatically installs the Portal APK
• If it fails, manually enable the Quash Portal accessibility service in Settings > Accessibility

"Invalid API key"

• Make sure you've run configure with a valid API key from quashbugs.com
• API keys start with mhg_ prefix
• Check your API key hasn't been revoked in the web portal

License

MIT