droidmind

DroidMind is an MCP server that bridges AI assistants with Android devices via ADB, enabling controlled device management, system analysis, file operations, app control, and UI automation through the Model Context Protocol. The server exposes its capabilities so MCP-compatible clients (such as Cursor, Claude, or other MCP-enabled IDEs/agents) can request actions like listing connected devices, reading logs, pushing/pulling files, installing apps, or performing UI interactions. When run via uvx (the recommended setup for IDE integrations), DroidMind can be launched directly from a GitHub source without manual installation, making it convenient for rapid experimentation and integration into your AI workflows. For local use, you can also run DroidMind with different transports (stdio for direct terminal or SSE for web/UIs) depending on your workflow.

To use DroidMind through MCP, configure your IDE or agent to load the server from the repository, then issue commands such as: listing devices, fetching logcat or crash logs, capturing bug reports, pulling or pushing files, installing or uninstalling apps, and performing shell commands on the device. The UI automation and ADB-backed capabilities provide a rich set of interactions suitable for debugging, testing, and automated workflows within an AI-assisted development loop.

Prerequisites:

• Python 3.13
• uv (Python package manager)
• Android device with USB debugging enabled
• ADB installed and available in your system PATH

Installation (quickest, via uvx):

1. Ensure you have Python 3.13 and uv installed.
2. Install and/or configure uvx as your MCP runner.
3. Use the MCP configuration to run DroidMind directly from its GitHub repository (no manual clone needed):

# If you don't have uvx installed yet, install it (example, via pipx or pip):
pipx install uvx  # or: python -m pip install uvx

# Run DroidMind via uvx using the MCP config approach (integrates with IDEs that support MCP):
uvx --from git+https://github.com/hyperb1iss/droidmind droidmind --transport stdio

Manual installation (from source):

1. Clone the repo
2. Create and activate a virtual environment
3. Install dependencies and run the server with the desired transport

git clone https://github.com/hyperb1iss/droidmind.git
cd droidmind
python -m venv .venv
source .venv/bin/activate  # on Unix/macOS
# Windows: .venv\Scripts\activate
pip install -r requirements.txt
# Run with stdio transport
droidmind --transport stdio
# Or run with SSE for web/UI clients
droidmind --transport sse

Docker (optional): Refer to the Docker guide in the repository for building/running in a container. Ensure ADB access from the container if you plan to interact with physical devices.

Tips and notes:

• The recommended MCP integration using uvx allows IDEs to launch DroidMind on demand without manual installation.
• If you plan to use SSE transport, ensure your environment can expose the SSE endpoint (usually sse://localhost:4256/sse) and that your client supports SSE connections.
• Typical environment variables are not required for basic operation, but advanced deployments may expose ADB paths, device filters, or security-related flags in your own integration.
• When running on a headless server, consider using a stable transport (stdio for local IDEs, SSE for web/UIs) and ensure the host machine has access to connected Android devices via USB or a compatible network/ADB setup.
• For security, DroidMind includes command validation and risk assessment; exercise caution with high-risk operations and review the Security Considerations in the documentation.