adb-controller

The adb-controller skill provides an interface for interacting with Android devices through the Android Debug Bridge (ADB). It is designed for developers, QA testers, and automation enthusiasts who need to bridge the gap between AI agent reasoning and mobile device execution. By utilizing the script-based architecture, the agent can translate natural language commands into precise shell actions, enabling hands-free operation of physical or emulated devices.

• Execute shell commands such as input tap, swipe, keyevent, and text injection.

• Automated screenshot retrieval for every action, allowing the agent to perform visual verification.

• Seamless integration with openclaw.json for per-device or per-server connection configuration.

• Supports standard ADB protocol, ensuring compatibility with any connected device or network-accessible ADB server.

• Feedback-loop enabled design where visual output is sent directly back to the chat interface.

• Users must configure their adbServer (serial number or IP:port) within the openclaw.json configuration file.

• The skill relies on the exec tool to run python scripts; ensure the environment has proper ADB permissions and that the device is authorized via USB debugging.

• Best suited for tasks involving UI navigation, automated testing, or performing repetitive gestures on Android.

• Always ensure the device is screen-on and unlocked, as ADB commands may fail if the device requires credential entry or suffers from system-level input restrictions.

• Visual output is essential for the agent; ensure the agent uses the message tool with the media parameter to correctly display the resulting screenshots.