The Agent Skills Directory

[COMMAND_EXECUTION]: The skill interacts with Android devices by executing adb (Android Debug Bridge) commands. The implementation in scripts/common.py uses subprocess.run with list-based arguments, which is a secure pattern that prevents shell injection on the host system. Tools like app_launcher.py and navigator.py leverage this to manage packages and perform UI interactions.
[SAFE]: The skill includes an example workflow in scripts/run_examples.sh that utilizes the Gradle wrapper (gradlew) to build a sample application. This is standard practice for Android development and is confined to the local examples/ directory.
[SAFE]: The skill processes UI hierarchy data through scripts/screen_mapper.py. While the UI content of third-party apps constitutes an external data source that an agent would process, the skill is designed as an automation tool where such inspection is the primary and intended purpose. No suspicious data exfiltration or credential harvesting patterns were detected.

android-emulator-automation