The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the Python subprocess module to execute system binaries including adb, gradlew, and the Android emulator. Specifically, scripts/common.py, scripts/build_and_test.py, scripts/log_monitor.py, and scripts/emulator_manage.py use subprocess.run and subprocess.Popen to perform lifecycle tasks. While these are necessary for the skill's primary purpose, they represent a significant capability surface.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it ingests and processes untrusted data from the Android device/emulator environment.
Ingestion points: scripts/screen_mapper.py dumps the UI hierarchy via uiautomator, and scripts/log_monitor.py streams system logs via adb logcat. Both scripts parse this external data and present it to the agent.
Boundary markers: There are no explicit boundary markers or 'ignore' instructions implemented in the output of the data-parsing scripts to distinguish between tool instructions and ingested UI/log content.
Capability inventory: The skill allows the agent to launch/terminate applications, input text, perform gestures, and trigger Gradle builds (scripts/app_launcher.py, scripts/navigator.py, scripts/build_and_test.py).
Sanitization: The skill employs some defensive measures, such as using shlex.quote in scripts/keyboard.py and scripts/navigator.py to sanitize text inputs before passing them to the shell, which reduces the risk of direct command injection.

android-emulator-skill