The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill executes raw JavaScript strings via the eval_script tool and loads local modules via the call_function tool based on the source and scriptPath arguments.
[COMMAND_EXECUTION]: The skill uses the Android Debug Bridge (ADB) to perform actions like port forwarding (adb forward) and starting application activities (am start -W) on connected devices.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it accepts and executes code or file paths provided by the agent. If the agent retrieves these values from an untrusted source, it could lead to the execution of malicious code on the Android target.
Ingestion points: The source, scriptPath, and functionName parameters are dynamically populated by the agent.
Boundary markers: There are no explicit boundary markers or instructions provided to the agent to ignore potentially malicious embedded code within the data it processes.
Capability inventory: The skill possesses the ability to execute arbitrary code within an app runtime, load local files as modules, and interact with the host system via ADB commands.
Sanitization: No sanitization, validation, or integrity checks are performed on the source code or the scriptPath before execution.

to-android-run