android-ui-journey-testing
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: This skill follows instructions provided in an external XML test suite to execute actions on an Android device. This creates a vulnerability surface where a malicious or compromised XML file could contain instructions that influence the agent to perform unauthorized actions on the device.
- Ingestion points: External XML journey specifications defined in the 'Parse the Journey Specification' step.
- Boundary markers: The skill does not explicitly provide the agent with delimiters or instructions to distinguish between the framework's operational instructions and the potentially untrusted natural language content within the XML
<action>tags. - Capability inventory: The skill possesses significant capabilities via ADB (Android Debug Bridge), including tapping, swiping, and typing text on the connected device or emulator.
- Sanitization: There is no evidence of validation or sanitization for strings being passed directly to the 'adb shell input text' command, which could potentially allowed for shell command injection on the target Android shell if special characters are present in the 'action' input.
Audit Metadata