Skills
Navigate
skills/bitwarden/android/interacting-with-android-device/Gen Agent Trust Hub

interacting-with-android-device

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill uses the capture tool to retrieve the UI hierarchy (XML) from the Android device. This data contains arbitrary text from third-party applications which may contain malicious instructions designed to influence the agent's behavior.
  • Ingestion points: view.xml UI hierarchy dump generated by the capture tool (SKILL.md).
  • Boundary markers: None identified; UI text is processed directly to find elements.
  • Capability inventory: Shell access via Bash(adb:*), file system access via Read/Glob, and specific interaction tools like tap_at and input_text.
  • Sanitization: No evidence of filtering or sanitizing UI text before processing.
  • [COMMAND_EXECUTION]: The skill relies on extensive shell access through Bash(adb:*) and the ./gradlew build system. While these are standard for Android development and interaction, they grant the agent significant control over the connected device and the local build environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 03:53 PM
Security Audit — agent-trust-hub — interacting-with-android-device