argent-test-ui-flow

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill references the use of the adb (Android Debug Bridge) command to perform port reversing (adb reverse tcp:8081 tcp:8081). This is a legitimate and common procedure in React Native development to bridge the mobile device to the local development server (Metro).\n- [PROMPT_INJECTION]: The skill architecture involves interpreting data from the mobile device's screen (screenshots and accessibility trees), which constitutes a potential surface for indirect prompt injection if the tested application displays malicious instructions. However, this is inherent to UI automation functionality.\n
  • Ingestion points: Element trees and screenshots from the describe, debugger-component-tree, and screenshot tools.\n
  • Boundary markers: No delimiters or ignore-instructions warnings are specified for the incoming UI data.\n
  • Capability inventory: The agent can perform UI gestures (tap, swipe, paste) and execute shell commands (adb).\n
  • Sanitization: No sanitization of the UI element content is mentioned.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 10:26 PM