device-interaction

Pass

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill provides standard developer utilities for iOS testing and interaction. All defined tools, such as session management and event synthesis, align with legitimate software development and verification workflows.\n- [PROMPT_INJECTION]: The subagent reads untrusted UI data (hierarchy and text) from the application under test, which presents an indirect prompt injection surface. Content within the application could attempt to influence the agent's logic.\n
  • Ingestion points: UI hierarchy and device screenshots processed via DeviceEventSynthesize.\n
  • Boundary markers: None identified; the agent is instructed to directly interpret UI content.\n
  • Capability inventory: touch simulation, keyboard input synthesis (kbd), hardware button control, and environment variable configuration.\n
  • Sanitization: The instructions do not specify any sanitization or validation for the text elements extracted from the device UI.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 22, 2026, 09:53 AM