device-interaction
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill serves as a technical adapter for device interaction tools and does not include executable scripts or unauthorized network operations. References to official Apple developer domains target a well-known and trusted service provider.
- [INDIRECT_PROMPT_INJECTION]: The skill identifies an attack surface where the agent processes untrusted UI data from applications under test.
- Ingestion points: UI hierarchy files and screenshots processed by the agent in apple/guide.md.
- Boundary markers: Explicit workflow instructions in SKILL.md mandate adherence to user scope and technical-only interpretation of provided guidance.
- Capability inventory: Includes application installation, process management, and user event synthesis (touch/keyboard) tools described in apple/guide.md.
- Sanitization: No automated filtering or sanitization of external UI text labels is specified.
Audit Metadata