ux-walker-native

Pass

Audited by Gen Agent Trust Hub on Oct 29, 2026

Risk Level: SAFEDYNAMIC_EXECUTIONPRIVILEGE_ESCALATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill includes a local Swift source file (driver/nativeui.swift) which is compiled into an executable binary using swiftc during the installation phase (driver/install.sh). This is used to create the driver that interacts with the macOS accessibility subsystem.
  • [PRIVILEGE_ESCALATION]: To drive native applications, the skill requires macOS Accessibility permissions. It instructs the user to manually enable these permissions for their terminal application within System Settings. This provides the agent with the ability to read UI content and synthesize keyboard/mouse input across the operating system.
  • [COMMAND_EXECUTION]: The skill executes various shell commands to manage its lifecycle and perform audits, including building the driver, running the compiled nativeui binary, capturing screenshots via screencapture, and inspecting system logs using log show.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the accessibility tree of target applications (such as labels, values, and hints). This data is interpolated into sub-agent prompts to guide the UX walk. There is a risk that a malicious or compromised application could place adversarial instructions in its UI elements to influence the agent's logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 29, 2026, 11:41 AM