ios-device-toolkit

Warn

Audited by Gen Agent Trust Hub on Jun 14, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill requires the execution of sudo pymobiledevice3 remote tunneld to support developer services on iOS 17+ devices, granting the tool root privileges on the host system.
  • [DATA_EXFILTRATION]: The skill includes numerous capabilities to access and extract sensitive data from connected devices. Evidence includes:
  • pymobiledevice3 apps pull: Extracting files from application sandboxes to the local machine.
  • pymobiledevice3 syslog live: Streaming live system logs which may contain private information.
  • pymobiledevice3 pcap: Capturing device network traffic for analysis.
  • pymobiledevice3 developer dvt screenshot: Capturing screen images of the device.
  • [EXTERNAL_DOWNLOADS]: The skill setup involves installing the pymobiledevice3 Python package from public registries using standard tools like uv or pipx.
  • [PROMPT_INJECTION]: The skill exposes the agent to untrusted data from the iOS device, creating a surface for indirect prompt injection.
  • Ingestion points: Data is ingested via syslog live, crash ls, pcap, and apps pull commands.
  • Boundary markers: No delimiters or protective instructions are provided to the agent regarding data retrieved from the device.
  • Capability inventory: The skill can execute shell commands through the pymobiledevice3 binary and write files to the host filesystem.
  • Sanitization: The retrieved content is processed without evidence of sanitization or validation.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 14, 2026, 09:41 PM
Security Audit — agent-trust-hub — ios-device-toolkit