The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of system commands to interact with xcrun simctl, idb, and xcodebuild for simulator management and app interaction. Analysis of scripts like app_launcher.py, navigator.py, and simctl_boot.py shows that these calls are implemented securely using argument lists in subprocess.run and subprocess.Popen, which prevents shell command injection.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources during app testing, creating a surface for indirect prompt injection.
Ingestion points: UI accessibility trees fetched via idb ui describe-all and application logs captured via xcrun simctl spawn ... log stream.
Boundary markers: None observed; element labels and log content are interpolated directly into the agent's context.
Capability inventory: Significant local capabilities including simulator creation/deletion (simctl_create.py, simctl_delete.py), permission management (privacy_manager.py), and hardware button control (keyboard.py).
Sanitization: Content is parsed as JSON but not explicitly sanitized for adversarial instructions. This is a characteristic of automation tools rather than a specific malicious intent.
[EXTERNAL_DOWNLOADS]: The scripts/sim_health_check.sh script identifies missing dependencies and provides the user with instructions to install the Pillow library via pip3. It does not execute these commands automatically.
[SAFE]: The skill is a legitimate developer tool suite authored by 'n0an'. All resource interactions (file paths, local configurations, and tool invocations) are consistent with the stated purpose of iOS app testing and simulator automation.

ios-simulator-skill