xcode

Warn

Audited by Gen Agent Trust Hub on Jun 21, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of an external global NPM package "xcode-cli" ("npm install -g xcode-cli") to enable the bridge functionality.
  • [COMMAND_EXECUTION]:
  • The skill suggests using "sudo xcode-select -s" to modify the system-wide path for Xcode command-line tools.
  • It establishes persistence on the host machine by installing a background LaunchAgent service via "xcode-cli-ctl install".
  • It executes powerful toolchain commands including "xcodebuild" and "xcrun simctl" for automated project and simulator management.
  • [DATA_EXFILTRATION]:
  • The skill can monitor and capture system-wide or app-specific logs using "/usr/bin/log stream".
  • It has the capability to capture screenshots and video recordings of the iOS/macOS simulator environment ("xcrun simctl io ... screenshot/recordVideo"), which may expose sensitive application data.
  • [PROMPT_INJECTION]:
  • The skill possesses a surface for Indirect Prompt Injection (Category 8).
  • Ingestion points: The skill ingests untrusted data from project files (".xcodeproj"), source code files ("XcodeRead"), and build/test logs ("GetBuildLog").
  • Capability inventory: The skill has extensive command execution capabilities ("xcodebuild", "simctl") and file system modification tools ("XcodeUpdate", "XcodeWrite").
  • Sanitization/Boundary markers: There is no evidence of output sanitization or the use of boundary markers to prevent malicious instructions embedded in project files or build logs from being interpreted as commands by the agent.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 21, 2026, 12:58 AM
Security Audit — agent-trust-hub — xcode