The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides scripts (scripts/record_trace.py, scripts/analyze_trace.py) that facilitate performance profiling by executing the Xcode xctrace utility via subprocess.Popen and subprocess.run. The commands are constructed using argument lists rather than raw shell strings, preventing command injection vulnerabilities.
[DATA_EXPOSURE]: The trace recording functionality includes a privacy-conscious design. The script scripts/record_trace.py requires an explicit --allow-system-wide-recording flag to capture data beyond the target app and features a redaction mechanism (_format_command_for_display) to ensure that environment variables passed during app launch are not leaked into the console logs.
[INDIRECT_PROMPT_INJECTION]: The scripts/analyze_trace.py script parses external data (Xcode .trace files) to generate performance reports. While this introduces an ingestion surface for untrusted data, the extracted strings (such as view names and symbols) are used strictly for informational reporting and filtering. The logic does not allow external data to influence the execution of system commands or other sensitive operations.
[SAFE]: All dependencies in the included Python scripts are restricted to the Python standard library. The skill does not perform any network operations or external downloads, relying entirely on local tools and bundled documentation.

swiftui