The Agent Skills Directory

[COMMAND_EXECUTION]: The skill incorporates Python-based diagnostic utilities (record_trace.py, analyze_trace.py) designed to interface with the system's xctrace utility for performance data collection.
The implementation uses secure subprocess APIs with list-based arguments, which mitigates risk of shell injection.
The record_trace.py script specifically includes logic to redact environment variable values from command logs to prevent accidental credential exposure during app launch profiling.
[PROMPT_INJECTION]: The skill's analysis workflow relies on processing external performance data (Xcode Instruments .trace files).
Ingestion points: External data is ingested and parsed in scripts/analyze_trace.py using the Python standard library's streaming XML parser.
Boundary markers: The skill does not implement explicit boundary markers for data extracted from the trace files.
Capability inventory: The skill has the capability to execute system commands via the provided profiling scripts.
Sanitization: Inputs derived from trace files are handled as data strings and are not executed; system interactions are strictly confined to parameters passed to the xctrace binary using secure list-based arguments.

swiftui