The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's test suite, found in the functions/ directory (e.g., test_exit_gate.py, test_phase_detector.py), utilizes subprocess.run to execute local Python scripts for validation purposes. These calls are confined to testing the skill's own CLI tools and do not involve untrusted user input or remote code execution.
[SAFE]: No network operations, data exfiltration patterns, or hardcoded credentials were detected. The skill operates entirely on local project configuration and provided task context.
[SAFE]: The autonomous 'loop mode' implementation includes a critical safety guard that requires a max_iterations limit when the never_pause_on_stagnation flag is enabled, effectively preventing infinite loop scenarios during unattended operations.
[SAFE]: The skill uses structured pilot-signal JSON blocks for status reporting, ensuring unambiguous state communication between the agent and the orchestration platform without relying on potentially injectable natural language instructions for core logic.

nav-loop