Diagnoses and resolves issues across NemoClaw agent sandbox, policy, and inference layers.
The NemoClaw Agent Debugger is a specialized skill designed to troubleshoot complex issues within NVIDIA NemoClaw environments. It provides a systematic, five-step diagnostic framework that guides users from the outermost sandbox layer down to the core inference engine. By verifying sandbox status, auditing network and filesystem policies, streaming real-time logs, and executing plugin health checks, this skill helps developers quickly identify why agents might be behaving unexpectedly, why network requests are being blocked, or why inference calls are failing.
Características Principales
010 GitHub stars
02Quick diagnostic collection for bug reporting and support
03Real-time log streaming with specific error pattern identification
04Automated plugin health monitoring via the 'doctor' command
05Step-by-step diagnostic checklist for systematic troubleshooting
06Detailed network and filesystem policy auditing
Casos de Uso
01Debugging credential injection and inference timeouts within an OpenShell container
02Investigating why an agent is unable to reach specific API endpoints or external services
03Fixing issues where a NemoClaw sandbox fails to initialize or reports a degraded state
