4090ctl is a remote server management skill for controlling NVIDIA RTX 4090 GPU servers via SSH. It provides a streamlined interface for monitoring Docker containers, restarting services, checking system status, and managing GPU workloads — all without manually SSH-ing into the server and running commands by hand.
Managing a GPU server for AI training, inference, or rendering typically involves frequent SSH sessions, Docker commands, and system monitoring. 4090ctl wraps these common operations into a skill that your AI assistant can execute on your behalf, saving time and reducing the risk of mistyped commands on production hardware.
Key Features
- Docker container monitoring — list running containers, check logs, restart failed services
- System status checks — view GPU utilization, memory usage, CPU load, and disk space
- Service management — start, stop, and restart Docker services and system daemons
- SSH-based execution — all commands run securely over SSH to your remote server
- GPU-specific monitoring — nvidia-smi integration for real-time GPU temperature, utilization, and memory stats
Use Cases
- Checking whether your AI training job is still running and how much GPU memory it is using
- Restarting a crashed Docker container on your GPU server without opening a terminal
- Monitoring system resources during heavy inference workloads to catch bottlenecks early
- Running routine maintenance tasks like clearing Docker caches or rotating logs
How It Works
4090ctl connects to your server via SSH using stored credentials and executes system commands on your behalf. When you ask to check GPU status or restart a container, the skill translates your request into the appropriate shell commands, executes them remotely, and returns the formatted output.
Getting Started
Install the skill from ClawHub and configure your server's SSH connection details (hostname, username, and key or password). Once connected, you can monitor and manage your 4090 server by simply describing what you need in natural language.
