packet-gpu-cli

Command-line interface for packet.ai GPU cloud.

Installation

npm install -g packet-gpu-cli

Quick Start

# Login to your packet.ai account
packet login

# List available GPU types
packet gpus

# Launch a GPU with VS Code pre-installed
packet launch --gpu rtx-pro-6000 --setup vscode

# Launch a bare GPU and wait for SSH
packet launch --gpu h100 --wait

# List your running instances
packet ps

# SSH into an instance
packet ssh <instance-id>

# View instance logs
packet logs <instance-id>

# Terminate an instance
packet terminate <instance-id>

Commands

Authentication

| Command | Description | |---------|-------------| | packet login | Authenticate with your API key | | packet logout | Remove stored credentials | | packet whoami | Show current account and balance |

GPU Management

| Command | Description | |---------|-------------| | packet gpus | List available GPU types and pricing | | packet launch --gpu <type> | Launch a new GPU instance | | packet ps | List your running instances | | packet ssh <id> | SSH into an instance | | packet logs <id> | View instance status and info | | packet terminate <id> | Terminate an instance |

Auto-Setup

| Command | Description | |---------|-------------| | packet setup list | List available setup presets | | packet setup <preset> <id> | Run a setup preset on an existing instance | | packet launch --setup <preset> | Launch with auto-setup |

packet launch

Launch a new GPU instance.

Options:

• -g, --gpu <type> - GPU type (e.g., rtx-pro-6000, h100)
• -n, --name <name> - Instance name
• -s, --setup <preset> - Auto-setup preset (see below)
• --gpus <count> - Number of GPUs (default: 1)
• -w, --wait - Wait for instance to be ready

packet setup

Auto-setup apps on GPU instances. Available presets:

| Preset | Description | Port | |--------|-------------|------| | vscode | VS Code in Browser (code-server) | 8080 | | jupyter | Jupyter Lab with data science packages | 8888 | | jupyter-torch | Jupyter Lab with PyTorch and CUDA | 8888 | | workspace | Persistent workspace linking | - | | full-dev | VS Code + Jupyter + Persistence | 8080, 8888 |

Launch with auto-setup:

packet launch --gpu rtx-pro-6000 --setup vscode
packet launch --gpu h100 --setup full-dev --name "my-dev-box"

Setup an existing instance:

packet setup vscode 12345
packet setup jupyter-torch 12345

List presets:

packet setup list

packet ssh <instance-id>

SSH into a running instance. Automatically uses the correct credentials.

Options:

• -c, --command <cmd> - Run a command instead of interactive shell
• --copy - Print the SSH command without connecting

packet logs <instance-id>

View instance status and connection info.

packet terminate <instance-id>

Terminate a running instance.

Options:

• -f, --force - Skip confirmation prompt

Configuration

Credentials are stored securely in your system's config directory:

• macOS: ~/Library/Preferences/packet-cli-nodejs/
• Linux: ~/.config/packet-cli-nodejs/
• Windows: %APPDATA%/packet-cli-nodejs/

Requirements

• Node.js 18.0.0 or higher
• An account at packet.ai
• sshpass for packet setup on existing instances (install with brew install sshpass or apt install sshpass)

Examples

Launch with VS Code pre-installed

packet launch --gpu rtx-pro-6000 --setup vscode --name "dev-server"

Launch with full dev environment

packet launch --gpu h100 --setup full-dev
# Includes VS Code (port 8080), Jupyter (port 8888), and persistent workspace

Setup Jupyter on an existing instance

# Get your instance ID
packet ps

# Install Jupyter + PyTorch
packet setup jupyter-torch 12345

CI/CD pipeline

# In GitHub Actions or similar
packet login --key $PACKET_API_KEY
INSTANCE=$(packet launch --gpu h100 --wait | grep "Instance ID" | awk '{print $3}')
packet ssh $INSTANCE -c "cd /workspace && python train.py"
packet terminate $INSTANCE -f

Support

• Website: packet.ai
• Documentation: packet.ai/cli
• Email: [email protected]

License

MIT