voxeltron ボクセルトロン
Install

Terminal UI

Voxeltron ships a rich terminal interface built with Bubbletea. It connects to the daemon over gRPC and gives you full control of your platform without leaving the terminal.

Overview

The voxeltron binary is a standalone Go program that communicates with voxeltrond over gRPC. Every action you can take in the TUI maps to one or more gRPC calls — there is no hidden state, no local database, and no background process. Close the TUI and nothing changes on the server.

The TUI is the primary interface for self-hosted Voxeltron. If you use Voxeltron Cloud, you also get a web dashboard, but the TUI works there too.

Key Views

The interface is organized into discrete views, each focused on one operational concern.

Dashboard

Live sparklines for CPU, memory, and request rate. Service health indicators and deployment status at a glance.

Deployments

Streamed build logs, release history, and one-key rollback. Filter by app or browse all deployments.

Databases

Provision, inspect, and manage PostgreSQL, MySQL, Redis, and MongoDB instances. View connection strings and trigger backups.

AI Panel

Ask questions in natural language, execute runbooks, and get contextual suggestions — all without leaving the terminal.

Configuration

Edit environment variables, secrets, scaling rules, and domain bindings. Changes are applied on the next deploy or instantly via hot-reload.

The TUI is fully keyboard-driven. No mouse required.

Key Action
Tab / Shift+Tab Switch between views
j / k Move up / down in lists
Enter Select / confirm
/ Search / filter
? Show help overlay
q Quit / back

AI Panel

Press Tab to navigate to the AI panel, then type a question or command in natural language. 

> Why is my Rails app returning 502s?

Checking deployment status for rails-app...
Last deploy: 4 minutes ago (v23 → v24)
Health check failing on /health — container OOMKilled (256Mi limit).

Suggested fix: increase memory limit to 512Mi.
Run `voxeltron config set rails-app MEMORY=512Mi`? [y/N]

The AI panel supports three interaction modes:

  • Natural language queries — ask about service health, logs, configuration, and costs.
  • Tool execution with confirmation — the AI suggests commands and waits for your explicit approval before running them.
  • Skill and runbook streaming — trigger multi-step playbooks (e.g. "rotate database credentials") and watch each step execute in real time.
Tool execution always requires confirmation. The AI will never make changes to your infrastructure without your explicit approval.

Connecting to the Daemon

The TUI connects to voxeltrond over gRPC with TLS. By default it looks for the daemon on localhost:7443. 

# Connect to local daemon (default)
voxeltron

# Connect to a remote server
voxeltron --host myserver.example.com

# Connect to a specific port
voxeltron --host myserver.example.com --port 8443

The connection uses mutual TLS by default. On first run, the TUI generates a client certificate and registers it with the daemon. You can also provide your own certificates: 

voxeltron --cert /path/to/client.crt --key /path/to/client.key

Installation

The TUI is distributed as a single static binary. Install it with Homebrew or download it directly.

Homebrew (macOS and Linux)

brew install voxeltron/tap/voxeltron

Direct Download

# macOS (Apple Silicon)
curl -fsSL https://get.voxeltron.dev/tui/darwin-arm64 -o voxeltron
chmod +x voxeltron
sudo mv voxeltron /usr/local/bin/

# Linux (x86_64)
curl -fsSL https://get.voxeltron.dev/tui/linux-amd64 -o voxeltron
chmod +x voxeltron
sudo mv voxeltron /usr/local/bin/

Verify the installation:

voxeltron --version
← First Deployment
Databases →