peters/horizon

GPU-accelerated terminal board that puts all your sessions on an infinite canvas

What it solves

Horizon is a GPU-accelerated terminal emulator that replaces traditional tabbed or tiled layouts with an infinite 2D canvas. This allows users to organize multiple terminal sessions, AI agents, and documentation in a flexible, non-linear workspace where they can pan, zoom, and group related tasks without losing context.

How it works

Built with Rust and the Alacritty terminal engine, Horizon renders terminal panels on a boundless surface. It uses workspaces to group related panels (with auto-arrange options like rows, columns, or grids) and includes a minimap for navigation. It integrates directly with AI agent CLIs (such as Claude Code and Gemini CLI) and provides a built-in Git status panel and Markdown editor. For accessibility and speed, it features an optional local speech-to-text system powered by transcribe.cpp for dictating commands directly into the terminal.

Who it’s for

Developers, DevOps engineers, and power users who manage complex workflows involving multiple remote hosts, AI coding assistants, and simultaneous monitoring of several processes.

Highlights

  • Infinite Canvas: A 2D workspace with pan/zoom capabilities and a minimap for fast navigation.
  • AI Agent Integration: First-class support for agents like Claude Code and Codex, including a usage dashboard to track token spend.
  • Local Speech Input: Opt-in, local transcription using Whisper GGUF models for push-to-talk dictation.
  • Session Persistence: Automatically restores panel layouts, scroll positions, and terminal history across restarts.
  • Remote Host Management: Fast discovery and connection to hosts via SSH config and Tailscale.
  • Integrated Tooling: Built-in Git status tracking and a split-view Markdown editor.

Related

  • Project
    nowledge-co/con-terminal

    A GPU-accelerated terminal emulator with an integrated AI harness for agent-native CLI workflows, designed for users who prefer a terminal-first approach to AI assistance.

  • Project
    eneskirca/nodeterm

    A node-based terminal manager that organizes terminals and AI agents on an infinite spatial canvas and kanban board to prevent context loss.

  • Project
    l0ng-ai/tty7

    A high-performance terminal workbench that provides persistent sessions and deep integration for AI coding agents, allowing them to be monitored and scripted.

  • Project
    alpbahadur/49Agents

    A 2D agentic IDE that replaces terminal tabs and SSH with a zoomable infinite canvas for managing multiple machines, terminals, and AI agents in one space.

  • Project
    maddada/Ghostex

    A native desktop workspace that integrates multiple AI agent CLIs into a single environment with persistent sessions, mobile access, and cross-agent orchestration.