The AI-native
|
A lightweight AI terminal with a built-in editor, AI agents, and live web preview. 7 MB on disk. 300 ms cold start. BYOK or fully local.
A terminal you'll actually want to live in.
xterm.js with a WebGL renderer, multi-tab, and a file explorer with Catppuccin icons. Keyboard-first, instantly fast.
- WebGL-rendered xterm.js - silky at any scrollback
- Multi-tab terminals, editors, and previews side-by-side
- File explorer with Catppuccin icons + keyboard nav
- Project-wide search at ripgrep speed
An editor with real Vim mode and AI autocomplete.
Built-in editor with context-aware completions, first-class Vim, and prebuilt themes. No language-server setup - it just works.
- Context-aware AI autocomplete
- Real Vim mode (motions, registers, marks)
- Catppuccin, Tokyo Night, and more
- Sub-millisecond keystroke latency
Agents that read, plan, and ship diffs.
Multi-agents and sub-agents browse files, run commands, and propose changes - every edit lands in a reviewable diff before it touches disk.
- AI Edit Diffs - review every change inline
- TERAX.md memory + per-project configs in your repo
- Snippets & skills - composable, callable workflows
- BYOK or fully local via LM Studio
- Voice input - talk to your agents while you code
Preview your app without leaving the shell.
Auto-detects dev servers - Vite, Next, anything - and opens them as a tab beside your editor.
- Auto-detect Vite, Next, Astro, and more
- Live preview tab next to your editor
- Plans, tasks, and TODOs tracked in-app
- Hot-reload at the speed of your dev server
More built in. No plugins required.
Talk to your terminal. Whisper or your provider of choice.
Run fully offline through LM Studio. Your code stays on disk.
Fast AI autocompletion inside code editor.
Per-project memory and config - versioned with your repo.
Break work into tracked steps. Agents pick up where you left off.
Atomone, Tokyo Night, and friends - drop-in.
Terminals, editors, and previews side by side.
Ripgrep-class speed across files, with keyboard navigation.
Review every agent change inline before it touches disk.
Modal editing first-class. Bring your dotfiles.
Reusable workflows your agents can call by name.
Under 10 MB on disk. Boots in 300 ms. No Electron tax.
Pick a build. Run it.
No account, no telemetry. Verified releases on GitHub. SHA256 checksums published with every tag.
Questions, answered.
Still curious? Open a discussion on GitHub - we're around.