Repositories
Keep your forge.
GitHub
Cloud + Enterprise
GitLab
Cloud + self-managed
Your forge
Available on request
Performance
Native speed. No compromises.
The whole fleet in one workspace — 100% Rust, rendered with Metal on your GPU. No Electron, no embedded browser. Ready before your fingers leave the keyboard.
<0.5s
Cold start
160MB
Memory at idle
1
Native process
100%
Rust · GPU-rendered
Your operating layer
Your stack stays. The chaos goes.
Your repositories. Your models.
A native control plane.
Agents & models
Choose per session.
Claude 
Codex 
Cursor 
Grok Oh My Pi 
OpenCode Pi Kimi Code 
Antigravity 
Gemini 
Copilot Factory Hermes 
Kiro Qwen Code Your agent Any model your provider makes available
super.engineering Control plane The full workflow
From first prompt to final merge.
Every task stays visible from isolated worktree to merge.
Worktrees
Isolated by task
Sessions
Terminal + chat
Context
Files + targets
Review & ship
Diff to merge
Local credentials · direct model traffic · human review
Teams & enterprise
Bring every agent.
Private by default · agent agnostic · rollout support
Orchestration
One lead.
Five specialists.
Put Fable 5 in the lead. Agents message one another, share coordination state, and hand work off—while every decision stays visible in one native workspace.
CLI control Same live workspace
sc team run Launch coordinated roles sc agent send Message any live agent sc coordination-state Share decisions and handoffs Lead · Fable 5
Plan · route · synthesize
Agent mesh · direct messages
Frontend
Grok
Backend
Claude
Tests
Codex
Research
Antigravity
Review
Pi
Shared state · direct handoffs · team memory
A coordinated result
Ready for human review
Message · share state · hand off · review
Flexibility
Coding agents scaled. The tooling didn’t.
Built from the ground up for agentic engineering.
Isolation
One task. One worktree.
Every agent gets an isolated branch, terminal, session history, and review diff — without stepping on another task.
Remote
Your Mac. Remote compute.
Run Git, terminals, and coding agents beside your code on an SSH-connected Linux or Mac host.
Layouts
Arrange the fleet.
Split sessions into views, tabs, and panes — or float one in Picture-in-Picture.
Review
Review before it lands.
Inspect diffs, checks, comments, and PR or MR state from the worktree that produced them.
Attention
Know who needs you.
Agent state and native notifications take you straight to approvals, input, or review.
Continuity
Resume everything.
Reopen worktrees with tabs, chats, provider sessions, and layouts restored.
Shared context
Cross-repo by design.
Group repos under one feature context — shared instructions, one review pass, and worktrees that appear as the task needs them.
Orchestration
Orchestrate a team.
Launch, steer, and follow labeled agents from the app or CLI. Every role stays visible in one workspace.
Git workflow
Your branch knows what comes next.
From local changes to merge, super.engineering surfaces the action your branch needs next—and routes it to the right provider, prompt, and view.
feat/workspace-previews +128 −24
Add instant workspace previews
feat/command-palette +84 −12
Make every action keyboard-first
fix/session-restore +62 −18
Restore sessions without drift
feat/review-handoff +96 −8
Ship reviewed changes with confidence
feat/product-narrative +74 −59
Sharpen the product narrative and responsiveness
Route selected action Create PR Global
In their own words
Engineers love us.
Public displays of affection.
FAQ
Questions & answers
What is super.engineering?
super.engineering is a native macOS app for running parallel AI coding agents. Built entirely in Rust, it gives you isolated git worktrees, a GPU-rendered terminal, customizable layouts and themes, and multi-agent orchestration in a single, keyboard-driven interface. Think of it as a purpose-built workspace for agentic engineering.
How is this different from other agent orchestration tools?
Most tools in this space are built on web technology and cap you at 5-7 parallel agents. super.engineering is a native app that runs unlimited parallel agents — scaling with your hardware, not an arbitrary limit. It launches instantly, renders the terminal on the GPU, and gives you deep customization: multiple panel layouts, picture-in-picture windows, per-workspace theming, configurable keybindings, and integration with 10+ editors and terminals.
What agents does it support?
The app is agent-agnostic, with 15 providers built in — Claude Code, Codex, Cursor, Gemini, Grok, Copilot, OpenCode, Kimi Code, Qwen Code, Antigravity, Factory, Kiro, Pi, Oh My Pi, and Hermes — plus any custom CLI agent. Agents run as native subprocesses using your existing accounts and API keys — nothing is proxied or wrapped.
What operating systems are supported?
macOS is currently supported with native Apple Silicon performance. Windows and Linux support are planned.
Join the Discord to hear when your platform shipsHow does super.engineering handle data?
super.engineering never sees your code, credentials, conversations, or project data — everything stays on your machine. AI agents run as native subprocesses using your own accounts and API keys, which are never stored or proxied through the app. We send minimal anonymous telemetry so we can see active usage, and you can opt out in Settings → Privacy.
Read the full Privacy PolicyWho's behind super.engineering?
A small team of engineers who built the tool we wanted to use. We started super.engineering because nothing else was fast enough or flexible enough for how we work.
How do I get started?
Download super.engineering, install it, and open any git repository to start your first session.
Stop babysitting terminals.
Download super.engineering and command your first fleet.