OpenEng is one platform with one account. Sign in once, then pick your track: an engineering app (Terminal, Kubernetes, Data, Git or Chat), or OpenEng AI.
Install a small engine, start it, and drive it from the browser. Free. The steps below use Terminal.
Author in the Console, publish a config, and run it on the openeng engine.
One account
Every OpenEng surface signs in with Google through auth.openeng.app: the five apps, the Console, the Web Console and the billing site. There is no separate registration: your Google sign-in is your OpenEng account, and the same account works everywhere. The engineering apps are free; for OpenEng AI pricing, see Plans & billing.
1. Pick an app
Every OpenEng engineering app is a small web UI backed by an engine that runs on your own machine. You install the engine once, start it, and the web app connects to it over a sealed local channel. Your code, credentials and data stay on your computer unless you point the engine somewhere else, such as a remote host, a connector in your own cloud or an external model server; none of it passes through OpenEng.
Each app is independent and ships its own engine and web UI. Start with whichever you need:
Terminal · terminal.openeng.app
A real local shell. Engine: openeng-terminal on 127.0.0.1:50052.
Kubernetes · k8s.openeng.app
Clusters, pods, logs, exec. Engine: openeng-kubernetes on 127.0.0.1:50054.
Data · data.openeng.app
Query SQL, document, key-value, graph, object and streaming databases. Engine: openeng-data on 127.0.0.1:50057.
Git · git.openeng.app
Repositories and history. Engine: openeng-git on 127.0.0.1:50053. The engine is not published yet.
Chat · chat.openeng.app
Local AI chat + a cross-app agent. Engine: openeng-chat on 127.0.0.1:50056. The engine is not published yet.
The Terminal, Kubernetes and Data engines are published. The Git and Chat web apps are live, but their engines haven't been published yet, so start with one of the other three. The rest of this guide uses Terminal; the steps are the same for Kubernetes and Data.
2. Install the engine
Install the engine with its one-line installer. Terminal and Data run on Linux x86_64 and macOS, Kubernetes on Linux x86_64 and ARM64 and macOS, and Windows through WSL; the macOS builds are older releases pending a rebuild:
curl -fsSL https://openeng.app/engine/<app>/install.sh | bash
For Terminal:
curl -fsSL https://openeng.app/engine/terminal/install.sh | bash
The installer downloads the engine binary for your platform, checks it against its published SHA-256 checksum (when it can fetch the checksum and finds sha256sum or shasum; it prints checksum verified when it did), and puts it on your PATH (in ~/.local/bin). App engines update themselves at start and every 30 minutes, so you stay current automatically. On Windows, run the installer inside WSL. Platforms, options and uninstalling are covered in Install, platforms and updates; the same commands are on openeng.app/download.
Note
Everything runs locally after install. The engine makes two kinds of network call to OpenEng: an update check against releases.openeng.app at start and every 30 minutes (the current macOS builds, which are older releases, still use the release service's former address; the next macOS builds move to releases.openeng.app), and its sign-in: the one-time pairing exchange at boot, which also fetches a browser-trusted certificate for the loopback engine so the https:// web app can talk to it with no manual trust step, and then hourly token renewal while it runs. When an update exists, the engine also downloads it from openeng.app. The Terminal engine also makes one call outside OpenEng: it downloads its on-device suggestion model (about 130 MB) once from huggingface.co, and sends none of your command history (Security).
3. Start the engine & connect
Open terminal.openeng.app and sign in with Google. Until an engine is reachable, the app shows its pairing screen, Connect to your terminal engine, with a start command that carries a one-time key. Copy it with Copy command and run it in a terminal on the same machine:
openeng-terminal serve --oauth=<one-time key>
Among the boot lines you will see:
openeng-terminal: serving gRPC + gRPC-Web (sealed) on https://engine.openeng.app:50052 — Ctrl-C to stop
Leave it running and return to the web app — it detects the engine and connects automatically. Refreshing the page is safe; in-flight work re-attaches. If you opened a deep link before pairing, the app takes you there once it connects; otherwise Terminal opens at /terminal.
About the one-time key:
- It is minted for you in the browser after you sign in, and it expires after about five minutes. A reload reuses the key while it's still valid; Regenerate mints a new one.
- The OpenEng API mints at most five unused keys an hour per account, shared by every OpenEng app (a key an engine has used no longer counts), and each app also counts the keys it mints in your browser.
- Once the engine connects, the app forgets the key, so the next pairing gets a fresh one.
Tip
The engine keeps its terminal quiet after the boot lines. Crash backtraces go to ~/ instead of your console (the Chat engine sends all of its output there).
Run your first command
Terminal opens with a tab ready; add another with the + in the tab strip, or open the command palette with ⌘+K (Ctrl+K on Linux and Windows) and choose New terminal in …. Type a command in the command bar at the bottom, for example ls -la, and press Enter. It runs on your machine, and the output appears as a block with its status and duration. See Run blocks for how blocks work.
4. Where to next
Terminal →
Groups, tabs, run blocks, suggestions, SSH and cloud shells.
Chat →
Add a model, create a group, and connect your other apps to the agent (the Chat engine isn't published yet).
Connectors →
Reach cloud clusters, databases and shells without handing over credentials.
Architecture →
Understand the sealed local model that makes this private by design.
Install →
Every engine's installer and platforms, updates, and how to uninstall.
OpenEng AI track
OpenEng AI is authored in the browser and run by the openeng engine on your machine.
Status on September 25, 2026
The OpenEng AI services behind the Console and the engine (the OpenEng AI side of the OpenEng API, and the Broker) aren't live yet: the Console can't load or save your work and no engine can register, whatever the plan. The steps below describe how it works once they are (OpenEng AI).
-
Sign in to the Console at console.openeng.app. The Console is where you define models, agents, templates and the rest, and publish them as versioned configs. The Free plan includes signing in and authoring.
-
Install the engine (Linux x86_64; on Windows, inside WSL):
bashcurl -fsSL https://openeng.app/engine/install.sh | bash -
Start it with your OpenEng key, an API key from the Console (Deploy → API keys):
bashopeneng serve --key=<your OpenEng key>The engine opens no local port: it registers with the OpenEng API and dials out to the OpenEng Broker, and you drive it from the Web Console at app.openeng.app. Registering an engine requires an active Pro plan, and Pro launches in Q4 2026; it can't be purchased yet (Plans & billing).
The OpenEng AI quickstart walks through the Console and your first publish, and How OpenEng works explains the Broker.