Skip to main content

Documentation Index

Fetch the complete documentation index at: https://www.rumus.ai/docs/llms.txt

Use this file to discover all available pages before exploring further.

This guide takes you from a fresh install to running commands with the AI assistant — first locally, then on a remote host.
Don’t have Rumus yet? Grab the latest build from rumus.ai/download, then come back here.

​
1. Open Rumus

Launch the app. Rumus opens straight into a local terminal tab running your default shell (zsh on macOS, PowerShell on Windows, bash on most Linux distros). There’s no setup wizard and no forced sign-in — you can start typing immediately. Run something familiar to confirm it works: 
ls

​
2. Try smart autocomplete

Smart autocomplete is on by default. As you type, Rumus suggests context-aware completions inline. Press → or Tab to accept. Try typing a few characters of a command you’ve used before, or start a git and let Rumus suggest the next word. To tune or turn it off later, see Settings → AI → Smart Autocomplete.

​
3. Open another local shell

Want to try a different shell — or jump into WSL on Windows? Click the chevron next to the new-tab button at the right side of the tab bar. The dropdown lists every shell Rumus detected on your machine:
  • macOS / Linux: zsh, bash, fish, sh, and any custom profiles you’ve added.
  • Windows: PowerShell, cmd, Git Bash, plus each installed WSL distribution shown by name (Ubuntu, Debian, etc.).
Pick one and a new tab opens with that shell.

​
4. Talk to the AI assistant

Open the AI sidebar by clicking the right-sidebar toggle at the far right of the title bar (the panel-right icon). Try a few prompts:

Explain a command

“What does find . -mtime -7 -type f do?”

Run something for you

“Show me the top 5 largest files in this directory.” The agent proposes a command and asks for approval before running it.

Debug from logs

Paste an error and ask: “What’s wrong and how do I fix it?”

Plan a multi-step task

Hand the agent a multi-step job. It drafts a plan first, then works through it — Plan mode kicks in automatically, no toggle needed.
By default the assistant uses Rumus’s built-in models. Want to use your own Anthropic, OpenAI, or local Ollama key instead? See Models & Providers.

​
5. Sign in

Have a Rumus account? Sign in now. It’s not required to use a local terminal, but it unlocks:
  • The built-in AI models included with paid plans.
  • Cloud sync of your settings, profiles, and preferences.
  • Encrypted vault sync, so the hosts and credentials you set up next can follow you to other devices.
Click the menu button at the far right of the title bar and choose Sign In. A browser window handles the OAuth flow; come back to Rumus once you’ve authorized.
Sign in before initializing your vault in the next step. Doing so binds the vault to your account so it can sync across devices.

​
6. Set up the vault to manage remote servers

To save SSH hosts and credentials, Rumus needs an encrypted vault. The first time you try to connect somewhere remote, the app walks you through creating it.
1

Open Remote Connection

Click the chevron next to the new-tab button in the tab bar and choose Remote Connection.
2

Initialize the vault

If this is your first time, Rumus prompts you to set up a secret key. The key encrypts everything in the vault on your device — Rumus servers only ever see ciphertext.
  • Create a new secret key — Rumus generates one and downloads it as a file. Keep this file in a safe place; it’s the only way to recover the vault on a new machine.
  • Recover an existing key — import the key file from another device.
Optionally set a PIN for fast unlock so you don’t re-enter the secret key every session.
3

Land in the host manager

Once the vault is ready, Rumus drops you into the host management view.

​
7. Add a remote host

In the host manager:
1

Click New Host

Hit the New Host button to start a new entry.
2

Fill in the connection details

  • Name — a label for your reference (e.g. prod-web-1).
  • Hostname / IP, Port (default 22), Username.
  • Auth — password, an SSH key file, or a key already saved in the vault.
  • Optional: a jump host chain, proxy, or environment variables.
3

Save and connect

Save the host, then double-click it to open a session in a new tab. Credentials never leave your device unencrypted.
For multi-hop access through a bastion, see Jump Hosts.

​
8. Use the AI on the remote session

With the remote tab focused, open the AI sidebar again. The assistant now operates in the context of the remote host — it can read remote files, propose remote commands, and help you debug live. Some prompts to try:
  • “What’s eating the disk on this box? Show me the 10 largest paths under /var.”
  • “Tail the nginx error log and summarize the last 20 entries.”
  • “This service is failing to start. Investigate and propose a fix.”
The agent asks for approval before running anything destructive. To pre-approve safe patterns, see Settings → AI → Command Approval.

​
You’re set

You now know how to:
  • Open a local terminal and use smart autocomplete.
  • Switch between local shells (and WSL on Windows).
  • Chat with the AI and let it run commands for you.
  • (Optionally) sign in for built-in models and sync.
  • Set up the vault, save remote hosts, and use the AI against them.
Stuck on a step? Ask in the Rumus community — the team and other users are happy to help.

​
Where to go next

Core concepts

Workspaces, the infinite canvas, profiles, and how the agent fits in.

AI assistant

Plan mode, sub-agents, MCP, command approval — the full toolbox.

Built-in models

What’s included, pricing, and how to view your usage.

Migrate from another terminal

Bringing settings over from iTerm2, Warp, or Windows Terminal.