Skip to content
kaiport

Quick start

Your first ten minutes with kaiport – an SSH connection, the file browser, setting up AI Pilot and approving its first command.

On this page

This guide takes you through your first ten minutes with kaiport. You save an SSH connection and connect to it, browse the server's files, set up AI Pilot and approve its first command. At the end you know the most important parts of the app.

Before you start

You need:

  • kaiport installed on your computer; see Installation.
  • An SSH server you can log in to: its host name or IP address, your user name, and a password, a private key or a running SSH agent.
  • For AI Pilot: kaiport Pro, and an API key from a provider such as Anthropic or OpenAI or a local model server such as Ollama or LM Studio. Steps 4 and 5 need it; the rest works in the free version.

Step 1: Create an SSH connection

The start window greets you with Welcome to kaiport and two steps: Create an SSH connection and Set up AI Pilot.

  1. Click Create an SSH connection, or New connection in the top right corner.
  2. Leave SSH selected under Connection type.
  3. Enter the server under Host, for example server.example.com or an IP address, and your user name under Username. Port stays at 22 unless your server uses another port.
  4. Under Authentication, choose how you log in:
    • Password: enter it, or leave the field empty to be asked when you connect.
    • SSH key: click Choose and select your private key. kaiport imports it into its encrypted vault. If the key is protected, enter its Passphrase.
    • SSH agent: kaiport uses the keys of your running SSH agent. The Mac App Store version has no agent; use SSH key there.
  5. Optionally enter a Name. Without one, the connection is called user@host.
  6. Click Create.

The connection now appears as a card in the start window. For all fields, including jump hosts and port forwarding, see SSH connections.

Step 2: Connect and trust the server

  1. Click the connection's card. A terminal window opens and kaiport connects. The start window closes; ⌘⇧HCtrlShiftH brings it back.
  2. The first time you connect to a server, the dialog Unknown host says that the authenticity of the server cannot be established. It shows the type and the fingerprint of the server's host key, for example SHA256:….
  3. Compare the fingerprint with the one of your server. Your hosting provider or administrator can tell you, or you run ssh-keygen -lf /etc/ssh/ssh_host_ed25519_key.pub on the server for its Ed25519 key.
  4. If the fingerprints match, click Trust. kaiport stores the key and doesn't ask again for this server.
  5. If you didn't save a password, the dialog Password required asks for it. Select Store encrypted for this connection if kaiport should remember it, then click Connect.

The shell of your server appears. kaiport keeps its own list of known hosts, separate from ~/.ssh/known_hosts, so it also asks once for servers you know from other SSH clients. You find the stored keys under Settings → Known hosts.

Step 3: Open the file browser

  1. Press ⌘BCtrlB, or click Files on the rail, the narrow bar at the right edge of the window.
  2. The file browser opens next to the terminal and shows your home directory on the server over SFTP.
  3. Double-click a folder to open it. Double-click a text file to open it in the editor window; other files are downloaded to a folder you choose.
  4. To upload, drag files from your computer into the list.

For a folder, Open in terminal changes the terminal to it with cd, and Ask the AI about this file hands a file to AI Pilot. See File browser.

Step 4: Set up AI Pilot

AI Pilot is part of kaiport Pro. You can fill in its settings without Pro; asking AI Pilot opens the offer for kaiport Pro, with 14 days free for new subscribers.

  1. Press ⌘,Ctrl, to open Settings, and scroll to AI Pilot. The step Set up AI Pilot in the start window opens the settings, too.
  2. Under Provider, click your provider: Anthropic, OpenAI, OpenRouter, Groq, Mistral, DeepSeek, Together, Ollama (local) or LM Studio (local). For another OpenAI-compatible service, enter its API URL instead.
  3. Paste your API key into API token. Anthropic keys start with sk-ant-. Local servers such as Ollama and LM Studio need no token.
  4. Click Load models and choose a Model from the list. You can also type a model name.
  5. Click Test connection. The message "Connection successful – the model responds." confirms that everything works.
  6. Click Save.

Your token is stored encrypted. For all other settings, such as Autopilot and whether AI Pilot may read file contents, see Set up AI Pilot.

Step 5: Ask AI Pilot and approve its first command

  1. Switch back to the terminal window of your server.
  2. Press ⌘JCtrlJ or click AI Pilot on the rail. The panel asks How can I help? and offers suggestions.
  3. Click the suggestion "How are disk and memory usage?", or type your own question and press Enter. ShiftEnter starts a new line.
  4. AI Pilot proposes its first command and waits. The card shows Awaiting approval and asks "Run this command in the terminal?".
  5. Check the command. Click Run to run it in your terminal, or Deny to refuse it.
  6. AI Pilot reads the output, continues step by step and sums up the result. Every further command asks for your approval again.
AI Pilot asks before it runs a command
AI Pilot waiting for approval of a command, with the buttons Run and DenyAI Pilot waiting for approval of a command, with the buttons Run and Deny

AI Pilot asks before it runs a command

The command runs in the terminal you see, so you can watch it and step in at any time. AI Pilot never asks for passwords in the chat: when a command such as sudo needs one, type it in the terminal and AI Pilot continues.

Next steps