Tempest AI is the assistant built into the Tempest terminal. It reads the command output you highlight, explains errors, generates commands, runs them with one click, and follows up on the results — inside your SSH session instead of a separate chat window.

Which AI models can I use?

Anthropic Claude, OpenAI GPT, and Google Gemini — or any self-hosted OpenAI-compatible endpoint such as Ollama, vLLM, or llama.cpp. You bring your own API key and switch providers in Settings → AI.

Does Tempest see my prompts or terminal data?

No. AI requests go directly from your device to the provider you configured — Tempest never proxies them. The assistant only receives what you explicitly highlight, attach, or type, and with a self-hosted model nothing leaves your network at all.

Is the AI included in the free plan?

The AI assistant is part of Tempest Pro. The free plan covers SSH, Mosh, SFTP, FTP, S3, and WebDAV; Pro adds the AI agent, RDP, VNC, Kubernetes, Telnet, RCON, serial, and unlimited sync.

Can I run Claude Code or Codex on a remote server with it?

Yes — every SSH host has a Use AI Agents preset that launches Claude Code, Gemini CLI, Codex CLI, or OpenCode in a chosen directory the moment you connect. There's also a Tempest MCP server so AI clients can drive SSH sessions themselves.

AI terminal

The terminal with an AI agent that sees your session

Stop copy-pasting stack traces into a chatbot in another window. Tempest's AI assistant lives inside the terminal: highlight an error, ask what happened, get the fix, and run it — in the same tab, with Claude, GPT, Gemini, or a model you host yourself.

Download Tempest See pricing

AI is part of Tempest Pro. Windows, macOS, Linux, Android, iOS & Web.

Tempest AI assistant diagnosing a failed command next to the SSH session it happened in

In the loop

Explain, generate, run — without leaving the tab

The assistant works on what you give it: highlight command output, attach a file, ask a question. Answers come back as runnable commands, not prose to retype.

Paste an error, get a plain-English diagnosis and the fix.

Ask for a one-liner, click once to run it in the terminal.

The AI sees the output of what it ran and follows up — investigate a full disk or a failing systemd unit in one conversation. AI assistant docs

AI chat panel proposing a command with a one-click run button in Tempest

Your model, your keys

Claude, GPT, Gemini — or a model on your own hardware

Tempest doesn't lock you into one provider or resell tokens. Configure your API key once and requests go directly from your device to the provider — they never pass through Tempest's servers.

Anthropic Claude, OpenAI GPT, and Google Gemini out of the box.

Self-hosted: point at any OpenAI-compatible endpoint — Ollama, vLLM, llama.cpp, or your enterprise gateway — and prompts never leave your network.

The assistant only sees what you highlight, attach, or type. It never scrapes your history, hosts, or keys. How Tempest protects your privacy

Tempest AI provider settings — Anthropic, OpenAI, Google, and self-hosted endpoints

For agent workflows

Built for Claude Code, Codex & MCP too

If your AI does the typing, Tempest is still the terminal it types into. Connection presets drop you straight into a coding agent on the remote host, and the Tempest MCP server lets AI clients open SSH sessions themselves.

Use AI Agents preset: pick Claude Code, Gemini CLI, Codex CLI, or OpenCode and a working directory — Tempest launches it on connect. AI agents guide

Install the Tempest MCP server in Claude Code, Codex, Cursor, and more. MCP server setup

Sessions keep running when you close the laptop — check on your agent from your phone.

Tempest running a coding agent on a remote host over SSH

Frequently asked questions

What is Tempest AI?: Tempest AI is the assistant built into the Tempest terminal. It reads the command output you highlight, explains errors, generates commands, runs them with one click, and follows up on the results — inside your SSH session instead of a separate chat window.
Which AI models can I use?: Anthropic Claude, OpenAI GPT, and Google Gemini — or any self-hosted OpenAI-compatible endpoint such as Ollama, vLLM, or llama.cpp. You bring your own API key and switch providers in Settings → AI.
Does Tempest see my prompts or terminal data?: No. AI requests go directly from your device to the provider you configured — Tempest never proxies them. The assistant only receives what you explicitly highlight, attach, or type, and with a self-hosted model nothing leaves your network at all.
Is the AI included in the free plan?: The AI assistant is part of Tempest Pro. The free plan covers SSH, Mosh, SFTP, FTP, S3, and WebDAV; Pro adds the AI agent, RDP, VNC, Kubernetes, Telnet, RCON, serial, and unlimited sync.
Can I run Claude Code or Codex on a remote server with it?: Yes — every SSH host has a Use AI Agents preset that launches Claude Code, Gemini CLI, Codex CLI, or OpenCode in a chosen directory the moment you connect. There's also a Tempest MCP server so AI clients can drive SSH sessions themselves.

Keep exploring

Free SSH client Kubernetes terminal Web SSH client MobaXterm alternative Termius alternative

The AI is already in the terminal.

Download Tempest Read the docs