---
title: What is Admiral
description: Admiral is a native control station for coding agents. The agents run on your machines.
section: Start
order: 1
---

# What is Admiral

Admiral is a native control station for coding agents. You tell it what to build. It starts a coding agent on your machine, shows the agent's work as it happens, and gives you the controls: send instructions, queue instructions, interrupt, and review the changes.

## What Admiral does

- **Starts agents.** Open the app, pick a project folder, pick an agent and model, and type a task.
- **Shows the work live.** The agent's reply, tool use, and file edits appear in a transcript as they happen.
- **Keeps your instructions.** Anything you send is queued. The agent receives it, even if you were offline when you sent it.
- **Shows the result.** The changes pane shows every file the agent changed, as a diff, file by file.
- **Works across your devices.** Start work on your desktop, then read and steer the same session from your phone.

## What Admiral is not

- **It is not an agent.** Admiral does not write code. It drives agents that do: OpenCode, Claude Code, and Gemini, plus others in legacy mode.
- **It is not a cloud service.** A fresh install works with no account and no network. Your code and sessions stay on your machines.

## Where it runs

- **Desktop.** A full application for macOS, Windows, and Linux.
- **Phone.** An iOS app that shows and steers sessions on your other devices.
- **Servers.** A headless mode that runs the engine with no window, for an always-on machine.

## The one-sentence summary

Your agents live on your machines. Admiral is the flight deck that runs them from anywhere.

## Next steps

- [Quickstart](quickstart.md): start your first session.
- [Projects and sessions](../concepts/projects-and-sessions.md): how Admiral organizes work.