Mel – the agentic terminal
The AI terminal for teams.
Learn · agentic coding desktop app
What is an agentic coding desktop app?
Updated
An agentic coding desktop app is an app you install on your computer where AI agents plan, write, run and test code in your real project. It gives the agents a window: you watch them work, approve what they may do and review what they changed.
Mel – the agentic terminal is an agentic coding desktop app for Mac, Windows and Linux. A command-line agent is powerful but lives in one terminal session. A cloud agent works on a copy of your code on someone else's servers. A desktop app sits in between: the agents work on your machine, with your tools, and the app shows you everything they do.
What a desktop app adds over a CLI agent
- A view of many agents at once: tabs, panes, and a status for each.
- Approval cards you can read, with the exact command or the diff of the edit before you say yes.
- Notifications when an agent in the background needs you.
- A review panel for every change, instead of scrolling a terminal.
- Your own tools. The agents use your shell, your installed runtimes and your git setup.
The kinds of agentic coding app
Desktop terminal app. The agents work where builds, tests and git already run. Mel is this kind.
Desktop editor app. The agents work inside a code editor, next to the files you are reading.
CLI tool. A program you start in any terminal. Light, but one session per window.
Cloud agent. You hand off a task and check back later, often as a pull request. Your code runs on the vendor's servers.
How Mel does it
- A native desktop app for macOS (Apple Silicon and Intel), Windows 10 and 11, and Linux.
- A team of agents that plans, builds, tests and ships a task, each worker in its own pane.
- Approval cards that show what will run. File-edit cards show the diff of exactly what will be written.
- The Runs screen, the sidebar and the inbox tell you which agent needs you.
- The VS Code extension brings Mel's agents into the editor.
- Beyond code, the same agents work with research, PDFs, photos and files.
- Coming soon: Mel on iPhone, iPad and in the browser, and Mel Remote.
How the kinds compare
| CLI tool | Cloud agent | Desktop app (Mel) | |
|---|---|---|---|
| Where the work runs | Your computer | The vendor's servers | Your computer |
| Uses your own tools and shell | Yes | No, a copy of the repo | Yes |
| Many agents in view | One per window | A list of jobs | Tabs, panes and the Runs screen |
| Approvals before each step | Varies | Review at the end | Configurable approvals |
| Notifications when blocked | No | Varies | Yes |
FAQ
What is an agentic coding desktop app? An app on your computer where AI agents plan, write, run and test code in your project, with a window to watch, approve and review their work.
Does my code leave my computer? The agents read, edit and run on your machine. Only the context a model call needs is sent to the model.
Is the Mel desktop app free? Yes, with your own API key, on Mac, Windows and Linux. Pro and Max add hosted credits; Business is $50 a seat a month.
Is there a Mel phone app? Not yet. Mel on iPhone and iPad is coming soon.
Mel – the agentic terminal · openmel.dev. A multi-agent, multi-model AI terminal for Mac, Windows, Linux and VS Code. Free with your own API key.