Mel – the agentic terminal
The AI terminal for teams.
Learn · terminal with command blocks
What is a terminal with command blocks?
Updated
A terminal with command blocks groups each command you run with its output, working folder and timing into one block. Instead of one long scroll of text, you get separate results you can select, copy, filter and hand to an AI agent one at a time.
Mel – the agentic terminal is a terminal with command blocks. A classic terminal is a single stream: the output of npm test runs straight into the next prompt and the command after it. Finding the one error you care about means scrolling and selecting by hand. Blocks give every run a clear start and end.
What a command block holds
- The command you typed.
- Its output, with the colours the program printed.
- Where it ran: the working folder.
- How it went: running, finished or failed, and how long it took.
What you can do with a block
- Select it. Click a block to select it on its own; others stay as they are.
- Copy the part you need: the whole block, only the command, only the output, or the working folder.
- Filter it. Search inside one block's output, as plain text or a regular expression, with or without matching case.
- Open what it mentions. Paths and links in output are live; Cmd-click (Ctrl-click on Windows and Linux) opens them.
- Hand it to an agent. Attach a block as agent context, so the agent reads the exact output instead of a pasted copy.
- Save it as a workflow to reuse the command later.
Why blocks matter more with AI
Agents run many commands. Without blocks, a long agent run is a wall of text. With blocks, each command an agent runs is its own result, with its own status, so you can see which step failed and what it printed. And when you want help with an error, you pass the agent one block, not a screenful of unrelated lines.
How Mel does it
- Every command is a block, in every pane and tab.
- Right-click or ⋮ opens the block menu: Copy, Copy command, Copy output, Copy working directory, Save as workflow, Attach as agent context and the block filter.
- Full-screen programs still work. Vim, top, SSH and CLI coding agents get a live terminal grid, and when they exit the final screen becomes a normal block.
- Colours are kept from the program's output all the way to the screen.
- A team of agents works in the same window, and each command it runs is easy to inspect.
How the options compare
| Classic terminal | Terminal with command blocks (Mel) | |
|---|---|---|
| Output layout | One continuous stream | One block per command |
| Copy one command's output | Select by hand | One menu item |
| Search one result | Search the whole scroll | Filter inside the block |
| See which command failed | Read the scroll | Each block shows its status |
| Give output to an AI agent | Copy and paste | Attach the block |
FAQ
What is a terminal with command blocks? A terminal that groups each command with its output, folder and status into one block you can select, copy and filter on its own.
Does a block-based terminal change my shell? No. Your shell runs as usual. Blocks only change how commands and output are shown.
Do full-screen programs like vim work with blocks? Yes. Mel shows them as a live terminal grid, and when they exit their last screen becomes a block.
Is Mel's terminal free? Yes. Mel is free on Mac, Windows and Linux, and its agents are free with your own API key.
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.