Agents or terminals
You can run agents in plain terminals. Skribly Agent builds on terminal multiplexers such as tmux and agent multiplexers such as herdr, then adds what sits around the agent. Here is what differs, and when a multiplexer is the better fit.
What plain terminals leave to you
- You watch each terminal to see which agent is waiting.
- You copy task descriptions from the tracker into a chat.
- You paste one agent's output into another.
- You remove
Co-Authored-Bytrailers nobody asked for.
Side by side
Scrollable table
| Topic | tmux and terminal multiplexers | Skribly Agent |
|---|---|---|
| Interface | Text interface in a terminal | Desktop app with a mouse, draggable dividers, dialogs and a command palette |
| After closing | A server keeps running, and you can detach and come back | Quitting ends the processes. The layout and native agent sessions come back |
| Workspaces | You name sessions and windows yourself | A workspace is a repository or a worktree, detected from the session directory |
| Beyond the terminal | None | Chat, tracker tasks, git and commits, pipelines, skills, audits and a browser |
| Scripting | tmux send-keys and capture-pane | The skribly command and an MCP server: start, send, wait for a state and read |
| Prefix | Configurable | Fixed Ctrl+B, plus ⌘D, ⌘⇧D and ⌘T on macOS |
What stays the same
- Agents run in real terminals with your login shell and your permissions.
- The layout of workspaces, tabs and panes and the
Ctrl+Bprefix come from terminal multiplexers. - Each agent keeps its own login and API keys.
When a multiplexer fits better
- You need processes to keep running after you close the window.
- You work only over SSH and want a text interface. Skribly Agent can open SSH panes, but the app itself runs on your computer.
- You want to change the prefix key.
Keep reading
- See all features
Grouped by job.
- Read the guide
From installation to a first commit.
- See the commands
CLI and MCP.