whisper

Idea: A physical "Enter console" for multi-agent AI coding workflows

With tools like Claude Code, Codex, Cursor, and other terminal AI agents, I often run multiple sessions in parallel.

The problem is that these tools frequently pause and wait for human confirmation:

  • "Continue?"
  • "Approve this step?"
  • "Press Enter to proceed"

When you have 4–6 agents running, you end up constantly switching between terminal panes just to press Enter.

So I started thinking about a simple hardware idea.

Continue Deck — a physical Enter key for multi-agent AI workflows

Continue Deck

A modular physical console made of small buttons.

Each button maps to a terminal pane and simply sends Enter.

Example —

Screen layout:

A   B
C   D

Desk layout:

[ A ] [ B ]
[ C ] [ D ]

So when Agent C pauses, you just press the bottom-left button.

No window switching. Just spatial memory.

Design idea

  • Small modular buttons
  • Magnetic connection
  • Each button = Enter
  • Layout matches terminal layout

Optional features:

  • LED status (running / waiting / error)
  • Works with tmux / Warp / iTerm
  • USB HID keyboard

Why keep it simple?

I deliberately did not design a full AI control panel.

No menus. No configuration UI.

Just Enter.

Because no matter how AI tools evolve, human approval will probably remain part of the loop.

Curious if other people running multiple AI agents feel this pain.

I don't have the hardware skills to build this, but wondering if something like this already exists — or if anyone would want to make one.

← Back to home