Skip to content

Commander View

Commander View is the power-user’s battlefield. Drop the map, drop the cards, and see every agent’s live terminal laid out in a grid. When you’re driving a team through a messy migration or a multi-front investigation, this is where the keyboard never leaves home row.

Commander View with four agents in parallel

How to open it

Click Commander View from the floating action bar (the rocket icon on the right edge), or press Alt + 2 to cycle. The dedicated button is always visible so you can snap back from anywhere.

On a fresh install the grid is empty until you spawn an agent.

Commander View, empty

What it gives you

Every terminal, one screen

One tile per agent. Live stdout streaming with the same tool-call formatting as the inspector. Tiles scale to fit the viewport — fewer agents, bigger tiles.

Type into any of them

Click into a tile’s input row and type. Enter dispatches to that agent’s CLI process exactly as if you were in its inspector.

Bulk commands

Multi-select tiles with Shift-click, type once, send to all. Every selected agent receives the same prompt in parallel.

Grouped by area

Tiles cluster by the area each agent belongs to, so teams working on the same codebase stay visually adjacent.

Typical workflow

  1. Spawn three or four agents pointed at the same repo (or different repos).
  2. Open Commander View; the grid is already populated.
  3. Pick the primary agent and use it as your driver.
  4. When a side-question comes up (“check how the old auth code worked”), type it into the tile of the agent best suited to investigate.
  5. When all the tiles’ tracking-status pills flip to need-review, triage one tile at a time and merge or discard what they produced.

Hotkeys that matter

  • Click tile header — open that agent’s full inspector in a side panel.
  • Shift-click — multi-select tiles for bulk send.
  • Ctrl/Cmd + K — Spotlight search jumps focus directly to any tile by agent name.
  • Esc — deselect all tiles; returns to passive viewing.

Gotchas