How to use Agent View panes.

Live panes for work you want to watch and steer by hand.

• Layout auto-arranged grid
• Per pane one CLI session
• Providers Claude, Codex, Gemini, local
• Best for watch and steer

1. Open a pane — ask Toad with watch-and-drive language like spawn here.
2. Pick provider and model — one CLI session per pane.
3. Compare providers — same prompt, side-by-side panes.
4. Steer live — nudge, inspect, or stop mid-run.

agent view — four panes

Open as many panes as you need

• Ask Toad with watch-and-drive language: spawn here, give me a pane, let me drive.
• Delegation defaults to background agents — that vocabulary routes it into a visible pane instead.
• The grid auto-lays panes out; you never drag or resize.

Choose the provider and model

• Each pane runs one CLI session against one model.
• Claude, Codex, and Gemini ship as defaults; add local models in Accounts.
• The brain runs on almost any model — not the Gemini/Antigravity CLI, which stays agents-only.

Run the same prompt across providers

• Open one pane per provider; send the identical prompt to each.
• Every pane shows its live session — compare approaches side by side.
• Rather let Toad route the work? Use Studio.

Keep a standby pane ready

• Leave an idle pane for quick manual jobs — a one-off command, a lookup, a sanity check.
• It stays on the grid, ready the moment you need it.

Steer and review the live output

• Nudge the agent mid-run, inspect its work, or stop it early.
• Review the result in the pane, then promote, discard, or follow up.

Four ways to use a pane

Reach for a pane whenever watching beats walking away.