Queue messages while Vera works

You can write a follow-up while Vera is working. Press Enter to add it to the current conversation’s queue without interrupting the active turn.

When Vera finishes a tool step and is about to call the model again, waiting prompts join the running turn in order. Vera reads them before its next step and answers them without stopping its work.

A prompt queued under a different model or effort, or with an image the current model cannot read, waits for its own turn. If the turn ends before a prompt joins, Vera starts the oldest queued prompt after a successful reply and continues one prompt and one reply at a time. You can also send waiting prompts together before the current reply finishes.

Send queued prompts

>chart the route to the island

Reading src/map.ts for the reef markers.

Two routes miss the reef. The north one is shorter.

>take the north route

Updating the route in src/map.ts.

>

While Vera works, type a follow-up.

The turn keeps running.

The prompt is queued, not sent. The line above the composer shows it.

With the composer empty.

Vera stops the turn and sends the oldest queued prompt. The answer picks up from there.

To send several at once, queue another follow-up.

It waits in the queue.

And one more.

+1 means one more is waiting. Queued prompts join the turn after its next tool step, or run one at a time after each reply.

With the composer empty.

Vera stops the turn and sends the whole queue.

Both prompts arrive as separate messages, and Vera answers them together.

1 of 9

With the composer empty and the queue visible:

Sending the whole queue preserves prompt order as separate user messages and produces one response to the batch. Sending only the oldest starts a turn for it; the rest join that turn at its next tool step, or wait for its reply.

If the composer contains text or an attachment, Enter adds that draft to the queue instead of sending existing queued prompts.

Leave and return

The queue belongs to the conversation. Switching away or returning from another client preserves it while the conversation remains live.

If you stop a batch or it fails, Vera reports that result once. Prompts added after the batch began remain queued.