AI Productivity

Wait for a Terminal to Finish

With orchestration on, Agent Input understands `wait for terminal <id> to finish, then implement the findings` — your follow-up is held until that terminal's turn has really ended, then sent for you. Chain several dependencies, address one by its Codex or Claude session id, and cancel any pending wait from Mission Control.

Screenshot coming soon

Why You Need This

How this feature boosts your daily productivity

Running two agents means being the message bus between them: you write the follow-up, then sit watching a spinner so you can paste it the moment the first one stops. The prompt is ready long before you are allowed to send it, and the only thing standing between the two is your attention. A passive wait removes you from that loop. Nothing is sent upstream while it waits — the agent you are waiting on is never nudged, never asked whether it is done, never has its context disturbed by a check-in; 1DevTool listens to the completion signal the agent itself emits and does nothing until it fires. That is the difference between this and a polling script, which has to touch the thing it is measuring. Waits are bounded and memory-only, so a restart drops them rather than firing a stale prompt into a session that has moved on, and an uncertain delivery is never retried. They refuse the cases that cannot mean anything — a terminal waiting on itself, a duplicate, a cycle — so you get told rather than getting a queue that silently never drains.

How It Works

Get started in 4 simple steps

1
1

Turn orchestration on

Passive waits live behind orchestration. With it enabled, Agent Input reads a dependency written in front of your prompt instead of sending straight away.

2
2

Write the dependency, then the work

Type `wait for terminal 3 to finish, then implement the findings`. The prompt is armed as a pending wait rather than delivered, and the upstream agent is never touched.

3
3

Chain or name the dependency

Wait on several at once — `wait for terminal a and terminal b to finish, then run the integration checks` — or address one by its Codex or Claude session id when exactly one open terminal owns it.

4
4

Watch it in Mission Control

Each pending wait is listed with what it is waiting on and a Cancel wait button. Copy original prompt gives your text back if you cancel.

5
5

Let it fire on real completion

The follow-up is delivered once every dependency’s turn has actually finished. Waits survive pane remounts and project switches, and refuse self-waits, duplicates and cycles.

Without 1DevTool

The manual way vs. the 1DevTool way

The Hard Way

# agent A is refactoring the data layer
# the next prompt is already written
# ...but B would read a half-rewritten repo
# so: watch the spinner
# check back every few minutes
# finally stopped -> paste the prompt by hand

The 1DevTool Way

wait for terminal 3 to finish, then implement the findings
Armed as a pending wait -- nothing sent upstream
Fires on the agent's own completion signal
Mission Control: Cancel wait / Copy original prompt

Key Capabilities

Everything included with Wait for a Terminal to Finish

Queue a prompt that sends itself when another agent finishes
Write the dependency in plain language in Agent Input
Chain several terminals so the follow-up waits for all of them
Address a dependency by its Codex or Claude session id
Leave the upstream agent completely untouched while you wait
See every pending wait and what it is waiting on in Mission Control
Cancel a wait and get your original prompt back
Keep waits alive through pane remounts and project switches
Get refused on self-waits, duplicates and cycles instead of a stuck queue

Why It Matters

The productivity impact of having ai productivity built into your workspace

You Stop Watching Spinners

The prompt you already wrote goes out on the real completion signal, so the gap between one agent finishing and the next starting is not your attention span.

The Upstream Agent Is Never Touched

No polling, no check-in, no nudge — the waiting is done by listening, so the agent you depend on runs exactly as it would have alone.

Sequencing Is Stated Once

The order you had in your head becomes a sentence, including the two-dependency case where you would otherwise wait on the slower one by eye.

Cancelling Costs Nothing

Cancel wait plus Copy original prompt means changing your mind does not throw away the thing you wrote.

Ready to boost your workflow?

Download 1DevTool for free and experience Wait for a Terminal to Finish along with 294 other features — all in one app.