Add a simple Pause / Resume button for running Agent tasks[FEATURE]:

Feature hasn't been suggested before.

• I have verified this feature I'm about to request hasn't been suggested before.

Describe the enhancement you want to request

I would like OpenCode to have a simple Pause / Resume feature for a running Agent task.

Currently, when an Agent is working on a long task and I need it to stop temporarily, I have to use Stop.

Then, when I want it to continue, I have to manually send a message such as:

continue

This is not ideal.

Current problem

After pressing Stop, I am never completely sure what will happen when I send continue.

For example:

Will the Agent continue exactly from the point where it stopped?

Will it repeat the previous step?

Will it restart part of the analysis?

Will it correctly remember what it was doing before I stopped it?

Will it rerun commands or checks that were already completed?

Also, sending continue requires another model interaction and takes additional time before the Agent understands the situation and starts working again.

For long development tasks, this creates unnecessary uncertainty.

What I would like

A simple control:

Pause

and then:

Resume

Pause

When I click Pause, OpenCode should temporarily stop the Agent while keeping its current task/session state.

The UI could simply show:

Paused

Resume

When I click Resume, the Agent should continue the same task from the point where it was paused.

I should not have to type:

continue

or explain again what the Agent was doing.

The important part is that OpenCode itself should know exactly where execution was paused and resume from that point.

Example

The Agent may be doing:

Repository audit
→ inspect files
→ run tests
→ analyze failures
→ fix issue
→ run tests again
→ final verification

If I pause while it is around:

analyze failures

I want:

Pause
→ Agent stops temporarily
→ Resume
→ Agent continues from the same workflow

I do not want:

Stop
→ manually type continue
→ wait for another model response
→ hope the Agent correctly determines where it stopped

Why this would be useful

Sometimes I only need to temporarily stop an Agent because:

I want to do something else on the computer.

I want to pause the work for a few minutes.

I want to inspect what it has done so far.

I need to temporarily stop the Agent without ending the task.

I want to resume later without having to prompt the model again.

This is especially useful for long-running coding and repository tasks.

Suggested UI

While running:

⏸ Pause
■ Stop

After pressing Pause:

▶ Resume
■ Stop

That is all I really need.

Expected behavior

The key requirement is:

Pause should preserve the current execution/task position, and Resume should continue from that exact point without requiring the user to send another prompt such as continue.

This would make long-running OpenCode Agent sessions much easier and more predictable to control.