---
title: "Execution and single-shot resolution"
canonical: "https://support.appfire.com/space/RS/3166273540/Execution%20and%20single-shot%20resolution"
format: markdown
---
> Macro (aura-html)

After QueueZero generates an implementation plan, what happens next depends on the **Single-shot mode** setting.

- When **Single-shot mode is enabled**, QueueZero automatically continues from planning to execution and creates a pull request. This is the default behavior.
- When **Single-shot mode is disabled**, QueueZero pauses after planning so you can review the implementation plan. When you're ready, select **Execute plan** from the **Plan** tab to start execution.

See [Review and refine](https://appfire.atlassian.net/wiki/spaces/RS/pages/3158213589) for more information about reviewing an implementation plan before execution.

## Monitor the execution

Open the **Execution** tab to follow the progress of the resolution.

![Resolver Execution tab while an execution is in progress](media://c7acf069-6eb1-4723-9652-ad4b765a07b0)

While QueueZero is working, the tab shows the [implementation steps](https://appfire.atlassian.net/wiki/spaces/RS/pages/3158213589) and their current status. After execution is complete, it includes the results of the run, a link to the created pull request, and additional execution details.

> ℹ️ Execution progress can remain at 99% even when all visible steps appear complete. QueueZero reserves the final 1% until the run has fully finished.

### Progress overview

The progress overview shows the current state of the execution, including the step QueueZero is working on and the number of completed and pending steps.

### Execution events

The **Execution events** section shows the implementation steps QueueZero performs during the resolution.

Expand a step to review additional information about the actions QueueZero took.

### Execution timeline

The **Execution timeline** shows the results of the implementation steps and helps you review how the execution progressed.

### Execution data

The execution details include the **Execution ID** and the start and end times of the run.

You can use the Execution ID in the standalone app when reviewing detailed logs for the resolution.

## Code review

As part of the resolution, QueueZero performs a separate AI code review of the generated changes before completing the run.

The review can go through multiple iterations. If the reviewer identifies an issue, QueueZero can make changes and review the result again.

The code review section on the **Execution** tab can show the following states:

- **Reviewing** — The review is in progress. The current iteration is displayed.
- **Approved** — The reviewer did not identify any remaining issues.
- **Issues Found** — The reviewer identified issues that need attention. Findings can be classified as **Critical**, **Major**, or **Minor**.
- **Failed** — The review could not be completed.
- **Iteration limit reached** — The maximum number of review rounds was reached while some issues remained.

If the review reaches its iteration limit with unresolved findings, QueueZero can still create the pull request. Review the remaining findings carefully before merging the changes.

Code review is included as part of the execution progress shown on the **Execution** tab.

## Resolve CI failures

If the pull request created by QueueZero has failing CI checks, the work item displays a **CI Failed** status.

You can select **Attempt automatic CI fix** to ask QueueZero to investigate the failure and attempt a fix.

While QueueZero is working on the failure, you can use **Attempt to fix further CI errors** to allow additional fix attempts if subsequent CI checks also fail. The interface shows how many attempts remain.

CI fixes are available only when **CI auto-fix** is enabled for both the organization and the space.

### Review CI fix attempts

After QueueZero attempts to fix a CI failure, open the work item’s **Logs** tab to review what happened.

Each attempt appears in a group such as **Fix CI #1**, **Fix CI #2**, or **Fix CI #3**. Expand **Details** to view more information about the attempt.

The final message indicates the outcome:

| **Message** | **What it means** |
| --- | --- |
| **"Pushed CI fix to <branch>"** | QueueZero created and pushed a fix commit to the pull request branch. |
| **"Agent made no changes — CI still broken"** | QueueZero investigated the failure but did not make any changes. |
| **"CI fix ended without changes (not_fixable)"** | QueueZero determined that it should not make an automatic change. |
| **"CI fix ended without changes (already_attempted)"** | QueueZero recognized that it had already attempted to fix the same failure. |
| **"CI fix ended without changes (unreported)"** | The attempt ended without a result QueueZero could safely use or commit. |
| "… **reverted**. Nothing shipped — this failure needs human review." | QueueZero made changes but rejected them before anything was pushed. |

### Why a CI fix can be reverted

QueueZero does not push a proposed fix if it detects that the change would:

- modify tests or CI configuration instead of application code;
- empty a file;
- undo the changes introduced by the pull request.

In these cases, QueueZero reverts the proposed changes and the failure requires manual review.

For successful fixes, the log also includes a **Fix:** summary describing what changed. The same summary is included in the fix commit message on the pull request.

## Cancel an execution

At any point before the execution completes, you can cancel the run by clicking **Cancel execution** in the right side of the Execution tab. Canceling a run before a PR is merged does not count against your resolution limits, but partial runs still incur a cost and may result in blocked work until limits refresh. 

## Rate a completed resolution

After a resolution finishes, you can provide feedback from the bottom of the **Execution** tab.

Select:

- **Helpful** to give the resolution a thumbs up.
- **Not helpful** to give it a thumbs down.

You can also add an optional comment describing what worked well or what could be improved.

Feedback totals are included in analytics reporting and exports.