---
title: "Understand a verification result"
description: "A verification result shows whether a checked expectation about your code holds, where a failure occurs, and what needs attention. Use it to understand a bug and review what changed after an agent fixes it."
url: https://panolayer.com/docs/verification-results
type: EXPLANATION
updated: 2026-10-20
---

# Understand a verification result

A verification result shows whether a checked expectation about your code holds, where a failure occurs, and what needs attention. Use it to understand a bug and review what changed after an agent fixes it.

Last updated 2026-10-20

<a id="twitter-example"></a>

## Follow the Twitter example

This guide follows the post-deletion bug in Panolayer’s Twitter example. You’ll read the issue, follow its explanation back to the code, and understand the verification summary before deciding whether to merge.

To follow along, open **Twitter Clone** under **Try a sample app**. It opens as `twitter-starter`. For the full setup and fix walkthrough, see [Fix your first bug with Panolayer](/docs/first-bug). Screenshots show one recorded run; your issue order and counts may differ.

<a id="start-in-issues"></a>

## Start in Issues

When you open a codebase, the **Panolayer verification engine** starts running. It parses the codebase down to its functions, symbols, and dependencies, builds the visual architecture, and checks the code to identify bugs.

Open **Issues**. In this example, the run has finished and the status reads **Verification complete**. Issues that need investigation appear under **Needs attention**.

![Twitter example Issues page with Verification complete above Needs attention; the first issue concerns post deletion.](/docs/first-bug/02-verification-issues.png "Verification has finished, and the post-deletion issue is first in the list. Search for canDelete or components/Post.tsx if it appears elsewhere in your run.")

**Verification complete describes the run.** It means the run has finished; read the results to see whether it found failures.

<a id="read-the-issue"></a>

## Read the deletion issue

The example’s first issue reports that a condition can stop a confirmed deletion:

> The extra `canDelete` guard can prevent deletion despite user confirmation, contrary to the stated current-post deletion behavior.

A **rule** is an expectation about a function’s behavior: what it should return, what action it should perform, or what other code expects from it. Here, the expectation concerns deleting the current post after the user confirms.

Read the issue’s file, function, status, and explanation together:

 | What you see | What it tells you |
| --- | --- |
| `components/Post.tsx` | The file containing the function to investigate. |
| `handleDelete()` | The function whose behavior was checked. |
| **Failed** | The checked behavior did not meet the rule. |
| **Severity 3 · Functionality** | The issue’s severity and category help you prioritize investigation. |
| **Details** | The explanation and source location help you understand why the issue was reported. |

 Click **Details**. The explanation says the function can return early when `canDelete` is false, before completing the deletion described by the rule. The source reference points to the relevant code.

![Expanded post-deletion issue showing handleDelete, the early return when canDelete is false, and the source reference at line 22.](/docs/verification-results/issue-details.png "The issue connects a specific behavior to a specific function. Follow the source reference to read the code in context; line numbers can change.")

Check that the rule matches the intended behavior before deciding what to change. If the expectation is wrong, correct the rule. If the code is wrong, fix the code. A change can also affect callers in other files, so follow any affected callers shown in the result.

<a id="statuses"></a>

## What do the statuses mean?

Read the status alongside its explanation and the code it applies to.

 | Result | How to read it |
| --- | --- |
| Passed | The evaluated rule passed for the code checked. This describes that rule, rather than every possible behavior of the application. |
| Failed | The evaluated rule did not hold. Read the explanation and inspect the source to decide what needs fixing. |
| Pending or running | The result is not ready. Wait for the check to finish before drawing a conclusion. |
| Unavailable | The check could not finish. Read the accompanying message and resolve the reported problem before checking again. |
| Not checked | The expectation has not been evaluated. It is not a passing result. |

 If verification cannot finish, follow the reported error and see [Resolve setup and agent problems](/docs/troubleshooting).

<a id="fix-and-verify"></a>

## Fix the bug and let verification run again

On the deletion issue, click **Fix**, then select **Panolayer Agent**. Panolayer passes the issue context to the agent, so you don’t need to write a prompt explaining the bug.

Panolayer Agent is built in. You can also use another supported agent after installing it; see [Supported agents](/docs/agents).

The agent works in its own Git worktree, keeping its edits separate from your branch. Follow the task in **My Agents**, or click **Open** on the run card to inspect its activity and changes.

As the code changes, Panolayer checks the affected functions and connected code again, including callers in other files. This helps catch regressions beyond the lines the agent edited.

A **regression** is a new problem introduced by the change. If verification catches one during a fix, the agent can repair it and run verification again before requesting a merge. A task can therefore have several verification runs before it is ready for your review.

<a id="read-the-summary"></a>

## Read the verification summary before merging

When this run finishes, Panolayer shows **Panolayer Agent is ready to merge**. The dialog summarizes what verification checked and repaired during the task.

![Twitter fix merge dialog showing one failure repaired, 14 rules passed of 21, one verification run, four files changed, and seven pre-existing failures.](/docs/first-bug/07-review-merge.png "This run repaired one failure. Read the counts together with the health message and the separate pre-existing-failures line.")

 | Summary field | This run | What it means |
| --- | --- | --- |
| **Failures repaired** | 1 | The work repaired a failure that was already present before the task started. |
| **Rules passed** | 14 of 21 | Fourteen rules passed out of 21 on the changed code. Read this alongside the health message and the separate pre-existing-failures line. |
| **Verification runs** | 1 | One verification run reached a result during this work. This counts verification runs, rather than individual tests or rules. |
| **Files changed** | 4 | The task changed four files. Review the full change set, including any tests and dependency files. |

 The health message says **No new failures, warnings, or unchecked rules.** Read that message alongside the counts: it describes the result of this work, while the next line explains the failures that remain from before it started.

### What if “Regressions caught” appears?

Some runs also show **Regressions caught**. It reports new problems verification detected during the work, helping keep them out of the target branch. Read the latest result to see what still needs attention; the count describes catches during the work, rather than the number of unresolved failures now.

This field is not shown in this screenshot. The deletion example shows one repaired pre-existing failure; it does not demonstrate a regression caught during the fix.

<a id="remaining-failures"></a>

## Why can it be ready to merge when failures remain?

Panolayer compares the result with the **baseline**: the state of the codebase before the change. This separates problems introduced by the task from problems that were already there.

- **A new failure** was introduced by the change and needs investigation before you accept the work.
- **A pre-existing failure** was present before the task started. It can still matter, but this task did not introduce it.
- **A repaired failure** was present before the task and was fixed during the work.

In the screenshot, the dialog reports **7 pre-existing failures** that do not block this merge. The selected deletion issue was repaired, and those other issues remain separate work.

The repository’s issue list and an agent’s verification summary can cover **different scopes**. The issue list shows what needs attention in the codebase; the merge summary describes the agent’s work against its baseline. An agent’s result can report no new failures while the repository still has pre-existing issues.

You can assign a pre-existing issue as another task using **Fix**. That gives the agent a specific issue to work on next.

**Ready to merge does not mean every issue in the repository is resolved.** Read what this task repaired, what it introduced, and what was already there.

<a id="your-decision"></a>

## Decide what reaches your branch

You control the merge. Before accepting the result:

1. Check that the repaired issue matches the bug you asked the agent to fix.
2. Read the verification summary, including any new failures, warnings, or checks that could not finish.
3. Open **Details** for more context and inspect the agent’s changed files. Review the relevant tests and run the application when the behavior needs a runtime check.
4. Confirm the target branch. Choose **Merge into main** when you’re satisfied, or **Not now** to keep reviewing.

Verification gives you evidence for the decision. Read it alongside the code and relevant tests; it does not cover every possible bug.

## Continue learning

- [Fix your first bug with Panolayer](/docs/first-bug)

- [Review changes and commit](/docs/changes)

- [When verification runs](/docs/concepts/when-verification-runs)

- [Troubleshoot Panolayer: setup, agents, and verification](/docs/troubleshooting)
