> ## Documentation Index
> Fetch the complete documentation index at: https://docs.intellixent.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Debugging Runs

> Inspect individual flow executions to find and fix errors in your automations.

Every time a published flow runs, DialogBot records the execution as a **run**. You can inspect any run to see exactly what happened at each step — whether it succeeded, what data it received, and what it returned.

## Viewing a run

<Steps>
  <Step title="Open the Runs view">
    Go to the **Dashboard** and click **Runs** in the left sidebar.
  </Step>

  <Step title="Find the run">
    Locate the run you want to inspect. You can filter by flow name, date, or status.
  </Step>

  <Step title="Open the run details">
    Click the run to open it. The flow builder displays in **view-only mode**, showing the same steps you built, with execution status indicators on each step.
  </Step>

  <Step title="Inspect step output">
    Click any step to open the **Run Details** panel. This panel shows the exact **input** the step received and the **output** it produced (or the error it returned).
  </Step>
</Steps>

## Reading run status

Each step displays a status icon when you view a run:

| Icon | Meaning                                                            |
| ---- | ------------------------------------------------------------------ |
| ✅    | The step executed successfully.                                    |
| ❌    | The step failed. Click it to see the error message and input data. |

## Diagnosing failures

When a step shows ❌, open its **Run Details** panel and check:

* **Input** — confirm the data passed into the step was what you expected. A missing or malformed value from an upstream step is a common cause of failures.
* **Error message** — the error returned by the action or external service. This often points directly to the problem (for example, a missing required field or an authentication error).

<Tip>
  If the input data looks wrong, trace it back to the step that produced it. You may need to update how an upstream step maps its output, or re-test it to regenerate accurate sample data.
</Tip>

<Warning>
  Runs are recorded only for **published** flows. If you are testing a draft and a trigger fires, no run is recorded.
</Warning>