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

# Field Fixer

> Use Field Fixer in the Data Preview to turn natural-language requests into reviewed changes to extraction fields.

Field Fixer lets you correct Auto-detected extraction fields with natural language. Tell Octoparse what the output should contain, review the proposed change and sample values, then decide whether to apply it.

Introduced in Octoparse 10.2, Field Fixer marks a step toward an agentic AI web scraper. It does more than explain how to fix a field: it understands the goal, inspects the current task, plans a field-level change, and prepares the result for your approval.

```text theme={null}
Describe the goal → review the proposed change → apply → test
```

## What Field Fixer does for you

Suppose Auto-detect creates a `Price` field, but it captures the crossed-out list price instead of the current selling price. You enter:

> Change Price to the discounted selling price, not the original price.

Field Fixer then works through the task with you:

1. **Identifies the target.** It connects your request to the existing `Price` field instead of treating the prompt as a general question.
2. **Inspects the current extraction.** It considers what `Price` currently extracts and the relevant elements on the loaded page.
3. **Plans the field change.** It determines that the field source needs to change while the surrounding list and workflow remain intact.
4. **Resolves uncertainty.** If several prices could match your description, it asks you to select the intended value on the page instead of guessing silently.
5. **Checks the result.** It tests the proposed field against sample items and presents the extracted values for review.
6. **Updates the task after approval.** The existing field changes only when you accept the proposal, after which Octoparse refreshes the Data Preview.

The result is not a tutorial telling you to edit XPath. It is a task change prepared by an agent and verified by you.

## How much work it removes

The improvement is clearest when Auto-detect has already built the right list and navigation, but the output columns need correction.

| Manual field repair                                    | With Field Fixer                                              |
| ------------------------------------------------------ | ------------------------------------------------------------- |
| Determine why the value is wrong                       | Describe the expected value                                   |
| Locate the correct element in the page                 | Confirm the element only when Field Fixer needs clarification |
| Decide whether to edit the selector or extraction type | Review the proposed field change                              |
| Edit XPath or field settings                           | Approve the proposal                                          |
| Refresh the preview and inspect the result             | Review sample values generated as part of the proposal        |
| Repeat the process if the first selector is wrong      | Ask for a revision without rebuilding the field manually      |

Field Fixer shifts field repair from **configuring how the scraper should extract** to **specifying and verifying what the dataset should contain**. It reduces the expertise needed for common field corrections, but it does not eliminate final testing.

## Changes you can request

| If Auto-detect...             | Ask Field Fixer to...                            |
| ----------------------------- | ------------------------------------------------ |
| Captures the list price       | `Use the discounted selling price instead.`      |
| Misses the product rating     | `Add the product rating as Rating.`              |
| Extracts link text            | `Extract the destination URL from Product link.` |
| Creates an unnecessary column | `Remove the Sponsored field.`                    |
| Generates an unclear name     | `Rename field_3 to Availability.`                |

Field Fixer can add, remove, rename, or adjust extraction fields while leaving the surrounding workflow intact.

## Fix a field

<Steps>
  <Step title="Create an Auto-detected task">
    Run [Auto-detect](/docs/en/platform/auto-detect), then review the fields and sample values in the Data Preview.
  </Step>

  <Step title="Open Field Fixer from Data Preview">
    Generate the workflow and Data Preview first. Then open Field Fixer from the Data Preview header or the ✨ control beside a field. The entry remains unavailable until an extraction workflow and preview exist.
  </Step>

  <Step title="Describe the expected result">
    State what the field should contain. Select the affected field first when possible. Be specific when the page contains similar values, such as a list price and a sale price.
  </Step>

  <Step title="Provide page context if requested">
    If the intended value is ambiguous, select the correct element on the webpage so Field Fixer can continue with the right target.
  </Step>

  <Step title="Review the proposal">
    Compare the proposed change and sample values with the webpage. Check several rows, including items with different layouts or missing values.
  </Step>

  <Step title="Apply and test">
    Apply the change only when the preview is correct. Run a small test before starting a full local or cloud run.
  </Step>
</Steps>

## Where Field Fixer stops

Field Fixer in Octoparse 10.2 focuses on fields in Auto-detected tasks. It does not create or repair an entire workflow from a prompt.

Use the [no-code builder](/docs/en/platform/no-code-builder) to change pagination, scrolling, detail-page clicks, login and form actions, wait conditions, loops, or the overall action sequence.

<Warning>
  AI-generated changes can be incorrect. Always review the sample output and test the task before running it at scale.
</Warning>

## Related pages

<Columns cols={3}>
  <Card title="Auto-detect" href="/docs/en/platform/auto-detect">
    Generate the initial extraction fields and workflow.
  </Card>

  <Card title="No-code builder" href="/docs/en/platform/no-code-builder">
    Configure page actions and complex workflow behavior.
  </Card>

  <Card title="Refine data" href="/docs/en/platform/refine-data">
    Clean and reformat extracted values.
  </Card>
</Columns>