> ## 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.

# Auto-detect

> Use Octoparse AI to detect lists, fields, and page structure, generate a starting workflow, and refine extracted fields with natural language.

Auto-detect scans a web page, identifies repeated data patterns, and generates a starting extraction workflow — including data fields and pagination logic. It sits between [templates](/docs/en/platform/templates/overview) (fastest, but limited to supported sites) and the [no-code builder](/docs/en/platform/no-code-builder) (full manual control) in terms of effort and flexibility.

## When to use Auto-detect

Use Auto-detect when:

* The page has repeated items such as products, listings, reviews, or search results
* You want a quick starting workflow
* You are not sure which elements to select manually
* You want Octoparse to suggest fields and pagination logic
* You want to quickly preview what data can be extracted from a target website before committing to a full configuration
* You plan to review and adjust the generated workflow afterward

Auto-detect is a starting point, not a guarantee that every field or action will be correct.

## How Auto-detect works

Auto-detect analyzes the page DOM structure and visual layout, uses similarity calculation and feature combination to locate repeated data regions, and generates extraction rules automatically.

<Steps>
  <Step title="Open the target page">
    Start from the page that contains the data you want to extract.
  </Step>

  <Step title="Run Auto-detect">
    Octoparse scans the page, detects repeated data regions, and generates extraction fields. It also attempts to identify pagination or next-page buttons so the task can move through multiple result pages automatically.
  </Step>

  <Step title="Review detected fields">
    Check whether the suggested fields match the data you need.
  </Step>

  <Step title="Confirm page navigation">
    Review pagination, scrolling, or next-page actions. Auto-detect may identify a next-page button or infinite scroll pattern — verify that it works correctly before running at scale.
  </Step>

  <Step title="Test the task">
    Run a sample and verify the output before scaling the extraction.
  </Step>
</Steps>

## What to review after detection

After Auto-detect generates a workflow, check:

| Area | What to verify | How to fix |
| - | - | - |
| Fields | Are the correct values captured? | Re-select the element or adjust the XPath in the Data Preview panel |
| Field names | Are column names clear and meaningful? | Rename fields before export |
| Pagination | Does the task move to the next page correctly? | Manually set the next-page button or scroll action |
| Detail pages | Does the workflow open item details when needed? | Add a click action to enter detail pages |
| Duplicates | Are repeated or unwanted elements included? | Delete unwanted fields or rows in the Data Preview panel |
| Missing values | Are some rows missing important fields? | Add fields manually in the Data Preview panel |

## Known limitations

Auto-detect works best on pages with clear, repeated data structures. It may produce incomplete or inaccurate results in certain situations:

| Situation | What may happen |
| - | - |
| Non-standard or irregular page layout | Fields may be misidentified or missing |
| JavaScript-rendered dynamic content | Data that loads after the initial page render may not be detected |
| Anti-scraping websites (e.g. Indeed, LinkedIn) | Auto-detect may fail to load or parse the page |
| Complex nested structures | The detected list may include incorrect or duplicate elements |
| Wrong field captured | The system may pick up promotional text instead of the actual value (e.g. a discount label instead of the real price) |

A common runtime error after Auto-detect is **"\[Loop Item] failed to find current loop item, exiting loop now"**, which usually means the generated XPath no longer matches the actual page elements. Re-running Auto-detect or manually adjusting the selector in the no-code builder typically resolves this.

## When manual editing is needed

Manual adjustments may be needed when:

* The page layout is irregular
* Important fields are outside the detected list
* The website loads content dynamically
* Pagination is not detected correctly
* The page requires login, filters, popups, or user interaction
* You need to extract data from detail pages

Use the no-code builder to refine the generated workflow.

If the workflow is generally correct and only its extraction fields need adjustment, use [Field Fixer](/docs/en/platform/field-fixer) to describe the desired result in natural language and review the proposed change before applying it.

<Warning>
  Do not assume Auto-detect output is production-ready. Always review fields, run a sample, and check the exported data.
</Warning>

## Related pages

<Columns cols={2}>
  <Card title="Templates" href="/docs/en/platform/templates/overview">
    Start with a prebuilt workflow for common websites.
  </Card>

  <Card title="No-code builder" href="/docs/en/platform/no-code-builder">
    Adjust or build workflows manually after Auto-detect.
  </Card>

  <Card title="Field Fixer" href="/docs/en/platform/field-fixer">
    Review and adjust Auto-detected fields with natural language.
  </Card>

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


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.