> ## Documentation Index
> Fetch the complete documentation index at: https://docs.brightdata.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Fix scrapers with the Self-Healing tool
> Use the Bright Data Scraper Studio Self-Healing tool to fix broken scrapers and add or remove output fields from plain-language prompts in 1 click.
The Self-Healing tool is an AI-powered code refactor assistant in Bright Data Scraper Studio that rewrites parts of a scraper from a plain-language prompt. Use it when a target site changes and your scraper stops returning the expected data, or when you need to add or remove output fields without editing the JavaScript by hand.
**When to use it:** the scraper stops returning expected data, the price or title field returns `undefined`, or you need to add new fields (`price`, `image`, `rating`) without writing code.
## Prerequisites
* An active [Bright Data account](https://brightdata.com/)
* An existing scraper in Bright Data Scraper Studio (saved in development mode)
* Access to the Scraper Studio IDE
## How do I fix a scraper with the Self-Healing tool?
In the Bright Data Scraper Studio IDE, open the scraper you want to fix and locate the Self-Healing tool panel.
> **Expected result:** a text input field appears, ready to accept your instruction.
Type your request in plain language. Be specific about what is broken and which fields are affected.
```txt Example prompts theme={null}
Add 'price' and 'image' fields to the output
The 'price' value is returning 'undefined', please fix
The 'price' field is returning incorrect data from HTML, switch to using tag_response() to capture '/api/price' instead
```
> **Expected result:** the AI processes your prompt and produces a code diff in the editor.
Refactoring can take up to 15 minutes. You do not need to stay on the page: Bright Data sends an email when the diff is ready, and you can review it at your convenience.
Examine the AI-generated diff in the code template. Verify it matches your intent before accepting.
* **Accept:** saves the changes to a draft
* **Decline:** discards the suggestion; your original code remains unchanged
> **Expected result:** if accepted, the editor shows the updated code and saves it to the draft.
After accepting, run a preview to confirm the scraper returns the expected data.
> **Expected result:** the output file contains the new or fixed fields with valid, non-undefined values.
Once the preview looks correct, save the scraper to production.
If you added or renamed fields, Bright Data Scraper Studio prompts you to update the output schema. Click **Update Schema**, then **Save to Production**.
> **Expected result:** the refactored scraper is live and collecting data with the new configuration.
## Troubleshooting
| Problem | Likely cause | Resolution |
| ------------------------------------------ | ------------------------------ | -------------------------------------------------------------------------------------- |
| AI-generated code does not fix the issue | Prompt was too vague | Re-prompt with specific field names and an example error message |
| Preview still returns undefined values | Target site structure changed | Inspect the live page and include the expected HTML element or selector in your prompt |
| Changes do not appear after accepting | Browser cache issue | Refresh the IDE and re-check the development mode draft |
| The refactor runs for more than 15 minutes | Complex change or heavy prompt | Break the request into smaller prompts (one field at a time) |
## Frequently asked questions
Yes. The Self-Healing tool works on any scraper saved in development mode in Bright Data Scraper Studio, regardless of whether the scraper was built by the AI Agent, from a template, or written from scratch in the IDE.
Yes. Accepted changes go to a draft first; they only affect production after you click **Save to Production**. If you already saved to production, open the **Versions** menu on the scraper dashboard to roll back to an earlier version.
The Self-Healing tool can add or remove fields, but you must confirm the schema change before saving to production. Bright Data Scraper Studio prompts you with **Update Schema** when the new output no longer matches the current schema.
## Related
Build a new scraper from natural-language prompts
Build and edit scrapers directly in the IDE