ayrisdev/crawl4ai
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Warn LLM against hashed/generated CSS class names in schema prompts

Browse Source 
Replace vague "handle dynamic class names appropriately" with explicit
rule: never use auto-generated class names (.styles_card__xK9r2, etc.)
as they break on every site rebuild. Prefer data-* attributes, semantic
tags, ARIA attributes, and stable meaningful class names instead.
This commit is contained in:
unclecode
2026-02-17 12:02:58 +00:00
parent d267c650cb
commit 4fb02f8b50
1 changed files with 6 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
crawl4ai/prompts.py
View File

@@ -348,8 +348,9 @@ When repeating items are siblings (e.g. table rows, flat divs):
- Include prices, dates, titles, and other common data types - Include prices, dates, titles, and other common data types
3. Always: 3. Always:
- Use reliable CSS selectors - Use reliable CSS selectors that will survive page rebuilds
- Handle dynamic class names appropriately - NEVER use auto-generated or hashed class names (e.g. `.styles_card__xK9r2`, `.css-1a2b3c`, `.sc-bdnxRM`). These are generated by CSS-in-JS tools and change on every build — your schema will break on the next crawl.
- PREFER: `data-*` attributes (`[data-testid="review"]`), semantic tags (`article`, `section`, `nav`), ARIA attributes (`[role="listitem"]`), and stable class names that reflect meaning (`.product-card`, `.review`).
- Create descriptive field names - Create descriptive field names
- Follow consistent naming conventions - Follow consistent naming conventions
</behavior_rules> </behavior_rules>
@@ -811,8 +812,9 @@ When repeating items are siblings (e.g. table rows, flat divs):
- Include prices, dates, titles, and other common data types - Include prices, dates, titles, and other common data types
3. Always: 3. Always:
- Use reliable XPath selectors - Use reliable XPath selectors that will survive page rebuilds
- Handle dynamic element IDs appropriately - NEVER use auto-generated or hashed class names (e.g. `styles_card__xK9r2`, `css-1a2b3c`, `sc-bdnxRM`). These are generated by CSS-in-JS tools and change on every build — your schema will break on the next crawl.
- PREFER: `data-*` attributes (`@data-testid='review'`), semantic tags (`article`, `section`, `nav`), ARIA attributes (`@role='listitem'`), and stable class names that reflect meaning (`product-card`, `review`).
- Create descriptive field names - Create descriptive field names
- Follow consistent naming conventions - Follow consistent naming conventions
</behavior_rules> </behavior_rules>