vodorod/Dorod-Sky
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Pedro.loop without explicit extraction (#3051)

Browse Source 
Co-authored-by: Shuchang Zheng <shu@skyvern.com>
This commit is contained in:
PHSB
2025-07-29 15:02:40 -07:00
committed by GitHub
parent c338c97ea7
commit 083d54f018
3 changed files with 230 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
skyvern/forge/prompts/skyvern/extraction_prompt_for_nat_language_loops.j2 Normal file
View File

@@ -0,0 +1,12 @@
Analyze the current webpage and extract information based on this request: {{ natural_language_prompt }}
You need to identify what values should be iterated over (loop_values). Each value should be the primary data that will be used in each loop iteration.
For example:
- If the request is "go to each product page", extract product URLs as strings
- If the request is "extract the authors of the top 4 posts", extract author names as strings
- If the request is "summarize the text of each article", extract article URLs as strings
- If the request is "download each file", extract file URLs as strings
- If the request is "check if these articles are AI-related", extract article titles as strings
Return the results in the specified schema format with loop_values containing an array of strings, where each string is the primary value to be used in the loop iteration.