Hi,
I have a long article (about 18,000 characters), and it’s already split into parts using a custom delimiter (---).
I want to build an n8n workflow that:
- Takes each block one by one, in order.
- Sends that block through a chain of 3–4 prompt-based OpenAI agent steps (for example: summarization → rewrite → keyword generation → style adaptation).
- Overwrites the original block in the article with the final result.
- Then moves on to the next block, and so on — until the entire article is processed.
The most important thing:
Each prompt chain must have access to the full article content as context, because the blocks are connected and rely on the surrounding meaning.
Can this be done in n8n?
If yes — any examples or tips on how to build this kind of loop logic would be appreciated.
Thanks.