How to pass data through a node without it being processed?

Questions
1

I think this is an extremely basic function that I can’t find on n8n: how to pass secondary data through a node, without it being processed?

Example of a concrete situation: the trigger receives several data (the photo of an expense report + a short explanatory text + a phone number).

The first node that I position in output of the trigger is an OCR node (Tesseract: perfect!). It extracts the data from the photo (the “main” data in this node) then sends it to the AI ​​agent that follows and will process it (to retain only the essential data and format it in a usable Json).

But in the meantime, what happened to the “secondary data” (text accompanying the photo and the phone number)? This data remained on the trigger. It would have been so much simpler if the nodes had integrated by default an option “pass the data without processing it”. What I would have activated on the OCR node for example.

Instead, you have to tinker with other nodes, like “merge” that you will put above and that will retrieve the data from the trigger and the OCR. But if it were just that, it would still be fine: the output of this “merge” generates a big indigestible Json in which the AI ​​can no longer find its little ones (no prompt, however refined it may be - generated by AIs - has managed to correctly extract the Json output of the “merge”).

Perplexity told me about a feature called “passthroughtData”. But it is unable to tell me how to execute it or find it.

Unless I missed something?

If anyone has an idea?

In my workflow, I would need to pass secondary data through nodes almost all the time without it being processed or lost. Wouldn’t this be a good option to include in a future update?

Information on your n8n setup

  • n8n version: 1.79.3
  • Database (default: SQLite):
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • Running n8n via Docker:
  • Operating system:
2

It looks like your topic is missing some important information. Could you provide the following if applicable.

  • n8n version:
  • Database (default: SQLite):
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • Running n8n via (Docker, npm, n8n cloud, desktop app):
  • Operating system:
3

hello @ThierryESPASA

the simplest thing is to split your input data into 3 parts (OCR, text, phone), then do whatever you need to prepare them and merge them into one item in the end, to pass the entire object to the AI.

there is no “passthrough” because dozens of nodes don’t expect any “garbage” in the payload they are working with. However, there are a bunch of options for how to combine payload from different nodes into one (code node, merge node, edit fields node)

1 Like
4

This is the best idea I’ve been offered so far. Thanks. But it’s still “tinkering” when a node option could have transferred this data. But thanks for the tip: it’ll save me.

5

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.