JSON error using code node after subworkflow execution

Questions
1

Describe the problem/error/question

After running a subworkflow I receive a json object. If i try to make a simple edit using node code, i get an error for the JSON output: Invalid output format [item 0] An output item contains the reserved key json. To get around this, please wrap each item in an object, under a key called json.
The only solution i found was to delete the metadata key from the item.
I believe this is not expected behaviour.

for (const item of $input.all()) {
delete item.metadata;
}
return $input.all();

Please share your workflow

Share the output returned by the last node

Invalid output format [item 0]
An output item contains the reserved key json. To get around this, please wrap each item in an object, under a key called json.

Information on your n8n setup

  • n8n version: 2.19.2
  • Database (default: SQLite): postgres
  • n8n EXECUTIONS_PROCESS setting (default: own, main): ?
  • Running n8n via (Docker, npm, n8n cloud, desktop app): docker
  • Operating system: ubuntu
2

Hey @jgggr, while you wait for a response, here are some things that might help:

Suggested resources

Automatically matched to your question.

Docs:

Forum:

@David_Streuli, @Matze202 - you’ve helped with similar issues before, can you take a look?

Automatically suggested by n8n’s community bot. It’s a pilot - please share feedback here.

3

Hi @jgggr
The Execute Sub-workflow node adds a top-level metadata key to every item it returns, which is what the “View sub-execution” link uses, and the Code node only allows json, binary, pairedItem and error as top-level keys on an output item, so return $input.all() gets rejected. Build the output items instead of returning the input ones:

return $input.all().map((item, i) => {
  const json = { ...item.json };
  if (json.agent_name === "example") json.agent_name = "other example";
  return { json, pairedItem: i };
});

pairedItem keeps item linking working for the nodes after it.

This was reported upstream and closed as stale, so building the items is the way around it for now:

5

Thanks for letting us know about this, We have created NODE-5788 as the internal dev ticket to look into it.

6

Thank you for your reply.

The solution workaround you describe I believe is the same thing as i’ve suggested (to delete the metadata key) right?

The GitHub issue you mentioned is spot on the problem I’ve described.

7

Same idea, and for the items that actually get cleaned the outcome is identical. The version in the workflow JSON you pasted isn’t equivalent though: delete item.metadata sits inside the case "example": branch, so only items whose agent_name matches get cleaned and every other item still carries metadata and trips the same error. Building the output item covers all of them without a branch, and it keeps working if the node ever returns another top-level key alongside metadata.