@Nate Herk i believe I paid an entire year in advance- does this effect that?

Even better use a set node in front of the HTTP node then you wont have a bunch of differently configured HTTP nodes in all your workflows... Example with Set + curl-powered HTTP Request Node 1. Set Node { "url": "https://api.example.com/data", "token": "Bearer your_api_token_here", "query": "?type=agent&limit=10" } 2. HTTP Request Node (Curl mode) Switch to Curl mode. Paste in this command, using n8n expressions: curl -s -H "Authorization: {{$json.token}}" \ -H "Accept: application/json" \ "{{$json.url}}{{$json.query}}" n8n will evaluate {{$json.token}}, {{$json.url}}, and {{$json.query}} at runtime using the output from the Set node.

@Djibrile Camara thanks

@Khalid El Mansouri thanks - I totally forgot about that - guess I can see if any try to use now that it is disabled.

@Nate Herk Like the sequential qualities of this workflow - For consideration would be a workflow component that assesses the readability of a section using something like the Flesch-Kincaid Reading Ease review at the section level and if that returns a lower score maybe consider adding a graphic to help with understanding things that might be difficult to understand or just to break things up a bit. Second thing I have thought of was to have a KAC or Devils Advocate workflow that intentionally challenges parts of the report and then force the flow to find different sources for the same section than have an editor type of workflow to do a comparative analysis to see if the same assumptions and statements align across both. In grad school I had to work on peer reviewed journals where we had to support our findings so that is how I would go about improving - I say this because in many cases to cost is not nearly important as the accuracy of the report. If we have workflows as peer reviewers then that would be a boon. I say this as I have used Elicit, Google Notebook and tried others and quite often find where statements are made and they sound good but those are not supported in fact or in the reference. Like Elicit just runs up tokens and spits out garbage if it has trouble finding content on your question.

thank you