Importing Blueprints: Pasting into scenario editor; Display invalid JSON errors

Questions
,
1

:bullseye: What is your goal?

Is there a way to import a blueprint just by pasting in the JSON. This would be very useful when AI generates a blueprint for me.

:thinking: What is the problem & what have you tried?

A related problem that I have with importing is that there is no feedback when an error is detected. If I import that same blueprint with errors via the API then I DO get told what the errors are.

3

No. However, you can paste a partial scenario flow, if that helps.

You can use my Make Blueprint Scrubber tool to convert a valid Make blueprint export into a pasteable module subflow.

Just toggle the “Convert to Subflow” option ON.

Another excellent suggestion! There is no such feature in Make yet.

If this is important to you, please submit this suggestion to the Idea exchange, under Platform Ideas and Improvements. However, do search for a similar request first (and also upvote), just in case it has already been suggested to avoid duplicates.

— @samliew

4

Thanks Sam. I must be doing something silly because pasting your subflow JSON into a scenario isn’t working. Nothing happens.

6

Probably because the scenario blueprint is invalid in the first place…

Care to attach it here?

7

Thanks. This is the Gemini 3 generated blueprint before I use your Make Blueprint Scrubber to convert it into a subflow (which it did without complaint).

{
  "name": "Salesforce Mismatch Reporter",
  "flow": [
    {
      "id": 1,
      "module": "salesforce:SearchRecords",
      "version": 1,
      "parameters": {
        "object": "Contact",
        "fields": [
          "Id",
          "Name",
          "Field_A__c",
          "Field_B__c"
        ],
        "limit": 100
      },
      "mapper": {},
      "metadata": {}
    },
    {
      "id": 2,
      "module": "builtin:ArrayAggregator",
      "version": 1,
      "parameters": {
        "feeder": 1
      },
      "filter": {
        "name": "Check for Mismatch",
        "conditions": [
          [
            {
              "a": "{{1.Field_A__c}}",
              "o": "text:not_equal",
              "b": "{{1.Field_B__c}}"
            }
          ]
        ]
      },
      "mapper": {
        "Name": "{{1.Name}}",
        "ValueA": "{{1.Field_A__c}}",
        "ValueB": "{{1.Field_B__c}}"
      }
    },
    {
      "id": 3,
      "module": "builtin:TextAggregator",
      "version": 1,
      "parameters": {
        "feeder": 2,
        "rowSeparator": "\\n"
      },
      "mapper": {
        "text": "Record: {{2.Name}} | Field A: {{2.ValueA}} | Field B: {{2.ValueB}}"
      }
    },
    {
      "id": 4,
      "module": "google-email:sendEmail",
      "version": 1,
      "parameters": {},
      "mapper": {
        "to": [
          "[your-email@example.com](mailto:your-email@example.com)"
        ],
        "subject": "Salesforce Data Mismatch Report",
        "content": "The following records have mismatched fields:\\n\\n{{3.text}}"
      }
    }
  ]
}
8

That blueprint is invalid, and cannot be imported as a scenario blueprint. You have to get Gemini to generate a valid Make blueprint (not just valid JSON).

9

Thanks. The prompt context was make.com and that it what it was asked to do…but I’ll ask it to try harder :wink:

10

Models used by Gemini (or other AI LLMs), are not trained on Make, so I don’t think you can just ask it to produce a Make blueprint. Let me know if you manage to successfully generate an importable scenario blueprint in future.

— @samliew

11

Gemini 3 is pretty good at reading an exported blueprint from make.com , explaining the logic and diagnosing bugs. But it seems not so good on producing a valid modified blueprint. But I will let you know if I have success.

1 Like
12

Yeah reading is easy, its just a JSON file and all of the modules out there can do that. But when you generate it you have to follow a strict structure that its usable by Make.

Unless a model is trained on creating that structure, its a monkeys with a typewriter type of situation. If you prompt it enough times its bound to get it right eventually, but it may take a while.

1 Like