I gave it the prompt: Create a JSON file named "flows.json" that is a valid Node-RED flow file that can be imported into the Node-RED editor. The flow must contain three nodes. They are as follows, an "Inject" node, a "Function" node, and a "Debug" node. The nodes must be wired in that order. When "Inject" is pressed a msg object is sent containing a payload expected to be a string to the "Function" node. The "Function" node then saves the payload to an array as a flow context variable, and returns a msg object with a payload of the string and an additional property with the key "test" and the value of true if the string was not in the array beforehand otherwise false. ChatGPT's response: Here is an example of a valid flows.json file that meets the requirements you specified: [ { "id": "n1", "type": "inject", "z": "", "name": "", "topic": "", "payload": "", "payloadType": "str", "repeat": "", "crontab": "", "once": false, "onceDelay": 0.1, "x": 120, "y": 100, "wires": [ [ "n2" ] ] }, { "id": "n2", "type": "function", "z": "", "name": "", "func": "if (typeof context.get('myArray') === 'undefined') {\n context.set('myArray', []);\n}\n\nconst myArray = context.get('myArray');\nconst payload = typeof msg.payload === 'string' ? msg.payload : '';\nlet test = true;\n\nif (!myArray.includes(payload)) {\n myArray.push(payload);\n test = false;\n}\n\nreturn {\n payload,\n test\n};", "outputs": 1, "noerr": 0, "x": 300, "y": 100, "wires": [ [ "n3" ] ] }, { "id": "n3", "type": "debug", "z": "", "name": "", "active": true, "tosidebar": true, "console": false, "tostatus": false, "complete": "false", "x": 480, "y": 100, "wires": [] } ] The "Inject" node sends a message with a string payload to the "Function" node. The "Function" node then checks if a flow context variable called myArray exists. If it doesn't, it creates the variable and initializes it to an empty array. The node then adds the string from the msg.payload property to the array if it is not already in the array, and returns a new msg object with the original string as the payload property and an additional property called test that is set to true if the string was already in the array, and false otherwise. Finally, the "Debug" node receives the msg object and outputs it to the debug sidebar in the Node-RED editor. All in all, it did pretty well. I made some slight changes to the resulting function: // I manually replaced the word "context" with "flow" throughout if (typeof flow.get('myArray') === 'undefined') { flow.set('myArray', []); }; const myArray = flow.get('myArray'); const payload = typeof msg.payload === 'string' ? msg.payload : ''; let test = true; if (!myArray.includes(payload)) { myArray.push(payload); test = false; }; // This line was manually added to save the array flow.set('myArray', myArray); return { payload, test }; What a wild tool to play with.

I've been using ChatGPT as well. Pretty cool but often doesn't work. The best success case I had was when I asked it to create a function node that takes as input a string and to single out specific pieces of that string separated by a space that match a certain pattern. It produced an array with exactly that info. I then told it that there would always only be a single result and therefore instead of an array I needed a string. It did that well. Then I asked that the output should also include the original message, which it did too. Negatives: Often it's code doesn't run. Also when I ask it to create a flows.json it can't always be imported and when it can it often has flaws.

I believe the reason it may struggle is that Node-RED is quite new. When something is very new online it can be hard to find for example, search results about it on Google. I noticed when I had it generating package.json files that it would always have old dependencies. Here is one it generated for example: { "name": "my-node-red-electron-project", "version": "1.0.3", "description": "A full-stack Node-RED and Electron project for a senior developer", "main": "flows.js", "scripts": { "start": "node-red", "dev": "node-red -s settings.js", "electron": "electron ." }, "dependencies": { "node-red": "^0.21.1", "node-red-node-email": "^0.3.6", "node-red-node-mongodb": "^0.2.4", "node-red-node-socketio": "^0.1.9", "electron": "^8.2.0" }, "devDependencies": { "eslint": "^6.1.0" } } I think this is a glimpse into ChatGPT's data-set, in a way revealing why it might struggle with Node-RED for now. I do find that the more direct and specific you are with your prompts, the better the responses will be. I guess it's no surprise with these tools you'll experience "bad data in, bad data out" kind of scenarios. It does a great job getting you 90% of the way there most of the time if you really think about your inputs. I don't mind filling in the last 10% and troubleshooting.

I would agree with you, but the problem is that the flows.json format is very difficult to debug. So if the ChatGPT output can't be imported at all into Node Red it isn't very helpful. If it can be imported but there are some flaws in the flow then I'd agree with you.

Just tested it out on JSONata [image]

It's aware of its own speech pattern! [image] cough! cough! SKYNET cough!

Just to let people know OpenAIs - ChatGPT (Chat Bot)is now closed for new users max users reached for some days at least

I've not been able to use it for the past week - just says error

I've just tried to login (I'm a registered user) - it took a few goes before it accepted me, but at least I can still access ChatGPT.

[image] HaroldPetersInskipp: I think this is a glimpse into ChatGPT's data-set I'm pretty sure that I read in their documentation that the dataset has nothing past 2021.

Node-RED

Node-RED Forum

ChatGPT does Node-RED

General

marcus-j-davies 12 January 2023 14:16 7

It's aware of its own speech pattern!

cough! cough! SKYNET cough!

ChatGpt and node red

Topic		Replies	Views
ChatGpt and node red General	15	1680	22 February 2023
Write Function Node Code with ChatGPT Share Your Nodes	21	1736	11 August 2023
[Announce] node-red-contrib-custom-chatgpt Share Your Nodes	6	728	10 March 2023
Lots of fun with ChatGPT recently General	9	304	18 November 2025
How far have you taken AI in Node-RED? General home-assistant , uibuilder	11	515	4 October 2026

Copyright OpenJS Foundation and Node-RED contributors. All rights reserved. The OpenJS Foundation has registered trademarks and uses trademarks. For a list of trademarks of the OpenJS Foundation, please see our Trademark Policy and Trademark List. Trademarks and logos not indicated on the list of OpenJS Foundation trademarks are trademarks™ or registered® trademarks of their respective holders. Use of them does not imply any affiliation with or endorsement by them.

ChatGPT does Node-RED

Related topics