Day 15: How you can turn your LLM to an n8n Co-pilot

Hi, this is G again! And this Day 15 of my build log.

I just want to drop this quick build log for anyone trying to level up their n8n workflows — or feeling stuck understanding someone else’s.

Lately, I’ve been using ChatGPT not just as a coding assistant, but as a workflow co-pilot. But you can use claude or gemini.

Here’s what that looks like:

I drop my exported workflow JSON into the chat (or describe it in steps)
t helps me analyze the structure, JSON flow, node logic, and expressions
It helps me debug errors I don’t fully understand yet
It helps me brainstorm better solutions or refactor messy parts
It teaches me why something works — not just how to fix it

I’ve even built a prompt framework to guide the process like a phased consultation. From clarification, to analysis, to breakdowns of JS nodes, to proposing multiple solutions — all with explanations tied back to JSON and core n8n logic.

The goal? I don’t just fix things faster — I learn deeper. Coz when you understand, you can build better.

This might be overkill for some. Hehe But if you’re a solo builder trying to understand workflows you didn’t build, or grow your skills without needing a dev mentor, this approach has been key 🔑.

If anyone here wants to try it, I added the systems prompt structure I use in chatgpt projects.

HOPE THIS IS USEFUL FOR SOME OF YOU! Happy to share and walk you through it.

CHEERIOS,

0 comments