Especially when:
-
output is wrong
-
prompt looks fine
-
and logs don’t really tell you what broke
So I built a small tool to help with this:
https://ai-debug-helper–tfreedom67.replit.app/
You can:
-
paste user input + prompt + AI output
-
and it tells you:
-
what likely went wrong
-
why it failed
-
and suggests a better prompt
-
It’s super simple, but already helping me debug faster.
I’m looking for a few people building AI workflows (especially WhatsApp / automation use cases) to try it and give honest feedback.
If you’ve had issues with inconsistent or weird AI outputs in n8n, would love your thoughts.
What I really want to know:
-
Did it actually help you debug faster?
-
What’s missing for real-world use?
Happy to improve it based on your workflows 