Deep n8n Observability with OpenTelemetry + Langfuse
Hey n8n community!
Many of us who self-host n8n have been looking for deeper insights into our workflow executions. Building on the fantastic foundational work shared by @stuartjohnson and the valuable iterations from @muxxi in their original OpenTelemetry post, I’ve put together a proof-of-concept specifically for Langfuse.
This guide provides all the code and a step-by-step tutorial to instrument your self-hosted n8n instance to send detailed traces directly to Langfuse. This is especially powerful for anyone building AI workflows, as it maps n8n nodes to Langfuse observation types (generation, agent, tool, etc.).
The result? You get a detailed trace for every workflow run, showing each node as a child span, complete with metadata and I/O.
You can find all the details on Github:
I hope this helps others in the community get the detailed observability they need. But most of all, I hope this will inspire the N8N development team, we need this ASAP! 
Feedback and improvements are very welcome.
Happy building.
Cheers