MCP Server in n8n stops after first trigger even after making it active

Hey,
I created an MCP Server and MCP Client with an agent. The flow works like this: MCP Client calls MCP Server → MCP Server triggers → does the action → returns the data.

The issue:
When the MCP Server gets triggered once, the workflow stops, even if it’s active. If I want to trigger it again, I have to manually run it or make it active again.

On top of that, MCP Server in n8n works like a trigger node (activated via URL), and in n8n trigger nodes can’t have previous nodes.
So I can’t connect anything before it for pre-processing — and that’s a big limitation.

so if i have to manually start the server workflow then it will be worst and not at all suitable for production

Hi @Microlent ,

welcome to the community

It sounds like the issue is that you’re using the test URL for the MCP Server in n8n, which only works when the workflow is manually executed and stops after one trigger. For production use, you’ll need to activate the workflow and use the production URL, which stays live and triggers automatically. If you still face the same problem on the production URL, it could be due to misconfigured webhook settings or conflicts, and you may need to adjust your setup

I was using Test url that was the issue

Thank You @Msquare_Automation

Hi @Microlent

Glad to know that your issue is resolved now.We would really appreciate if you mark the answer as solution if it helped you. This will keep the community healthy and organized.