Difference between using the “MCP Server Trigger” node and simply enabling “Available in MCP” in workflow settings

Questions
1

Hi everyone! :slight_smile:

I’m using n8n together with ChatGPT via MCP (Model Context Protocol), and I’m a bit confused about the difference between two ways of exposing a workflow to MCP.

What’s the difference between:

  1. Using an “MCP Server Trigger” node inside a workflow

  2. Simply enabling “Available in MCP” in the workflow settings

I have a chat pipeline that starts with a Chat Trigger node (“When chat message received”).
This workflow is used for RAG — retrieving information from my vector database and responding through the ChatGPT interface.

Right now I’m only enabling the “Available in MCP” toggle in the workflow settings (no MCP Server Trigger node inside the flow).

Through ChatGPT, when I select the MCP tool, I can send a message and the workflow executes correctly using the Chat Trigger. Everything works fine.

Since the workflow already works through MCP without the MCP Server Trigger node:

  • What exactly is the role of the MCP Server Trigger node?

  • What additional functionality does it provide compared to just enabling “Available in MCP”?

  • When should I use the MCP Server Trigger node, and when is the settings toggle enough?

I would really appreciate clarification on how these two approaches differ and in which situations each one should be used.

Thanks in advanced!