Resolving Tools Name Conflicts with Multiple MCP Node in AI Agent

Questions
,
1

Hi,

I’m currently working with the AI Agent in n8n and want to use multiple MCP nodes as tools for the agent.

I have two MCP nodes that, because they are instances of the same MCP Server type, inherently share the same underlying tool name from the AI Agent’s perspective. Although I’ve configured them with different custom names in the n8n interface, different SSE Endpoints, and distinct settings within their respective MCP Servers, I encounter the following error when I run the workflow:

You have multiple tools with the same name: 'xxxx', please rename them to avoid conflicts

It appears the AI Agent is identifying them as the same tool despite their different configurations.

Could anyone advise on how to resolve this issue? Is there a way to properly differentiate these MCP nodes so the AI Agent can recognize them as distinct tools?

Thanks in advance for your help!

2

Take a look at the configuration of your agents to the mcp client. Each mcp server only needs one mcp client regardless of the number of tools it has.

connecting multiple mcp clients to the agent, which use the same server gets interpreted as a duplicate of the tool

3

Thanks for your reply!

I’d like to clarify my setup a bit further, as I think there might be a slight misunderstanding.

You’re right that a single MCP server instance generally only needs one MCP client. However, in my scenario, I am running multiple instances of the same type of MCP server – specifically, mcp-grafana. Each mcp-grafana server instance is configured differently because each connects to a distinct Grafana service endpoint (e.g., one for a development Grafana, another for a production Grafana).

Because I have these separate mcp-grafana server instances (each with its unique configuration and Grafana target), I do need to configure a corresponding MCP Client node in n8n for each one to make them available as tools to the AI Agent.

The core issue I’m facing is that even though these MCP Client nodes connect to different mcp-grafana server instances (with different settings), they are still identified by the n8n AI Agent as having the same underlying tool name, leading to the conflict: You have multiple tools with the same name: 'xxxx'. This is because they are all of the mcp-grafanatype.

For context, in other environments like Cursor or Windsurf, it’s possible to configure multiple instances of the same MCP server type, and they can be differentiated, often by a custom name defined in the configuration, which then allows them to be treated as distinct tools. For example:

{
  "mcpServers": {
      "dev-grafana": { // Custom name differentiates this mcp-grafana instance
          "url": "http://localhost:7823/sse"
      },
      "prod-grafana": { // Custom name differentiates this mcp-grafana instance
          "url": "http://localhost:7824/sse"
      }
  }
}

Is there a similar way in n8n to differentiate these MCP client nodes when they connect to different instances of the same MCP server type, so the AI Agent can treat them as unique tools despite originating from the same base mcp-grafana tool?

4

Can someone help me?