Help on attaching a Remote MCP Server as 'MCP Client Tool' to the AI Agent node

Questions
1

Describe the problem/error/question

I created a simple email agent that connects to a Remote MCP server that exposes several tools. I can run the agent and it also answers, but it fails to do dynamic MCP tool calls, it always fails during the parameter mapping in the MCP Client Tool node.

Now my question, is this MCP Client Tool node even able to do dynamic tool invocations? Or is there a specific AI Node that I need to use for that?

I am used from LangGraph agent configs that I just let the agent know my Remote MCP server and the agent does the tool execution planning.

I did set the environment variable for executing tools:

  • N8N_COMMUNITY_PACKAGES_ALLOW_TOOL_USAGE = true

What is the error message (if any)?

MCP error -32602: Received unexpected input arguments: [toolCallId]; Please check your request

Please share your workflow

Share the output returned by the last node

McpError: MCP error -32602: Received unexpected input arguments: [toolCallId]; Please check your request at Client._onresponse (/usr/local/lib/node_modules/n8n/node_modules/.pnpm/@[email protected]/node_modules/@modelcontextprotocol/sdk/src/shared/protocol.ts:450:27) at StreamableHTTPClientTransport._transport.onmessage (/usr/local/lib/node_modules/n8n/node_modules/.pnpm/@[email protected]/node_modules/@modelcontextprotocol/sdk/src/shared/protocol.ts:295:22) at StreamableHTTPClientTransport.send (/usr/local/lib/node_modules/n8n/node_modules/.pnpm/@[email protected]/node_modules/@modelcontextprotocol/sdk/src/client/streamableHttp.ts:497:39) at processTicksAndRejections (node:internal/process/task_queues:105:5)

Information on your n8n setup

  • n8n version: 1.119.1
  • Database (default: SQLite): SQLite
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • Running n8n via (Docker, npm, n8n cloud, desktop app): Docker self hosted
  • Operating system: Docker
2

The MCP Client Tool node in n8n currently doesn’t support dynamic tool invocations - it’s designed for static, predefined tool calls rather than the dynamic agent-driven tool selection you’re used to from LangGraph.

For dynamic MCP tool usage, you’ll need to use the **AI Agent node** instead, which can automatically discover and invoke MCP tools based on the conversation context. Make sure to:

• Connect your MCP server directly to the AI Agent node’s tools section

• Remove the separate MCP Client Tool node from your workflow

• Let the AI Agent handle both the tool selection and execution automatically

The error you’re seeing (`toolCallId` parameter issue) is because the MCP Client Tool node expects manual parameter mapping, while your agent is trying to pass dynamic tool call parameters that the node doesn’t recognize.

1 Like
3

Actually, N8N fixed the MCP dynamic tool selection by fixing a bug.

1 Like
4

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.