n8n MCP Server with Google Calendar and Custom Functions

This n8n workflow demonstrates how to set up a Model Context Protocol (MCP) server that can interact with an AI Agent. It includes an example of using custom functions (tools) to extend the AI Agent's capabilities, specifically for interacting with Google Calendar.

What it does

This workflow acts as an MCP server, receiving requests from an MCP client and routing them to an AI Agent. The AI Agent is configured with a chat model, memory, and several tools, including a custom n8n workflow tool and an MCP client tool.

Here's a step-by-step breakdown:

1. Listens for MCP Requests: The workflow starts with an "MCP Server Trigger" node, which acts as the entry point for requests from an MCP client.
2. Initializes AI Agent: Upon receiving a request, an "AI Agent" node is activated. This agent is the core of the intelligent processing.
3. Configures AI Agent:
   • It uses an "OpenAI Chat Model" for its language understanding and generation capabilities.
   • A "Simple Memory" node provides conversational memory, allowing the AI Agent to remember previous interactions within a session.
   • It's equipped with two tools:
     • "Call n8n Workflow Tool": This tool allows the AI Agent to execute other n8n workflows as custom functions. This is where you would define specific actions the AI can take (e.g., interacting with Google Calendar).
     • "MCP Client Tool": This tool enables the AI Agent to potentially make requests to other MCP servers or even itself, allowing for complex, multi-agent interactions.
4. Processes AI Agent Output: The output from the "AI Agent" is then passed to an "Edit Fields (Set)" node, which is likely used to format or modify the AI's response before further processing.
5. Conditional Routing (Switch): A "Switch" node is present, suggesting that the workflow can route the AI Agent's response based on certain conditions. This allows for different actions depending on the AI's output.
6. Debug/Output: A "DebugHelper" node is included, which is useful for inspecting the data flow and debugging the workflow during development. An "HTTP Request" node is also present on an output path, indicating that the workflow might make external API calls based on the AI's decision.
7. External Workflow Execution: An "Execute Workflow Trigger" node is available, which could be used by the "Call n8n Workflow Tool" to trigger a specific sub-workflow (e.g., one that interacts with Google Calendar).
8. Chat Trigger: A "Chat Trigger" node is also present, which could be used as an alternative or additional entry point for chat-based interactions with the AI Agent, separate from the MCP server.

Prerequisites/Requirements

• n8n Instance: A running n8n instance (self-hosted or cloud).
• OpenAI API Key: For the "OpenAI Chat Model" to function.
• Google Calendar Account: If you intend to implement the Google Calendar integration using the "Call n8n Workflow Tool" to trigger a separate workflow.
• MCP Client: Another n8n workflow or application capable of sending requests to this MCP server.

Setup/Usage

1. Import the Workflow:
   • Download the provided JSON file.
   • In your n8n instance, go to "Workflows" and click "New".
   • Click the "Import from JSON" button and paste the workflow JSON.
2. Configure Credentials:
   • OpenAI Chat Model: Configure the "OpenAI Chat Model" node with your OpenAI API Key.
   • Google Calendar (if applicable): If you implement the Google Calendar integration, ensure you have Google Calendar credentials set up in your n8n instance and configured in the sub-workflow that the "Call n8n Workflow Tool" will trigger.
3. Activate the Workflow: Enable the workflow in n8n.
4. Interact via MCP Client: Send requests from your MCP client to the webhook URL provided by the "MCP Server Trigger" node. The AI Agent will process these requests, using its language model and available tools to generate a response.
5. Customize Tools:
   • To add custom functionality (like Google Calendar interactions), create a separate n8n workflow that performs the desired action.
   • Configure the "Call n8n Workflow Tool" in this MCP server workflow to trigger your custom workflow, passing relevant data as input.
   • Ensure the custom workflow is designed to receive input from the "Call n8n Workflow Tool" and return a meaningful output for the AI Agent.