Chapter 30: MCP Integrations

Last Updated: 2026-03

30.2 The n8n MCP Server

The n8n MCP server exposes the n8n API as a set of tools that Cursor can call during a session. This enables:

• Creating, validating, and publishing workflows from Cursor
• Listing existing workflows and their status
• Triggering test executions
• Reading workflow logs

Configuration file: ~/.mcp.json (workspace root)

{
    "mcpServers": {
        "n8n": {
            "url": "http://localhost:5678/mcp-server/http",
            "headers": {
                "Authorization": "Bearer YOUR_N8N_API_KEY"
            }
        }
    }
}

Note: The n8n API key is stored in .mcp.json. This file should not be committed to Git if it contains credentials. Add it to .gitignore if needed.

---

30.3 n8n MCP Tool Reference

The MCP server exposes these tools:

Tool	Purpose
get_sdk_reference	Load the n8n Workflow SDK documentation
search_nodes	Find n8n node types by keyword
get_node_types	Get exact parameter definitions for specific nodes
get_suggested_nodes	Get curated node recommendations for workflow technique categories
validate_workflow	Validate workflow code before creating/updating
create_workflow_from_code	Create a new workflow in n8n
update_workflow	Update an existing workflow
publish_workflow	Activate a workflow (make it live)
archive_workflow	Deactivate and archive a workflow

---

30.4 Using MCP Tools in Cursor

MCP tools are invoked through the Cursor AI — you describe what you want and Cursor calls the appropriate MCP tool.

Example: Create a new workflow

"Create an n8n workflow that receives a webhook, calls the Dify API to retrieve 
relevant knowledge base content, then calls Claude to generate a response, 
and returns the response to the caller."

Cursor will:

1. Call get_sdk_reference to load SDK patterns.
2. Call search_nodes to find webhook, HTTP Request, and AI Agent nodes.
3. Call get_node_types to get exact parameter definitions.
4. Write the workflow code.
5. Call validate_workflow to check for errors.
6. Call create_workflow_from_code to save it.
7. Call publish_workflow to activate it.

Example: Debug an existing workflow

"The iti-career-coach-session workflow is failing. Check its recent executions 
and tell me what's wrong."

---

30.5 Checking MCP Server Status

Verify the n8n MCP server is accessible:

curl -s http://localhost:5678/mcp-server/http \
  -H "Authorization: Bearer YOUR_N8N_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"method": "tools/list", "params": {}}' | python3 -m json.tool

If this returns a list of tools, the MCP server is operational.

---

30.6 Adding New MCP Servers

To add a new MCP server:

1. Install or configure the MCP server (see that server’s documentation).
2. Add the server configuration to .mcp.json:

{
    "mcpServers": {
        "n8n": { ... existing config ... },
        "my-new-server": {
            "url": "http://localhost:PORT/mcp",
            "headers": {
                "Authorization": "Bearer TOKEN"
            }
        }
    }
}

1. Restart Cursor to load the new server.
2. Verify the server appears in Cursor’s MCP tool list.

---

30.7 MCP vs Direct API Calls

Approach	Use When
MCP tool in Cursor	Building or modifying n8n workflows during development
Direct API call (curl/code)	Automated scripts, CI/CD, production integrations
n8n web UI	Visual workflow editing, execution monitoring

---

Previous: Chapter 29 — Cursor Rules | Next: Chapter 31 — Team Roles