TopolearnMCP connects n8n to the wider AI ecosystem in both directions, with real security implications.
MCP (Model Context Protocol) is an open standard that lets AI assistants discover and call tools provided by other systems. An MCP server offers tools; an MCP client (such as Claude Desktop, Cursor or an n8n agent) uses them. n8n can play both roles.
n8n as a server: instance-level MCP. An owner or admin turns on Enable MCP access in Settings. Assistants then connect to one URL for the whole instance (ending in /mcp-server/http), signing in with OAuth or an API key. Nothing is exposed by default: you mark each workflow (or a whole project or folder) as available in MCP. Connected assistants can search workflows, run the exposed ones, and in recent 2.x versions also create and edit workflows and data tables. Access follows the signed-in user's permissions.
n8n as a server: MCP Server Trigger. This node turns a single workflow into its own MCP server with specific tools attached. Use it for a small, self-contained tool set; use instance-level access for broad use.
n8n as a client. Give an AI Agent the MCP Client Tool to use tools from an external MCP server. Recent versions also offer some popular MCP servers directly in the agent's tools panel.
Design advice:
- Expose workflows with clear names, descriptions and defined inputs (sub-workflow style), not your whole instance.
- Treat every MCP call like a webhook from the outside: validate inputs.
- Assistants act on whatever they read, so keep write actions behind Human review.
- Revoke API keys and OAuth grants you no longer use.
Remember
- Instance-level MCP exposes chosen workflows to assistants through one URL
- MCP Server Trigger makes one workflow an MCP server; MCP Client Tool consumes others
- Expose little, validate inputs, and keep approvals on write actions