n8n Workflow Builder MCP Server: Managing Workflows with Claude and Cursor

The n8n Workflow Builder MCP Server is an MCP server provided by the salacoste/mcp-n8n-workflow-builder repository, allowing Claude AI and Cursor IDE to create, manage, and monitor n8n workflows through the Model Context Protocol. The core change is not the introduction of a new workflow engine, but rather handing over some operations that typically require editing JSON or accessing the n8n UI to AI clients that support MCP via natural language.

For those with existing n8n automations, the traditional operation path usually involves configuring nodes in the n8n interface, adjusting workflows, and then viewing execution results. According to the project's README, users can directly describe their needs, and Claude Desktop or Cursor IDE can call the MCP Server to build and manage workflows. In other words, it aims to make the AI client an n8n management interface, rather than replacing n8n itself.

The project's documentation outlines capabilities covering the creation, updating, deletion, activation, and execution of workflows, as well as monitoring, retrying, and analyzing run records, in addition to tag management. The README also states that a single MCP Server can connect to multiple n8n environments, listing production, pre-release, and development environments as use cases. This design is suitable for users who need to switch between different environments without manually entering different n8n backends each time.

The installation and connection methods are also written in the project's README: the project offers an npm global installation solution and provides MCP configuration examples for Claude Desktop and Cursor IDE, with examples launching @kernel.salacoste/n8n-workflow-builder via npx. Before running, Node.js, npm, an n8n instance with API access, and Claude Desktop or Cursor IDE are required. A single n8n instance can be configured via N8N_HOST and N8N_API_KEY, while multiple instances use .config.json to set n8n_host and n8n_api_key for each environment.

The assessment of capabilities in this article is primarily based on the repository's README and the project's API coverage documentation. The latter takes the 23 methods of the n8n REST API v1 as a scope, stating that 19 are fully supported and 4 are partially supported. These numbers indicate the interfaces the project authors aim to cover and do not equate to actual performance across every deployment environment and workflow scenario.

The documentation also points out a clear limitation: activate_workflow and deactivate_workflow are labeled as partially supported due to limitations of the n8n API, with related operations only possible through the UI. Therefore, while this MCP Server can centralize workflow creation, maintenance, and execution review to AI clients, it cannot fully relocate all operations away from the n8n backend.

From this perspective, it is more suited as a management interface for teams or individuals already using n8n. While natural language can reduce repetitive configuration and search steps for run records, API permissions, environment configuration of n8n instances, and operations that still need to be completed in the UI remain boundaries of this solution.