n8n

A comprehensiveModel Context Protocol (MCP) serverthat provides AI assistants with direct access to your n8n automation platform. This server enables seamless integration between AI tools (like Claude Desktop) and n8n workflows, variables, credentials, and executions.

- list_workflows- List all workflows
- get_workflow- Get workflow details by ID
- create_workflow- Create new workflows
- update_workflow- Update existing workflows
- delete_workflow- Delete workflows
- activate_workflow- Activate workflows
- deactivate_workflow- Deactivate workflows

- list_variables- List all variables
- get_variable- Get variable by key
- create_variable- Create new variables
- update_variable- Update existing variables
- delete_variable- Delete variables

- list_credentials- List all credentials (sanitized)
- create_credential- Create new credentials
- delete_credential- Delete credentials

- list_executions- List workflow executions
- get_execution- Get execution details by ID

- self_test- Test server connectivity and permissions

- MCP Protocol: Full JSON-RPC 2.0 compliance via stdio transport
- HTTP Bridge: Health checks and testing endpoints
- Auto-detection: Automatically switches between modes

- Node.js 18+
- n8n instance running and accessible
- n8n API key configured

git clone <repository-url> cd n8n-mcp

cp .env.example .env # Edit .env with your settings: # N8N_API_KEY=your-api-key-here # N8N_BASE_URL=http://localhost:5678 # MCP_PORT=3001

# Test HTTP endpoints node index.js & curl http://localhost:3001/health # Test MCP protocol node test-all-tools.js

The server runs as a stdio-based MCP server for AI clients:

Claude Desktop Configuration(~/.claude_desktop_config.json):

{ "mcpServers": { "n8n": { "command": "node", "args": ["index.js"], "cwd": "/path/to/n8n-mcp", "env": { "N8N_API_KEY": "your-n8n-api-key-here", "N8N_BASE_URL": "http://localhost:5678" } } } }

When run in a terminal (TTY), the server provides HTTP endpoints:

node index.js # Server starts on http://localhost:3001 # Available endpoints: # GET /health - Health check # POST /test - Run self-test # GET / - Usage instructions

# Initialize connection echo '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"test","version":"1.0.0"}}}' | node index.js # List tools echo '{"jsonrpc":"2.0","id":2,"method":"tools/list","params":{}}' | node index.js # Call a tool echo '{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"self_test","arguments":{}}}' | node index.js

Run the complete test suite to validate all 18 tools:

- Test MCP protocol compliance
- Validate all tool definitions
- Check n8n API connectivity
- Verify error handling
- Provide detailed results

# Health check curl http://localhost:3001/health # Quick self-test curl -POST http://localhost:3001/test # Individual tool test echo '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"list_workflows","arguments":{"limit":5}}}' | node index.js

- Store API keys in environment variables
- Use.envfiles for local development
- Never commit API keys to version control

- Credential data is automatically sanitized in responses
- Only metadata (ID, name, type) is exposed
- Sensitive credential data is never returned

- HTTP server binds to localhost by default
- CORS headers configured for cross-origin requests
- No sensitive data exposed via HTTP endpoints

1. "N8N_API_TOKEN not configured"

# Solution: Set your API key export N8N_API_KEY=your-api-key-here # Or add to .env file

2. "Connection refused" errors

# Solution: Check n8n is running curl http://localhost:5678/api/v1/workflows?limit=1 -H "X-N8N-API-KEY: your-key"

3. "License does not allow for feat:variables"

# This is expected for n8n Community Edition # Variables require n8n Pro/Enterprise license # The tool will still work but return license errors

4. "GET method not allowed" for credentials

# Some n8n configurations restrict credential access # Check your n8n security settings

# Solution: Kill existing process or change port pkill -f "node index.js" # Or set different port: MCP_PORT=3002 node index.js

# Test n8n connectivity curl -H "X-N8N-API-KEY: your-key" http://localhost:5678/api/v1/workflows?limit=1 # Test MCP server node test-all-tools.js

# Basic health curl http://localhost:3001/health # Detailed system test curl -X POST http://localhost:3001/test | jq '.result.summary'

The server logs all tool executions and provides timing information:

- Tool execution time
- n8n API response time
- Error rates and types
- Fork the repository
- Create a feature branch
- Make your changes
- Run tests:node test-all-tools.js
- Submit a pull request
- Add tool definition insetupToolHandlers()
- Implement the tool method
- Add test case intest-all-tools.js
- Update documentation

MIT License - see LICENSE file for details.

- n8n- Workflow automation platform
- Model Context Protocol- Protocol specification
- Claude Desktop- AI assistant with MCP support

Automated document processing and extraction