Error: Unrecognized node type: n8n-nodes-mcp.mcpClient

I have an issue with the community mcp client. I get this error on each workflow that the mcp client is added, even if its not connected.

error: Unrecognized node type: n8n-nodes-mcp.mcpClient

at LoadNodesAndCredentials.getNode (/usr/local/lib/node_modules/n8n/src/load-nodes-and-credentials.ts:434:10)

at NodeTypes.getByNameAndVersion (/usr/local/lib/node_modules/n8n/src/node-types.ts:51:45)

at new Workflow (/usr/local/lib/node_modules/n8n/node_modules/.pnpm/n8n-workflow@file+packages+workflow/node_modules/n8n-workflow/src/workflow.ts:99:30)

at JobProcessor.processJob (/usr/local/lib/node_modules/n8n/src/scaling/job-processor.ts:110:20)

at processTicksAndRejections (node:internal/process/task_queues:105:5)

at Queue.<anonymous> (/usr/local/lib/node_modules/n8n/src/scaling/scaling.service.ts:102:5)

Problem with execution 2156: Error: Unrecognized node type: n8n-nodes-mcp.mcpClient. Aborting.

Error: Unrecognized node type: n8n-nodes-mcp.mcpClient (execution 2156)

Only running or waiting executions can be stopped and 2156 is currently error

I have tried various workarounds like local installation on the docker container, setting env, etc, but nothing helps, and each workflow that has an mcp client node is not executed, and this error is present.
This is the docker compose file im using:

services:

n8n:
image: n8nio/n8n:latest
restart: always
environment:

  • DB_TYPE=postgresdb
  • DB_POSTGRESDB_HOST=postgres
  • DB_POSTGRESDB_PORT=5432
  • DB_POSTGRESDB_DATABASE=n8n
  • DB_POSTGRESDB_USER=n8n
  • DB_POSTGRESDB_PASSWORD=******
  • GENERIC_TIMEZONE=Asia/Jerusalem
  • TZ=Asia/Jerusalem
  • N8N_ENFORCE_SETTINGS_FILE_PERMISSIONS=true
  • N8N_COMMUNITY_PACKAGES_ALLOW_TOOL_USAGE=true
  • N8N_COMMUNITY_PACKAGES_ENABLED=true
  • N8N_COMMUNITY_NODES_ENABLED=true
  • N8N_COMMUNITY_PACKAGES_INSTALL=n8n-nodes-mcp
  • N8N_MCP_ENABLED=true
  • N8N_FEATURE_FLAG_MCP=true
  • N8N_RUNNERS_ENABLED=true
  • OFFLOAD_MANUAL_EXECUTIONS_TO_WORKERS=true

Redis configuration for queue mode and AI agent memory

  • EXECUTIONS_MODE=queue
  • QUEUE_BULL_REDIS_HOST=redis
  • QUEUE_BULL_REDIS_PORT=6379
  • N8N_ENCRYPTION_KEY=******
ports:
 - "5678:5678"
volumes:
  - n8n_data:/home/node/.n8n
depends_on:
  - redis
  - postgres

n8n-worker:
image: n8nio/n8n:latest
restart: always
environment:

  • DB_TYPE=postgresdb
  • DB_POSTGRESDB_HOST=postgres
  • DB_POSTGRESDB_PORT=5432
  • DB_POSTGRESDB_DATABASE=n8n
  • DB_POSTGRESDB_USER=n8n
  • DB_POSTGRESDB_PASSWORD=*****
  • EXECUTIONS_MODE=queue
  • QUEUE_BULL_REDIS_HOST=redis
  • QUEUE_BULL_REDIS_PORT=6379
  • GENERIC_TIMEZONE=Asia/Jerusalem
  • TZ=Asia/Jerusalem
  • N8N_ENFORCE_SETTINGS_FILE_PERMISSIONS=true
  • N8N_COMMUNITY_PACKAGES_ALLOW_TOOL_USAGE=true
  • N8N_PROCESS=worker
  • N8N_ENCRYPTION_KEY=*****
  • OFFLOAD_MANUAL_EXECUTIONS_TO_WORKERS=true
  • N8N_FEATURE_FLAG_MCP=true
  • N8N_COMMUNITY_PACKAGES_ENABLED=true
  • N8N_COMMUNITY_NODES_ENABLED=true
  • N8N_COMMUNITY_PACKAGES_INSTALL=n8n-nodes-mcp
  • N8N_MCP_ENABLED=true
  • N8N_RUNNERS_ENABLED=true
    depends_on:
  • redis
  • postgres
    command: worker

redis:
image: redis:alpine
restart: always
volumes:

  • redis_data:/data
    command: redis-server --appendonly yes
    labels:
  • traefik.enable=false
    postgres:
    image: postgres:15
    container_name: n8n-postgres
    restart: always
    environment:
    POSTGRES_USER: n8n
    POSTGRES_PASSWORD: *****
    POSTGRES_DB: n8n
    volumes:
  • pgdata:/var/lib/postgresql/data

watchtower:
image: containrrr/watchtower
restart: always
volumes:

  • /var/run/docker.sock:/var/run/docker.sock
    command: --schedule “0 0 * * *” --cleanup
    environment:
  • TZ=Asia/Jerusalem

volumes:
n8n_data:
external: false
traefik_data:
external: false
redis_data:
external: false
workspace:
external: false
pgdata:
external: false

Why are you still using the community mcp node? MCP was introduced natively in n8n which you can make use of directly without having to try and install community nodes

Assentily, I can use the built-in mcp client, but it doesn’t support stdio and thats the reason I used the community node instead.

May I ask what is the need for stdio? What is the use case youre trying to solve for?

From my experience, that’s the most prevalent and effective method for interacting with MCP-based tools. I can run any mcp tool locally with npx or uv, without installation, and blazingly fast. By the way, that’s how I use it most of the time with local agents

Right, ok I think I was confusing this with shell stdout. You’re referring to stdio vs sse, which I think n8n uses sse.

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.