n8n-mcp

MCP运维云端

提供方 aidvgg

运维云端 Streamable HTTP API Key 社区 / 第三方 Claude DesktopClaudeCursorVS Code
查看接入方式

n8n-mcp is an MCP server that provides AI assistants with 29 typed tools to create, run, and fix n8n automation workflows, suitable for AI coding environments such as Claude, Cursor, and VS Code.

一句话选型结论

按资料字段整理

n8n-mcp 属于运维云端类 MCP 服务器,Streamable HTTP,API Key,可直接接入 Claude Desktop、Claude、Cursor 等,有可直接复制的远程端点配置。接入前先确认权限范围;涉及写入、删除或付款的操作,建议先在测试环境验证。

分类运维云端 传输方式Streamable HTTP 鉴权API Key 部署方式远程部署
GitHub 星标 1 仓库热度,无公开仓库时显示未标注
工具数 未标注 暴露给模型的工具数量
资源 / 提示词 0 / 0 资源与提示词模板数量
维护方 社区 / 第三方 核验等级 仅资料核对

工具与能力

运维云端
数量来自来源目录的标注,具体工具清单以官方文档为准
工具数量未标注暴露给模型可调用的工具个数
资源数量0可读取的资源条目
提示词数量0内置提示词模板数量

能力标签

Blend Deployment MCP Server Workflow Automation TypeScript AI Coding Assistant n8n Integration Self-Healing Diagnosis Workflow Validation

支持客户端

适用场景

Blend DeploymentMCP ServerWorkflow Automation

安装与配置

Streamable HTTP API Key
新一代远程传输,替代 SSE
node /absolute/path/to/n8n-mcp/dist/server.js --stdio

这条命令只是入口,具体启动参数请按官方文档补齐,不要直接照搬到别的服务器。

  1. 1
    准备运行环境

    确认你的客户端支持远程端点,并拿到服务地址。

  2. 2
    安装服务

    复制上面的安装命令执行;命令里带占位符的位置,换成自己的密钥或路径。

  3. 3
    写入客户端配置

    在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,让新配置生效。

  4. 4
    验证是否可用

    连接成功后随便调一个只读工具,确认能返回结果,再接入正式流程。

典型使用场景

3 类
按分类与标签整理的常见用法,具体以官方文档为准
云资源查询

查实例、存储、账单等云上资源状态。

部署与发布

把部署、重启、回滚这类操作接进对话。

监控与告警

拉取监控指标,异常时让模型帮你判断。

选择建议

选之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 有现成的安装命令,不用自己找入口,照着跑就能起服务。
  • 已适配 Claude Desktop、Claude、Cursor,换客户端不用重新研究配置。

先想清楚的地方

  • 远程端点依赖网络与服务方可用性,注意密钥与调用额度。
  • 使用前要准备好API Key对应的凭据,并按最小权限配置。
  • 资料没标注工具数量,实际能力要装好后确认。
  • 版本更新与行为细节以官方仓库为准。

更适合

  • 需要在运维云端场景里让模型直接调用外部能力的人。
  • 已经在用 Claude Desktop、Claude 的用户,接入路径最短。
  • 想先快速试一下、能接受在本地或自有环境跑服务的人。

不太适合

  • 不方便管理密钥或第三方授权的人。
  • 需要厂商级 SLA 和技术支持的生产环境。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
MCP 服务器工具数传输方式部署方式安装命令最后核验
n8n-mcp当前 未标注 Streamable HTTP 远程部署 有 2026-09-24
lobehub 未标注 stdio · 本地进程 本地部署 有 2026-04-11
TrendRadar 未标注 stdio · 本地进程 本地部署 有 2026-04-13
Ui Tars Desktop 8a825770 未标注 stdio · 本地进程 本地部署 有 2026-05-17
ruflo 未标注 stdio · 本地进程 本地部署 有 2026-04-13
FastGPT 未标注 stdio · 本地进程 本地部署 有 2026-04-11

服务器档案

运维云端
字段来自项目仓库与公开索引,中文说明由本站标注
提供方aidvgg上游标注的维护方 aidvgg
分类运维云端本站按上游分类展示
连接方式Streamable HTTP新一代远程传输,替代 SSE
鉴权方式API Key远程服务建议用最小权限凭据
维护方社区 / 第三方核验等级 仅资料核对
包名 / 运行时未标注 · node本地安装时用到的包标识与运行时
部署类型远程部署上游给出的部署形态
最近核验2026-09-24上游最后核验时间

详细介绍

8.2 KB
来自项目仓库与上游目录的原文,未做删改

n8n-mcp

Lets an AI assistant build, run and fix n8n automations for you.

Connect it to Claude, Cursor or VS Code and describe the automation you want. The assistant gets 29 typed tools: it checks a workflow against a catalogue of 300+ n8n nodes before creating it, runs it, reads the failure when a step breaks, and gets back concrete fix suggestions it can apply. The checking happens before anything touches your n8n instance.

Built on the MCP TypeScript SDK and the n8n public API. Runs over Streamable HTTP (for remote clients like Claude, Cursor, VS Code) or stdio (for local desktop clients).

Features

  • 29 typed tools spanning workflow management, execution, diagnostics, node intelligence, templates, and validation.
  • Pre-creation validation - validate_workflow checks node types, required parameters, connection integrity, duplicate names, orphan nodes, credentials, and typeVersion mismatches before anything hits n8n.
  • Self-healing loop - self_heal_workflow executes a workflow, diagnoses failures, and returns concrete fix suggestions the agent can apply via update_workflow.
  • Node catalogue kept in sync with n8n nodes-base v2.14.0, including current typeVersion values for HTTP Request, Postgres, Slack, Gmail, OpenAI, and the newer AI nodes (AI Transform, Data Table, Guardrails, Evaluation, MCP Server Trigger).
  • Golden-path examples - annotated workflow templates for common patterns (webhook-transform-respond, schedule-fetch-filter-notify, error handling, batch loops).
  • HTTP server hardening - CORS allow-listing, rate limiting, security headers, structured logging via pino, and graceful shutdown.
  • Cloud client - a tiny CLI that talks to a remote MCP endpoint over curl, for environments where claude mcp add is unavailable.

Quick start```bash

git clone https://github.com/aidvgg/n8n-mcp.git
cd n8n-mcp
npm install
npm run build


Create a `.env` file:```env
N8N_API_URL=https://your-n8n-instance.example.com/api/v1
N8N_API_KEY=your-api-key-here
PORT=3000
NODE_ENV=development
ALLOWED_ORIGINS=https://claude.ai,https://cursor.sh

Run it:```bash
npm start # HTTP server on PORT (default 3000)
npm run start:stdio # stdio mode for local desktop clients
npm run dev # watch mode (bun)


Health check:```bash
curl http://localhost:3000/health

Connecting an MCP client

Claude Code, Cursor, VS Code (Streamable HTTP)

Point your client at http://localhost:3000/mcp (or your deployed URL). For Claude Code:```bash
claude mcp add n8n https://your-deployment.example.com/mcp


### Claude Desktop (stdio)

In `claude_desktop_config.json`:```json
{
  "mcpServers": {
    "n8n": {
      "command": "node",
      "args": ["/absolute/path/to/n8n-mcp/dist/server.js", "--stdio"],
      "env": {
        "N8N_API_URL": "https://your-n8n.example.com/api/v1",
        "N8N_API_KEY": "your-api-key"
      }
    }
  }
}

Cloud sessions (no mcp add available)

Use the bundled cloud client to call tools directly:```bash
node dist/cloud-client.js https://your-deployment.example.com/mcp list-tools
node dist/cloud-client.js https://your-deployment.example.com/mcp call list_workflows '{}'
node dist/cloud-client.js https://your-deployment.example.com/mcp call execute_workflow '{"workflowId":"123"}'


The default URL can be overridden with the `MCP_SERVER_URL` environment variable.

## Available tools

### Workflow management

| Tool | Arguments | Description |
|------|-----------|-------------|
| `list_workflows` | `{active?, tags?, name?, limit?}` | List workflows with filters |
| `get_workflow` | `{workflowId}` | Full workflow definition |
| `create_workflow` | `{name, nodes, connections, settings}` | Create a workflow |
| `update_workflow` | `{workflowId, name?, nodes?, connections?, settings?}` | Update a workflow |
| `delete_workflow` | `{workflowId}` | Delete a workflow |
| `activate_workflow` / `deactivate_workflow` | `{workflowId}` | Toggle active state |

### Execution

| Tool | Arguments | Description |
|------|-----------|-------------|
| `list_executions` | `{workflowId?, status?, limit?}` | List executions |
| `get_execution` | `{executionId}` | Execution details and run data |
| `delete_execution` | `{executionId}` | Delete an execution record |
| `execute_webhook` | `{webhookPath, data?, username?, password?}` | Trigger via webhook |
| `execute_workflow` | `{workflowId, payload?, timeoutMs?}` | Execute and wait for results |

### Diagnostics and self-healing

| Tool | Arguments | Description |
|------|-----------|-------------|
| `diagnose_execution` | `{executionId}` | Analyze errors in an execution |
| `self_heal_workflow` | `{workflowId, payload?, timeoutMs?}` | Execute, diagnose, suggest fixes |

### Node intelligence

| Tool | Arguments | Description |
|------|-----------|-------------|
| `get_node_types` | `{category?}` | List nodes by category |
| `get_node_schema` | `{nodeType}` | Parameter schema for a node |
| `search_nodes` | `{query}` | Keyword search across the catalogue |
| `get_expression_help` | `{topic?}` | n8n expression syntax reference |

### Templates and examples

| Tool | Arguments | Description |
|------|-----------|-------------|
| `get_workflow_templates` | `{}` | List built-in templates |
| `get_workflow_template` | `{templateName}` | Fetch a template |
| `list_workflow_examples` | `{}` | List golden-path examples |
| `get_workflow_example` | `{exampleName}` | Annotated example workflow |

### Validation

| Tool | Arguments | Description |
|------|-----------|-------------|
| `validate_workflow` | `{nodes, connections}` | Check a definition before creating it |

### Metadata

| Tool | Arguments | Description |
|------|-----------|-------------|
| `list_tags` / `create_tag` | `{}` / `{name}` | Workflow tags |
| `list_credentials` | `{}` | List credentials |
| `get_credential_schema` | `{credentialType}` | Credential parameter schema |
| `list_variables` | `{}` | Environment variables |
| `run_audit` | `{categories?}` | n8n security audit |

## Recommended workflow development cycle

1. `list_workflow_examples` - find a similar pattern
2. `get_node_schema` - confirm parameters for each node
3. `validate_workflow` - verify the definition before creation
4. `create_workflow` - deploy it
5. `self_heal_workflow` - execute and collect fix suggestions
6. `update_workflow` - apply the fixes
7. Repeat 5–6 until every node passes

## Configuration

| Variable | Default | Description |
|----------|---------|-------------|
| `N8N_API_URL` | `http://localhost:5678/api/v1` | n8n REST API base URL |
| `N8N_API_KEY` | _(required)_ | n8n API key |
| `PORT` | `3000` | HTTP server port |
| `NODE_ENV` | `development` | In `production`, CORS requires `ALLOWED_ORIGINS` |
| `ALLOWED_ORIGINS` | _(empty)_ | Comma-separated CORS allow-list |
| `REQUEST_TIMEOUT` | `30000` | n8n client request timeout (ms) |
| `MAX_RETRIES` | `3` | n8n client retry count |
| `DOTENV_CONFIG_PATH` | _(unset)_ | Load `.env` from a custom path (useful when running the server from another project) |

## HTTP endpoints

| Path | Method | Purpose |
|------|--------|---------|
| `/mcp` | POST | MCP JSON-RPC (Streamable HTTP) |
| `/mcp` | DELETE | Session cleanup acknowledgment |
| `/health` | GET | Liveness check with uptime and n8n target |
| `/docs` | GET | Markdown reference designed for AI agents calling the server via curl |

## Development```bash
npm run dev         # watch mode via bun
npm run build       # compile TypeScript to dist/
npm run typecheck   # type-check without emitting
bun test            # run the test suite
bun test --coverage # with coverage report
npm run inspector   # launch the MCP Inspector against this server

The codebase is laid out as:src/ server.ts # HTTP + stdio entry point, transport wiring tools.ts # Tool registration and validation logic resources.ts # MCP resource handlers (workflows, executions) n8n-client.ts # n8n REST API client with retries and timeouts node-catalog.ts # Node schemas, categories, search examples.ts # Golden-path workflow examples expressions/ # n8n expression reference templates/ # Built-in workflow templates nodes/ # Per-integration node definitions cloud-client.ts # Standalone CLI for remote MCP calls logger.ts # pino setup

License

MIT.

常见问题

8 问
答案由本站依据条目字段整理,不引用上游原文
n8n-mcp 怎么安装?

在客户端按下面的安装命令接入:node /absolute/path/to/n8n-mcp/dist/server.js --stdio。装完在客户端配置里引用同一个包或端点。

n8n-mcp 怎么在 Claude Desktop、Claude、Cursor 里配置?

它的传输方式是Streamable HTTP,在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,配置才会生效。

n8n-mcp 收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于服务本身和背后的接口额度,正式使用前按官方说明确认。

n8n-mcp 需要 API Key 吗?

资料标注需要API Key认证,使用前准备好对应凭据,并按最小权限配置。

n8n-mcp 支持哪些客户端?

资料标注支持 Claude Desktop、Claude、Cursor、VS Code;其他客户端只要能配置 MCP Server,也能用同样的命令或地址接入。

n8n-mcp 能提供哪些能力?

资料显示能力标签包括 Blend Deployment、MCP Server、Workflow Automation、TypeScript、AI Coding Assistant、n8n Integration。具体工具清单以官方文档为准。

n8n-mcp 需要什么运行环境?

运行时为 node;属于远程端点,不需要在本地装运行时。

n8n-mcp 是本地还是远程?

它是远程端点,连服务地址即可。传输方式为Streamable HTTP。

热门专题

按标签挑服务器
全部专题 →

服务器信息以作者仓库与官方文档为准;分类、传输方式与鉴权方式的中文说明为中文整理,最近核验 2026-09-24。