🌐

n8n MCP Server

Updated June 2026Trust grade A94/100

N8n-MCP gives an AI assistant deep knowledge of n8n rather than a remote control for it: 2,412 nodes (829 core plus 1,583 community, 1,340 of them verified) with 99% property coverage and 66.5% operation coverage, 86% documentation coverage, a library of 2,352 workflow templates, and validators that check a node config or a whole workflow before anything is deployed. Built by czlonkowski, it is community-built and best for API & Web.

by czlonkowski

About

n8n-MCP gives an AI assistant deep knowledge of n8n rather than a remote control for it: 2,412 nodes (829 core plus 1,583 community, 1,340 of them verified) with 99% property coverage and 66.5% operation coverage, 86% documentation coverage, a library of 2,352 workflow templates, and validators that check a node config or a whole workflow before anything is deployed. Seven documentation tools always load — search_templates, get_template, search_nodes, get_node, validate_node, validate_workflow and tools_documentation — and get_node is deliberately tiered because full node detail costs 3,000-8,000 tokens. Supplying N8N_API_URL and N8N_API_KEY unlocks a further sixteen management tools that create, partially update, validate, autofix, deploy and inspect workflows on your own instance, plus executions, credentials, folders, data tables and health checks; without credentials the server cannot execute anything at all. MCP_MODE=stdio is required for stdio clients, or log output corrupts the JSON-RPC stream; WEBHOOK_SECURITY_MODE=moderate is required when N8N_API_URL points at localhost or host.docker.internal, because the default SSRF gate rejects loopback. Install with npx n8n-mcp, the ghcr.io/czlonkowski/n8n-mcp Docker image, or the maintainer's hosted instance at dashboard.n8n-mcp.com (100 tool calls/day free). This is a community project by Romuald Czlonkowski under MIT, not published by n8n, and it is distinct from n8n's own MCP Server Trigger node, which does the reverse — exposing an n8n workflow as an MCP server over SSE or Streamable HTTP. The project's own headline warning is never to let an AI edit production workflows directly.

A
Reliable94/100
low confidence · 1 measured signal

Grade A (94/100, reliable) from 1 measured signal, based on repository evidence. Only one signal stands behind it, so treat the grade as provisional.

What was measured

  • Repository maintenance100/100 · weight 20

    The repository has been pushed to or released within the last six months. — last push 2026-07-25, last release 2026-07-23 (v2.65.2).

  • Source verification100/100 · weight 25

    The repository URL was confirmed to resolve against the live GitHub API and is not archived.

  • Provenance65/100 · weight 10

    Community-built. That is not a mark against it — most of the ecosystem is community-built — but there is no vendor accountable for keeping it working.

  • Listing ↔ repository match100/100 · weight 5

    The listing name lines up with the linked repository czlonkowski/n8n-mcp.

What could not be measured

These contributed nothing to the score — not a penalty, not a zero. They are why the confidence reads the way it does.

  • Live MCP handshakeunknown

    No remote endpoint to handshake — this server installs and runs locally over stdio, so there is nothing to probe from the outside.

  • Measured uptimeunknown

    No probe history recorded for this server yet.

  • Tool-schema stabilityunknown

    Drift is a difference between two successive checks, and this server has none recorded.

Installation

npm / npx
npx n8n-mcp

n8n-mcp confirmed live on the npm registry — checked August 17, 2026.

n8n-MCP is not a workflow execution engine, and expecting it to be one is the single biggest source of disappointment with it. What it is: a searchable reference to 2,412 n8n nodes (829 core, 1,583 community) with 99% property coverage, 66.5% operation coverage, 2,352 workflow templates, and validators that check a node config or a whole workflow *before* you deploy it. Give it your instance's API credentials and it gains a second half — 16 management tools that create, update, deploy and inspect workflows on that instance. It is also a community project by Romuald Czlonkowski (22.6k★, MIT), not something n8n publishes, and it is unrelated to n8n's own MCP Server Trigger node, which points the other way: that node makes an n8n workflow *into* an MCP server for clients to call.

Setting up n8n-MCP

  1. 1.Fastest path — the hosted instance

    dashboard.n8n-mcp.com is the maintainer's hosted version: sign up, take an API key, connect your client. Free tier is 100 tool calls per day and the node/template data is kept current, which is the part that goes stale in a local install.

  2. 2.Local — add it to Claude Code, docs tools only

    MCP_MODE=stdio is required, not cosmetic: without it the process writes log lines to stdout and the client reports JSON parse errors like "Unexpected token…". LOG_LEVEL=error and DISABLE_CONSOLE_OUTPUT=true exist for the same reason.

    shell
    claude mcp add n8n-mcp \
      -e MCP_MODE=stdio \
      -e LOG_LEVEL=error \
      -e DISABLE_CONSOLE_OUTPUT=true \
      -- npx n8n-mcp
  3. 3.Add your instance to unlock the management tools

    The n8n API credentials are optional. Without them you get documentation, templates and validation. With them you additionally get create/update/execute/inspect against that instance. For a local n8n, N8N_API_URL is http://localhost:5678 — or http://host.docker.internal:5678 if n8n-MCP itself runs in Docker.

    shell
    claude mcp add n8n-mcp \
      -e MCP_MODE=stdio \
      -e LOG_LEVEL=error \
      -e DISABLE_CONSOLE_OUTPUT=true \
      -e N8N_API_URL=https://your-n8n-instance.com \
      -e N8N_API_KEY=your-api-key \
      -- npx n8n-mcp
  4. 4.Docker, including the loopback trap

    The image is ghcr.io/czlonkowski/n8n-mcp:latest. If N8N_API_URL points at localhost or host.docker.internal you must also set WEBHOOK_SECURITY_MODE=moderate — the SSRF gate covers both webhook triggers and the API client, and the default `strict` mode rejects loopback addresses outright. `moderate` allows localhost while still blocking RFC1918 ranges and cloud metadata endpoints.

    claude_desktop_config.json
    {
      "mcpServers": {
        "n8n-mcp": {
          "command": "docker",
          "args": [
            "run", "-i", "--rm",
            "-e", "MCP_MODE=stdio",
            "-e", "LOG_LEVEL=error",
            "-e", "DISABLE_CONSOLE_OUTPUT=true",
            "-e", "N8N_API_URL=http://host.docker.internal:5678",
            "-e", "N8N_API_KEY=your-api-key",
            "-e", "WEBHOOK_SECURITY_MODE=moderate",
            "ghcr.io/czlonkowski/n8n-mcp:latest"
          ]
        }
      }
    }
  5. 5.Behind Cloudflare Access

    Set N8N_CF_CLIENT_ID and N8N_CF_CLIENT_SECRET; they are sent as CF-Access-Client-Id / CF-Access-Client-Secret on API requests, health probes and webhook executions. The token is scoped to the N8N_API_URL origin only — a webhook on a different host deliberately does not receive it.

The 23 tools, in the order you actually use them

Seven documentation tools always load; the sixteen n8n_* management tools appear only when API credentials are present. `DISABLED_TOOLS` and `DISABLED_TOOL_OPERATIONS` (e.g. `n8n_executions:delete`) let you remove capabilities you do not want an agent to have.

search_templates / get_template

Start here. 2,352 templates, searchable by node type — `{searchMode: "by_nodes", nodeTypes: ["n8n-nodes-base.slack"]}`. Using a template requires the documented attribution line crediting its author.

search_nodes

Find nodes by keyword. `includeExamples: true` pulls real configurations; `source: "community"|"verified"` filters the 1,583 community nodes.

get_node

The token-economics tool. `detail: "minimal"` is ~200 tokens, `"standard"` is the default, `"full"` is 3,000–8,000 tokens per node. `mode: "search_properties"` with a propertyQuery finds one property without loading the schema; `mode: "docs"` returns readable markdown.

validate_node

Run `mode: "minimal"` first (<100ms, required fields only), then `mode: "full", profile: "runtime"` for full validation with suggested fixes.

validate_workflow

Whole-workflow check — connections and expressions, not just per-node fields. The documented order is minimal → full → workflow.

n8n_create_workflow / n8n_update_partial_workflow

Management. Partial (diff-style) updates need n8n 2.32+; they let an agent change one node without rewriting the workflow JSON.

n8n_autofix_workflow / n8n_validate_workflow

Validate what is already deployed, and attempt repairs against the live instance.

n8n_test_workflow / n8n_executions

Trigger a workflow and read execution history — the closest this server comes to "running" things, and only with API credentials.

n8n_manage_credentials / n8n_audit_instance / n8n_health_check

Instance-level operations. Folder management needs n8n 2.19+; some evaluation features need 2.30+.

What people use it for

Build a workflow that works on the first deploy

Search templates for a Slack-to-Google-Sheets workflow. If nothing fits, find the nodes, get their standard details with examples, configure every parameter explicitly, then validate each node and the whole workflow before creating it.

This is the documented pattern, and the reason it exists is the failure mode below: relying on default parameter values is the primary cause of runtime failures.

Explain a node you have never used

Get the docs for the n8n LangChain AI Agent node and list which properties are required versus optional, with an example config.

No API credentials needed — this half of the server is a documentation index, which is also why it is useful before you have an instance at all.

Fix a broken deployed workflow

Validate workflow <id> on my instance, then use a partial update to fix only the nodes that fail validation. Show me the diff before applying it.

Partial updates (n8n 2.32+) keep the change small, which matters because the project's own top-line warning is never to let AI edit production workflows directly.

Which one should you use?

n8n MCP Server Trigger (built into n8n)

Use this when you want your n8n workflows to be tools an agent can call. SSE or Streamable HTTP, bearer/header auth, no stdio. It is the mirror image of n8n-mcp, not a competitor.

dashboard.n8n-mcp.com (hosted)

Pick the hosted tier when you do not want to keep a node/template database current. 100 tool calls/day free; self-host when you need more or cannot send queries out.

Zapier

If the goal is "let the agent trigger automations" and you have no n8n instance, Zapier's MCP surface is action-oriented out of the box. n8n-mcp is for building n8n workflows, which is a different job.

Every command, environment variable, and endpoint above was read from the project’s own documentation on 2026-08-10: czlonkowski/n8n-mcp README, n8n-mcp — Claude Code setup guide, n8n-mcp — self-hosting guide, n8n docs — MCP Server Trigger node.

Browse all MCP server setup guides.

Frequently Asked Questions

Can n8n-MCP execute my n8n workflows?
Only indirectly, and only with credentials. The server itself is a documentation, template and validation layer; workflow execution happens through the n8n_* management tools calling your instance's API, which requires N8N_API_URL and N8N_API_KEY. With no credentials it cannot run anything at all.
Is n8n-mcp the official n8n MCP server?
No. It is a community project by Romuald Czlonkowski (MIT, 22.6k stars) and is not published by n8n. The confusion is worth clearing up because n8n does ship MCP functionality of its own — the MCP Server Trigger node — which does the opposite thing: it exposes an n8n workflow as an MCP server over SSE or Streamable HTTP, with bearer or header auth, at a randomly generated URL path. It does not support stdio.
Why do I get "Unexpected token" JSON errors in Claude Desktop?
MCP_MODE is not set to stdio. Without it, log output goes to stdout and corrupts the JSON-RPC stream. Set MCP_MODE=stdio, and LOG_LEVEL=error plus DISABLE_CONSOLE_OUTPUT=true alongside it.
Why does it fail to reach my local n8n instance?
The SSRF gate. Default WEBHOOK_SECURITY_MODE is strict, which rejects loopback addresses for both webhook triggers and the API client. Set WEBHOOK_SECURITY_MODE=moderate when N8N_API_URL is localhost or host.docker.internal — it still blocks private networks and cloud metadata.
Why does the agent build workflows that fail at runtime?
Because it left parameters at their defaults. The project states this directly: default values are the primary cause of runtime failures, so every parameter controlling node behaviour must be configured explicitly. Coverage is also uneven — 66.5% of node operations are documented, and `includeExamples` availability varies by node popularity.
How much context does it consume?
Enough to plan for. `get_node` with `detail: "full"` costs 3,000–8,000 tokens per node, so use `minimal` or `standard`, or `mode: "search_properties"` to fetch one property. Minimal validation returns in under 100ms.
Is it safe to point at production?
The maintainer's own answer is no: copy the workflow first, test in development, export backups, validate before deploying. If you need a harder guarantee, use DISABLED_TOOLS or DISABLED_TOOL_OPERATIONS to remove destructive operations from the tool list entirely.
What is n8n MCP Server?
n8n is an MCP server built by czlonkowski. N8n-MCP gives an AI assistant deep knowledge of n8n rather than a remote control for it: 2,412 nodes (829 core plus 1,583 community, 1,340 of them verified) with 99% property coverage and 66.5% operation coverage, 86% documentation coverage, a library of 2,352 workflow templates, and validators that check a node config or a whole workflow before anything is deployed. Seven documentation tools always load — search_templates, get_template, search_nodes, get_node, validate_node, validate_workflow and tools_documentation — and get_node is deliberately tiered because full node detail costs 3,000-8,000 tokens. Supplying N8N_API_URL and N8N_API_KEY unlocks a further sixteen management tools that create, partially update, validate, autofix, deploy and inspect workflows on your own instance, plus executions, credentials, folders, data tables and health checks; without credentials the server cannot execute anything at all. MCP_MODE=stdio is required for stdio clients, or log output corrupts the JSON-RPC stream; WEBHOOK_SECURITY_MODE=moderate is required when N8N_API_URL points at localhost or host.docker.internal, because the default SSRF gate rejects loopback. Install with npx n8n-mcp, the ghcr.io/czlonkowski/n8n-mcp Docker image, or the maintainer's hosted instance at dashboard.n8n-mcp.com (100 tool calls/day free). This is a community project by Romuald Czlonkowski under MIT, not published by n8n, and it is distinct from n8n's own MCP Server Trigger node, which does the reverse — exposing an n8n workflow as an MCP server over SSE or Streamable HTTP. The project's own headline warning is never to let an AI edit production workflows directly.
Who built n8n MCP Server?
n8n MCP Server was built by czlonkowski.
Is n8n MCP Server free?
Yes, n8n MCP Server has a free option. The MCP server is free and open-source. n8n: Free and open-source (self-hosted). Cloud Starter: $24/mo. Pro: $60/mo. Enterprise: Custom.
How do I install n8n MCP Server?
Install n8n MCP Server with npm: npx n8n-mcp
What does n8n MCP Server integrate with?
n8n MCP Server integrates with Claude Desktop, Cursor, VS Code, Windsurf, Cline.

Repo Health

Actively maintained

Local/stdio install — runs on your machine, so there is no remote endpoint to verify live. Trust signal below is from the source repo.

Last commit
1mo ago
Last release
v2.65.2 · 1mo ago
Install
npm
Package
n8n-mcp

Quick Info

Install Type
npm
Author
czlonkowski
Categories
3
Integrations
5

Sponsored

Better Stack

Free Plan

Get alerted when your APIs, browser tests, payment pipelines, or MCP server dependencies go down. Used by 100K+ developers.

Start monitoring free →