AI Developer Stack
How to Connect GoHighLevel to n8n With the MCP Server
By Marnix Geerkens. Published 2026-05-30. Updated 2026-05-30.
In short
You connect GoHighLevel to n8n with the built-in MCP Client node, no custom plugin needed. Point the node at the official MCP endpoint, choose the HTTP Streamable transport, and authenticate with a Private Integration Token in a header. Then n8n can read and write your CRM as part of any workflow, so a new contact can trigger an AI-drafted follow-up automatically. You need n8n version 1.104 or newer for the streamable transport.
- Use the official MCP Client node that ships with n8n. No community plugin required.
- Set the URL to https://services.leadconnectorhq.com/mcp/ and the transport to HTTP Streamable.
- HTTP Streamable transport needs n8n 1.104 or newer. Check your version before you start.
Before you start
- A GoHighLevel account (start the 30-day free trial through our link if you do not have one).
- An n8n instance on version 1.104 or newer (cloud or self-hosted).
- A Private Integration Token from GoHighLevel that starts with pit-.
Step by step
Step 1. Confirm your n8n version and create the token
Open n8n and check the version. HTTP Streamable transport landed in n8n 1.104, so anything older will not connect. Update first if you need to.
In GoHighLevel, go to Settings, then Private Integrations, and click Create New Integration. Pick only the scopes your automation needs, save, and copy the token. It starts with pit-. This must be a Private Integration Token, not a legacy API key.
Step 2. Add the MCP Client node
In your workflow, add a node and search for MCP. Pick the official MCP Client node with "Model Context Protocol" in its name, the one that ships with n8n. Do not use a community plugin.
In the node, set the endpoint URL and select HTTP Streamable as the transport, not SSE.
MCP Client node settings URL: https://services.leadconnectorhq.com/mcp/ Transport: HTTP Streamable Auth: Header AuthStep 3. Authenticate with a header credential
Create a Header Auth credential in n8n so the token is stored securely and reused. Set the header name to Authorization and the value to "Bearer " followed by your token.
If your automation always acts on one sub-account, add a second header named locationId with your location ID. This header is optional, so you can leave it out and name the location in the prompt instead.
Header Auth credential Name: Authorization Value: Bearer pit-YOUR_TOKEN Name: locationId Value: YOUR_LOCATION_ID (optional)Step 4. Build one automation end to end
Start simple: a new GoHighLevel contact triggers an AI-drafted follow-up. Add a trigger that fires on a new contact (a GoHighLevel webhook or a poll), then pass the contact to an AI step that drafts a short message.
Send the draft back through the MCP Client node, calling a conversations or messaging tool to create the message on that contact. Leave it as a draft for the first run so you can review the wording before anything goes out.
The MCP server exposes tools across contacts, conversations, calendars, opportunities, and payments. The exact set depends on the scopes on your token and the tool list changes over time, so check the official docs for what is current rather than hardcoding a count.
Step 5. Turn it on and watch the first runs
Activate the workflow and create a test contact in GoHighLevel. Watch the execution in n8n to confirm the trigger fired, the AI step produced text, and the MCP node wrote the message.
Once a few runs look right, switch the message step from draft to send, or add a human approval step in between if you want a person to sign off first.
How to test it worked
Create a test contact in GoHighLevel and open the n8n execution log. A green run with all nodes passing means the trigger, the AI step, and the MCP write all worked.
Open the contact in GoHighLevel and confirm the drafted message or note is there. If it is, your n8n plus GoHighLevel MCP stack is live.
Troubleshooting
The MCP node will not connect: confirm n8n is on 1.104 or newer and that the transport is set to HTTP Streamable, not SSE. Older versions cannot use the streamable transport.
401 unauthorized: the header is wrong. The value must read "Bearer " then your pit- token, and it must be a Private Integration Token, not a legacy API key.
A tool is missing: the scope is not on the token. Edit the integration in GoHighLevel, add the scope, and save. Grant only what the workflow needs, since some scopes can delete data.
Wrong sub-account: set or fix the locationId header, or name the correct location in the prompt you pass to the MCP node.
You need a GoHighLevel account to use the MCP server. Start the 30-day free trial through our link, longer than the standard 14-day trial.
Frequently asked questions
Do I need a custom plugin to connect GoHighLevel to n8n?
No. The official MCP Client node ships with n8n and connects to the GoHighLevel MCP server directly. You do not need a community plugin or a custom node. There is also an official GoHighLevel help article that walks through the no-plugin setup.
What n8n version do I need?
You need n8n 1.104 or newer. That is the version where HTTP Streamable transport became available, which is the transport the GoHighLevel MCP server uses. Older versions cannot connect over this transport.
How do I authenticate the MCP node?
Use a Header Auth credential. Set the Authorization header to "Bearer " followed by your Private Integration Token. Optionally add a locationId header to lock the workflow to one sub-account.
What can the automation actually do?
Anything the token scopes allow across the MCP tool set: read and update contacts, send or draft messages, check calendars, create opportunities, and more. A common first build is a new contact triggering an AI-drafted follow-up message.
Is this better than the native GoHighLevel nodes in n8n?
It is different. The MCP route gives an AI step access to many GoHighLevel tools through one connection, which suits AI-driven automations. For simple fixed actions, the native GoHighLevel nodes are still fine. Many builders use both.