Connect your AI to the help centre

Our free MCP server lets Claude, ChatGPT, Cursor, Copilot and other assistants search and read all 2,893 GHL Customer Care articles, then answer with links to the exact section. No account, no API key.

https://connect.docs.ghlcustomercare.com/mcp

Streamable HTTP
No sign-in
Read-only
9 tools
2,893 articles

What is MCP?

The Model Context Protocol is an open standard that lets AI assistants use outside tools. Point your assistant at our server and it can look things up in these docs by itself, instead of guessing or relying on old training data.

Grounded answers

Your assistant searches the real articles and quotes the relevant section, so answers match how the platform actually works.

Every answer cited

Each result carries its link on this site, down to the heading, so you can check the source in one click.

Always current

The server reads the same content as this site and refreshes when articles change. Nothing to update on your side.

Set it up in your client

Pick your tool. Most take under a minute.

Claude.ai

Easiest

Works in Claude on the web, and the connector then follows you to the desktop and mobile apps.

1Open Claude and go to Settings → Connectors.
2Click Add custom connector.
3Name it GHL Customer Care Docs and paste https://connect.docs.ghlcustomercare.com/mcp as the remote MCP server URL. Leave the advanced OAuth fields empty: no sign in is needed.
4Click Add. In any chat, open the tools menu (the + or sliders button) and make sure the connector is switched on.
5Ask a question such as “Using the GHL Customer Care docs, how do I connect a custom domain to a funnel?”

On Team and Enterprise plans an Owner adds the connector once under Organization settings → Connectors, then members switch it on. The Free plan allows one custom connector.

Claude Desktop

The quickest route is the same Connectors screen as Claude.ai. Prefer a config file? Bridge it with mcp-remote.

1Option A: in Claude Desktop open Settings → Connectors → Add custom connector and paste https://connect.docs.ghlcustomercare.com/mcp. Done.
2Option B: open Settings → Developer → Edit Config to open claude_desktop_config.json.
3Add the server below (needs Node.js 18 or newer installed), save, and fully quit and reopen Claude Desktop.
4Look for the tools icon in the chat box: the ghl-docs tools are listed there.

claude_desktop_config.json

{
  "mcpServers": {
    "ghl-docs": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://connect.docs.ghlcustomercare.com/mcp"
      ]
    }
  }
}

Claude Code

One command

Add it once and every Claude Code session can look things up in the help centre while it works.

1Run the command below in your terminal.
2Add --scope user to make it available in every project, or --scope project to share it with your team through .mcp.json.
3Check it with claude mcp list, or type /mcp inside a session.

Terminal

claude mcp add --transport http ghl-docs https://connect.docs.ghlcustomercare.com/mcp

All projects

claude mcp add --transport http --scope user ghl-docs https://connect.docs.ghlcustomercare.com/mcp

ChatGPT

ChatGPT connects to remote MCP servers through developer mode, on Plus, Pro, Business, Enterprise and Edu plans on the web.

1Open Settings → Apps & Connectors → Advanced settings and turn on Developer mode.
2Back in Apps & Connectors, click Create.
3Enter the name GHL Customer Care Docs, a short description, and the MCP server URL https://connect.docs.ghlcustomercare.com/mcp. Set Authentication to No authentication, tick the trust checkbox and click Create.
4Start a new chat, open the + menu, choose Developer mode and enable the connector for that chat.
5The server also exposes the search and fetch tools that ChatGPT deep research and company knowledge expect, so it works there too.

On Business and Enterprise workspaces an admin may need to allow developer mode or custom connectors first.

OpenAI API & Agents SDK

Give your own app or agent the help centre as a hosted MCP tool. OpenAI calls the server directly, so there is nothing to run.

1In the Responses API, add a tool of type mcp pointing at the server.
2require_approval: "never" is safe here: every tool is read-only.
3In the Agents SDK, wrap the same config in HostedMCPTool.

Responses API (Python)

from openai import OpenAI

client = OpenAI()
response = client.responses.create(
    model="gpt-5",
    tools=[{
        "type": "mcp",
        "server_label": "ghl_docs",
        "server_url": "https://connect.docs.ghlcustomercare.com/mcp",
        "require_approval": "never",
    }],
    input="How do I connect a custom domain to a funnel? Cite the docs.",
)
print(response.output_text)

Agents SDK (Python)

from agents import Agent, HostedMCPTool, Runner

agent = Agent(
    name="Support assistant",
    instructions="Answer from the help centre and cite the article URLs.",
    tools=[HostedMCPTool(tool_config={
        "type": "mcp",
        "server_label": "ghl_docs",
        "server_url": "https://connect.docs.ghlcustomercare.com/mcp",
        "require_approval": "never",
    })],
)
print(Runner.run_sync(agent, "How do I register for A2P 10DLC?").final_output)

Cursor

Use the one-click install, or add the server to your MCP config by hand.

1Click Add to Cursor below and confirm, or open Cursor Settings → Tools & MCP → New MCP Server.
2If editing by hand, add the server to ~/.cursor/mcp.json (all projects) or .cursor/mcp.json (this project).
3The server shows a green dot and its tools once connected. Ask in Agent chat.

mcp.json

{
  "mcpServers": {
    "ghl-docs": {
      "url": "https://connect.docs.ghlcustomercare.com/mcp"
    }
  }
}

Add to Cursor

VS Code & GitHub Copilot

Copilot Chat in agent mode can call the help centre tools directly (VS Code 1.102 or newer).

1Click Install in VS Code below, or run MCP: Add Server from the Command Palette, pick HTTP and paste https://connect.docs.ghlcustomercare.com/mcp.
2To share it with your team, commit the config below as .vscode/mcp.json.
3Open Copilot Chat, switch to Agent mode and check the tools picker lists the ghl-docs tools.

.vscode/mcp.json

{
  "servers": {
    "ghl-docs": {
      "type": "http",
      "url": "https://connect.docs.ghlcustomercare.com/mcp"
    }
  }
}

Install in VS Code

Windsurf

Cascade picks up remote servers from its MCP config file.

1Open Windsurf Settings → Cascade → MCP Servers and choose View raw config (the file is ~/.codeium/windsurf/mcp_config.json).
2Add the server below. Windsurf uses serverUrl for remote servers.
3Save and press Refresh in the MCP panel.

mcp_config.json

{
  "mcpServers": {
    "ghl-docs": {
      "serverUrl": "https://connect.docs.ghlcustomercare.com/mcp"
    }
  }
}

Gemini CLI

Add it from the command line or in settings.json.

1Run the command below, or add the server to ~/.gemini/settings.json.
2Start gemini and type /mcp to see the tools.

Terminal

gemini mcp add --transport http ghl-docs https://connect.docs.ghlcustomercare.com/mcp

settings.json

{
  "mcpServers": {
    "ghl-docs": {
      "httpUrl": "https://connect.docs.ghlcustomercare.com/mcp"
    }
  }
}

Any other client

Anything that speaks MCP over Streamable HTTP can connect with just the URL.

1Transport: Streamable HTTP. URL: https://connect.docs.ghlcustomercare.com/mcp. Authentication: none.
2For a client that only supports local (stdio) servers, bridge it with npx -y mcp-remote https://connect.docs.ghlcustomercare.com/mcp.
3Test it with the MCP Inspector: npx @modelcontextprotocol/inspector, then connect to the URL.

Tool reference

Your assistant chooses these for you. Every tool is read-only and every result includes the article's link on this site.

Tool	What it does	Inputs
find_answer	Answers a question with the most relevant sections from across the help centre, each with a link straight to that heading.	question, category?, limit?
search_docs	Ranked search across all 2,893 articles, with snippets, the best-matching section and paging.	query, category?, folder?, limit?, offset?
get_article	The complete article as Markdown, with every screenshot described in words, plus its sections and FAQs.	article (id, URL, path or slug), section?
list_categories	Every category and folder with article counts. A good place to start browsing.	include_folders?
get_category	One category with all of its folders and the articles inside them.	category
list_articles	Paged article lists by category or folder, in site order, most recently updated or A to Z.	category?, folder?, sort?, offset?, limit?
get_related_articles	Articles on the same topic and from the same folder, for next steps.	article, limit?
search / fetch	The pair ChatGPT deep research and connectors expect: search returns ids, fetch returns full text.	query / id

Also available: every article as a resource, the llms.txt index, and three ready-made prompts (troubleshoot, how_to, explain_feature) that show up as slash commands in clients that support them.

Things to ask

Once connected, just ask. Mentioning the docs nudges the assistant to use them.

Using the GHL Customer Care docs, walk me through setting up a round robin calendar.
My SMS messages are failing with error 30007. What does it mean and how do I fix it?
What do I need to prepare before registering for A2P 10DLC? Cite the articles.
Compare workflow triggers for form submissions and survey submissions.
Find every article about connecting a custom domain and summarise the steps.
Write a checklist for onboarding a new sub-account, based on the help centre.

Prefer plain text?

For crawlers, RAG pipelines or pasting into a chat, the same content is published as static files.

/llms.txt

Index of every article with a one-line summary, in the llmstxt.org format.

/llms/<category>.txt

Every article in a category, in full, as one Markdown file. Linked from llms.txt.

Any article + .md

Add .md to any article URL for a clean Markdown copy with screenshots described.

Frequently asked questions

Does it cost anything, or need an account or API key?

No. The MCP server is free and public, like the help centre itself. There is nothing to sign up for and no key to paste.

Can it see or change anything in my account?

No. It only reads the public help articles, and every tool is marked read-only. It has no access to your GHL Customer Care account, contacts or data.

Why is there no sign-in step?

Because the docs are public, the server needs no authentication. If a client asks about OAuth, leave those fields empty or choose "No authentication".

How current are the answers?

The server reads the same content as this site and reloads automatically whenever the help centre is updated, so answers match what you see here.

Are my questions stored?

The server only sees the tool calls your assistant makes, such as a search phrase. It does not store them or link them to you. Your conversation itself stays with your AI provider.

Is there a usage limit?

Each connection can make 120 requests a minute, far more than a person chatting will use. If you hit it, wait a minute and try again.

It will not connect. What should I check?

Make sure the URL is exactly https://connect.docs.ghlcustomercare.com/mcp with no trailing text, that the transport is Streamable HTTP (not SSE), and that authentication is set to none. Restart the client after editing a config file. Still stuck? Email support@ghlcustomercare.com.

What is the difference between this and llms.txt?

llms.txt is a static index of every article for crawlers and for pasting into a chat. The MCP server is live: your assistant can search, read and cite articles on demand while it answers.

Connect a client Talk to support