Hugo | MCP API

Quickstart

Updated on March 30, 2026

Connect your tools and services to Hugo using the Model Context Protocol (MCP).

What is MCP?

MCP (Model Context Protocol) is an open standard that defines how AI assistants connect to external tools and data sources. Think of MCP as REST API for AI agents — a universal, language-agnostic protocol that any AI can use to interact with your systems.

MCP allows Hugo to:

Execute actions in your systems (create tickets, update records, process orders)
Fetch data from your databases, CRMs, or APIs
Perform complex workflows by combining multiple tools

Because MCP is a standard protocol, you can build your server in any language. Official SDKs are available for:

Language	SDK
TypeScript	modelcontextprotocol/typescript-sdk
Python	modelcontextprotocol/python-sdk
Java	modelcontextprotocol/java-sdk
C#	modelcontextprotocol/csharp-sdk
PHP	modelcontextprotocol/php-sdk
Ruby	modelcontextprotocol/ruby-sdk
Rust	modelcontextprotocol/rust-sdk

MCP servers are simple to build — you can even use AI coding assistants to generate them. Just describe what tools you need, and the AI can scaffold the entire server for you.

No-code options: Tools like Zapier and n8n can also create MCP servers without writing code, letting you connect Hugo to thousands of apps visually.

For more information, visit the official MCP documentation.

Configuring Hugo

Step 1: Open Hugo Integrations

Go to app.crisp.chat
Navigate to AI Agent > Integrations & MCP
In the External MCP servers section, click Add MCP server

Step 2: Enter Your Server URL

Enter the public URL of your MCP server and click Add MCP Server.

Step 3: Configure the Integration

Name: Give your integration a descriptive name
Description: Describe when Hugo should use this MCP server
Enable: Toggle the integration on
Save: Click Save changes

Your MCP server is now connected!

Hugo will automatically use your tools when relevant to customer conversations.

Server Description

The description is important — it helps Hugo understand when to use your MCP server. Be specific about what the server does and when it should be called.

Example:

This server provides access to the Acme store database. Use it to retrieve customer orders, get product information, and check inventory levels.

Your MCP server can provide its own description. Hugo will fetch it automatically, but you can override it in the integration settings if needed.

Exposing Local Servers

If you're running your MCP server locally, Hugo needs a public URL to reach it. Use a tunneling service:

Using Cloudflare Tunnel:

cloudflared tunnel --url http://localhost:3000

Using ngrok:

ngrok http 3000

Both tools will give you a public URL (e.g., https://abc123.ngrok.io) that you can use in Hugo.

Authentication

Hugo supports three authentication methods:

Method	Use Case
None	Public APIs or non-sensitive data
Bearer Token	API token authentication
Basic Auth	Username/password authentication

To configure authentication:

Select your MCP integration in Hugo
Under Authentication, choose your method
Enter your credentials

For maximum security, combine authentication with Request Signing.

User Verification

You can require users to verify their identity before Hugo calls sensitive tools. When enabled, Crisp sends the user an OTP code via email or SMS.

To enable:

Select your MCP integration
Click Manage on the tool
Toggle Require user identification verification

Recommended for tools that perform actions like canceling orders or updating account information.

Testing Your Integration

Use the Playground to test your MCP integration:

In Hugo, go to Automate > Playground
Select your AI model
Ask questions that should trigger your tools

Example questions:

What's the status of order #12345?

Can you search for wireless headphones?

I need to cancel my last order.

Next Steps

Build your own MCP server with code examples
Set up request signing for security