Using Huwise's MCP server

MCP server functionality is not automatically included in your plan. For more information, please contact our support team.

The Model Context Protocol (MCP) is an open standard that lets AI systems and applications talk to each other. Huwise's MCP server lets AI agents query your data portal — searching and analyzing assets (datasets, visualizations, showcases, pages, reuses) — you can use an AI assistant to search and analyze assets in your Huwise portal without having to make the API requests yourself.

This article explains how to connect the Huwise MCP server to commonly used MCP clients.

We're working on making the Huwise MCP server available as a connector in ChatGPT and Mistral AI, but these connectors are not yet available.

For more detailed technical documentation of the Huwise MCP server, including its tools, parameters, and endpoints, see mcp.huwise.com/docs.

Authentication

Every request to the MCP server needs to identify which portal to query and who's asking, so you must provide your portal URL and API key. These are sent as HTTP headers or, if your MCP client doesn't support custom headers, as query parameters.

Header	Required	Description
`X-Domain`	Yes	The URL of your Huwise data portal (e.g. `https://your-portal.com`)
`X-Apikey`	Yes	Your API key

To define an API key, see Managing API keys.

Configuring Claude Desktop

Prerequisites

Claude Desktop installed (Windows, macOS or Linux)
Node.js version 18 or higher
npm or yarn for package management
A valid API key

Locate the configuration file

macOS:

~/Library/Application Support/Claude/claude_desktop_config.json

Windows:

%APPDATA%\Claude\claude_desktop_config.json

Linux:

~/.config/Claude/claude_desktop_config.json

MCP configuration

Open the configuration file for your operating system and add the following configuration. Replace the placeholders with your Huwise portal URL and API key.

If you've already configured other MCP servers, add the huwise entry to your existing mcpServers object rather than replacing the entire file.

{
  "mcpServers": {
    "huwise": {
      "command": "npx",
      "args": [
        "-y",
        "supergateway@3.4.3",
        "--streamableHttp",
        "https://mcp.huwise.com/mcp",
        "--header",
        "x-domain: <URL OF YOUR HUWISE PORTAL>",
        "--header",
        "x-apikey: <API KEY>"
      ]
    }
  }
}

Save the file and restart Claude Desktop for the new server to appear.

Configuring Cursor

Prerequisites

Cursor installed (Windows, macOS or Linux)
A valid API key

Locate the configuration file

Cursor reads MCP servers from an mcp.json file, at either scope:

Project: .cursor/mcp.json at the root of your project
Global: ~/.cursor/mcp.json in your home directory

MCP configuration

Cursor connects directly to the Huwise MCP server over HTTP, so you don't need to install a local proxy.

Open the mcp.json file you want to use and add the following configuration. Replace the placeholders with your Huwise portal URL and API key.

If you've already configured other MCP servers, add the huwise entry to your existing mcpServers object rather than replacing the entire file.

{
  "mcpServers": {
    "huwise": {
      "url": "https://mcp.huwise.com/mcp",
      "headers": {
        "x-domain": "<URL OF YOUR HUWISE PORTAL>",
        "x-apikey": "<API KEY>"
      }
    }
  }
}

Reload the MCP settings or restart Cursor for the new server to appear.

Configuring a Microsoft Copilot assistant

Prerequisites

A Microsoft 365 Copilot license for users and makers
Permissions to access Copilot Studio

Creating the agent

In Copilot Studio, add the Huwise MCP server as a tool for your agent. Follow Microsoft's instructions for connecting to an existing MCP server.

Swagger configuration

Use the following Swagger definition when configuring the Huwise MCP server as a tool in Copilot Studio.

swagger: '2.0'
info:
  title: Huwise MCP server
  version: 2.0.0
host: mcp.huwise.com
schemes:
  - https
paths:
  /mcp:
    post:
      summary: Huwise MCP Server
      x-ms-agentic-protocol: mcp-streamable-1.0

Required configuration policies

Add one policy per header your agent needs to send:

Setting	X-Domain	X-Apikey
Type	Set HTTP header	Set HTTP header
Header name	`x-domain`	`x-apikey`
Value	Your portal URL	Your chosen API key
Action	Override	Override
Applies to	Request	Request

Available tools

The Huwise MCP server provides four tools for searching and analyzing your portal's assets, using the Explore API and ODSQL:

Tool	What it's for	Typical usage
`analyze_catalog`	Analyze the structure and composition of a portal's catalog using ODSQL (`select`, `where`, `order_by`, `group_by`, `facet`…)	"How many assets are there?", "What's the breakdown by type?", "What changed recently?"
`search_assets`	Search a portal's assets by semantic similarity	Discovering assets by topic when you don't know the exact one
`get_asset`	Retrieve the full metadata of a single asset by its slug	You already know which asset you want and need its details, schema or sample records
`search_asset_records`	Query and filter the records of a specific asset by its slug, using ODSQL	Filtering, sorting or paginating within a known asset's data

These are the tools exposed by the default (latest) version of the server. Earlier tool sets (beta, old) remain available for backwards compatibility — see the full reference for details.

Note that tool names changed with the server's v2.0 update — explore_get_datasets, explore_get_dataset and explore_get_dataset_records are now search_assets, get_asset and search_asset_records respectively, and now cover all asset types (not just datasets). If your client still references the old names, update it to point at the tools above.