ePrivacy and GPDR Cookie Consent by Cookie Consent Skip to main content

Using the Meiro MCP Server with Claude Desktop

The Model Context Protocol (MCP) Server allows Meiro CDP to integrate with AI tools such as Claude Desktop. This makes it possible to search customers, build segments, retrieve attributes, and run analytics directly from your AI client.


Prerequisites

Before installing the MCP server, you will need:

  • Claude Desktop installed (available for Windows and macOS from Claude’s official download page).

  • Python (version 3.8 or newer) installed for your operating system.

  • A valid Meiro CDP account (domain, username, password).


Quick Start

1. Install prerequisites

  • Install Claude Desktop (Windows/macOS)
  • Install Python 3.8+
  • Install uv:
    curl -LsSf https://astral.sh/uv/install.sh | sh   # macOS/Linux  
    powershell -c "irm https://astral.sh/uv/install.ps1 | iex"   # Windows

2. Install MCP server

uvx pip install meiro-mcp

3. Configure Claude Desktop

  • Name: Meiro MCP
  • Command: uvx
  • Arguments: meiro-mcp
  • Env vars: 
    {
      "MEIRO_DOMAIN": "https://your-instance.meiro.io",
      "MEIRO_USERNAME": "your-username",
      "MEIRO_PASSWORD": "your-password"
    }

4. Test & use

  • Restart Claude Desktop and run a customer search via MCP.

Full setup guide (step-by-step)

Step 1 – Install uv

The uv package and project manager are recommended for running Python-based MCP servers.

macOS / Linux:

curl -LsSf https://astral.sh/uv/install.sh | sh

Windows (PowerShell):

powershell -c "irm https://astral.sh/uv/install.ps1 | iex"

Check the installation:

uv --version

Step 2 – Create a Virtual Environment (Optional but Recommended)

Creating a dedicated environment helps avoid conflicts with other Python packages:

uv venv

Step 3 – Install the Meiro MCP Server

Using uvx, install the Meiro MCP server package:

uvx pip install meiro-mcp

Step 4 – Configure Claude Desktop

You can connect MCP to Claude Desktop either via the in-app UI or by editing the configuration file.

Option 1: Configure via Claude Desktop Settings

  1. Open Claude Desktop and go to Settings → MCP Servers.
  2. Click Add MCP Server.
  3. Fill in:
    • Name: Meiro MCP
    • Command: uvx
    • Arguments: meiro-mcp
  4. Add environment variables:
    • MEIRO_DOMAIN – your Meiro CDP domain (with https://)
    • MEIRO_USERNAME – valid Meiro CDP username
    • MEIRO_PASSWORD – password for that user
  5. Save changes and restart Claude Desktop.

Option 2: Configure by Editing the Config File

If you prefer editing the configuration JSON directly:

Windows:

%APPDATA%\Claude\claude_desktop_config.json

macOS:

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

Add or merge the following JSON snippet:

{
  "mcpServers": {
    "Meiro MCP": {
      "type": "stdio",
      "command": "uvx",
      "args": ["meiro-mcp"],
      "env": {
        "MEIRO_DOMAIN": "https://your-instance.meiro.io",
        "MEIRO_USERNAME": "your-username",
        "MEIRO_PASSWORD": "your-password"
      }
    }
  }
}

Replace values with your actual Meiro CDP credentials. The user must exist in the instance and have permission to access customer data, segmentation, and analytics.


Capabilities

The Meiro MCP server opens up a powerful set of tools that allow you to work with customer data in a natural, conversational way through Claude Desktop. Here’s what’s possible:

Customer Management

  • Search Customers – Find customers in the CDP simply by typing a name, email, phone number, or other search text.
    Example: Quickly pull up a customer profile when handling a support request.

  • Get Customer Attributes – Retrieve all available data for a specific customer using their entity ID.
    Example: Instantly view a customer’s purchase history, preferences, and activity.

Segmentation

  • List Segments – See all customer segments currently defined in your CDP.
    Example: Review “High-Value Customers” or “Inactive 90 Days” groups.

  • Create Segment – Build new customer segments using a DSL that supports complex conditions based on attributes, operations, and logical operators.
    Example: “Customers who purchased in the last 30 days AND live in London OR Paris.”

  • Get Segment Details – See customer counts, applied conditions, and insights for any segment.
    Example: Check the size and definition of a “Loyalty Program Members” segment.

Data Structure

  • List Attributes – Get all available customer attributes, including compound attributes with sub-attributes. Only enabled attributes are shown.
    Example: Discover what customer data points are tracked, from basic contact info to behavioral data.

  • List Events – See all tracked events with IDs, names, descriptions, and example payloads.
    Example: View how “Purchase Completed” events are structured in the CDP.

Analytics & Funnels

  • List Funnels – Retrieve funnels grouped by category with definitions and steps.
    Example: See your “Onboarding Funnel” to understand the customer journey.

  • Get Funnel Group Data – Get conversion counts over date ranges, with optional segment filters.
    Example: Measure how many “New Sign-Ups” completed all onboarding steps last month.


Debugging MCP in Claude Desktop

Claude Desktop includes built-in debugging to help diagnose connectivity issues:

  • Go to Help → Developer Tools → MCP Debugging.

  • Check the logs for MCP connection attempts and error messages.

If errors occur, see the Troubleshooting section below.


Troubleshooting

1. MCP Server Does Not Start

  • Cause: uv not installed or not in PATH.
  • Fix: Reinstall uv and restart your terminal.

2. Authentication Failed

  • Cause: Wrong username/password or insufficient permissions.
  • Fix: Update credentials and verify the account exists in your Meiro instance.

3. Invalid Domain

  • Cause: Domain missing https:// or inaccessible from network.
  • Fix: Use full URL, e.g., https://your-instance.meiro.io.

4. Feature Not Working in Claude

  • Cause: Outdated MCP server or feature disabled in Meiro.
  • Fix: uv pip install --upgrade meiro-mcp

5. Slow Responses

  • Cause: Large queries or heavy filters.
  • Fix: Narrow the scope — shorter date ranges, fewer segment conditions.