Skip to main content

MCP server for Google Tag Manager

Project description

GTM MCP Server

PyPI version Python Version License: MIT

A Model Context Protocol (MCP) server that enables Claude to interact with Google Tag Manager.

Table of Contents

Features

⬆ top

  • List GTM accounts and containers
  • Manage tags, triggers, and variables
  • Create and publish container versions
  • Full workspace management

🚀 Quick Start

⬆ top

Prerequisites

  • Python 3.10 or higher
  • Claude Desktop (or any MCP-compatible client like Cursor)
  • A Google account with access to Google Tag Manager

📋 Complete Setup Guide

⬆ top

Part 1: Install the Package

pip install gtm-mcp

See PyPi


Part 2: Choose Your Authentication Method

⬆ top

GTM MCP supports two authentication methods. Choose the one that best fits your needs:

Method A: Application Default Credentials (Recommended for Servers)

Best for:

  • ✅ Server environments and automation
  • ✅ No browser interaction required
  • ✅ Flexible credential sources
  • ✅ Better for CI/CD pipelines
  • ✅ Works on Google Cloud (GCE, GKE, Cloud Run)

Pros:

  • No OAuth consent screen setup required
  • No browser popup for authorization
  • Automatic credential discovery from multiple sources
  • Automatic token refresh
  • Works with gcloud CLI, service account files, or cloud environment

Cons:

  • Requires initial credential setup (service account or gcloud CLI)
  • Need to manually grant service account access to GTM

→ Go to Application Default Credentials Setup


Method B: OAuth 2.0 (Original Method)

Best for:

  • ✅ Personal desktop use
  • ✅ Interactive environments
  • ✅ When you want user-level permissions

Pros:

  • Uses your personal Google account
  • Natural authorization flow
  • Direct access through browser

Cons:

  • Requires OAuth consent screen setup
  • Requires browser interaction on first use
  • More complex initial setup

→ Go to OAuth 2.0 Setup


Application Default Credentials Setup (Method A)

⬆ top

Application Default Credentials (ADC) is Google's recommended way to authenticate applications. It automatically discovers credentials from multiple sources in the following order:

  1. GOOGLE_APPLICATION_CREDENTIALS environment variable pointing to a service account JSON file
  2. gcloud CLI credentials (gcloud auth application-default login)
  3. Cloud environment service accounts (GCE, GKE, Cloud Run, etc.)

Choose the option that works best for you:

Option 1: Service Account File (Most Common)

Step 1: Create a Google Cloud Project
  1. Go to Google Cloud Console
  2. Click on the project dropdown (top left)
  3. Click "New Project"
  4. Enter a project name (e.g., "GTM MCP Service")
  5. Click "Create"
  6. Wait for the project to be created and select it
Step 2: Enable Tag Manager API
  1. In your project, go to "APIs & Services""Library"
  2. Search for "Tag Manager API"
  3. Click on it and click "Enable"
  4. Wait for it to enable (may take a minute)
Step 3: Create Service Account
  1. Go to "APIs & Services""Credentials"
  2. Click "Create Credentials""Service Account"
  3. Fill in the service account details:
    • Service account name: gtm-mcp-service
    • Service account ID: (auto-filled)
    • Description: Service account for GTM MCP Server
  4. Click "Create and Continue"
  5. Skip the optional steps and click "Done"
Step 4: Create and Download Service Account Key
  1. Click on the service account you just created
  2. Go to the "Keys" tab
  3. Click "Add Key""Create New Key"
  4. Select "JSON" format
  5. Click "Create"
  6. The JSON file will download automatically
  7. Save this file securely - you'll need its path later
  8. Copy the service account email from the JSON file (looks like: gtm-mcp-service@your-project.iam.gserviceaccount.com)
Step 5: Grant Service Account Access to GTM
  1. Go to Google Tag Manager
  2. Select your GTM account
  3. Click "Admin" in the top navigation
  4. Under "Account", click "User Management"
  5. Click the "+" button to add a user
  6. Enter the service account email from Step 4
  7. Select appropriate permissions:
    • Read: For read-only access
    • Edit: For creating/modifying tags, triggers, variables
    • Approve: For publishing container versions
    • Publish: For full publishing rights
  8. Click "Invite"
Step 6: Configure Claude Desktop (Service Account File)

Edit your Claude Desktop config file:

  • Linux: ~/.config/Claude/claude_desktop_config.json
  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
  • Windows: %APPDATA%\Claude\claude_desktop_config.json

Add your configuration:

{
  "mcpServers": {
    "gtm-mcp": {
      "command": "gtm-mcp",
      "env": {
        "GTM_AUTH_METHOD": "service_account",
        "GOOGLE_APPLICATION_CREDENTIALS": "/path/to/your/service-account-key.json"
      }
    }
  }
}

Important: Replace /path/to/your/service-account-key.json with the actual path to your downloaded JSON file.

Example paths:

  • Linux/macOS: /home/username/gtm-service-account.json
  • Windows: C:\\Users\\YourName\\gtm-service-account.json

Option 2: gcloud CLI Credentials (Simplest for Local Development)

If you have the Google Cloud SDK installed and want to use your own Google account credentials:

Step 1: Install Google Cloud SDK

If not already installed, follow the Google Cloud SDK installation guide.

Step 2: Authenticate with Application Default Credentials
gcloud auth application-default login

This will open a browser window to authenticate with your Google account and save credentials locally.

Step 3: Grant Your Account Access to GTM

Ensure your Google account has the appropriate permissions in Google Tag Manager (see Step 5 in Option 1, but use your Google account email instead of a service account email).

Step 4: Configure Claude Desktop (gcloud CLI)

Edit your Claude Desktop config file:

{
  "mcpServers": {
    "gtm-mcp": {
      "command": "gtm-mcp",
      "env": {
        "GTM_AUTH_METHOD": "service_account"
      }
    }
  }
}

Note: No GOOGLE_APPLICATION_CREDENTIALS is needed when using gcloud CLI credentials.


Option 3: Cloud Environment (GCE, GKE, Cloud Run)

If you're running in a Google Cloud environment, ADC will automatically use the attached service account.

Step 1: Attach Service Account to Your Resource

When creating your GCE instance, GKE pod, or Cloud Run service, attach a service account with the necessary permissions.

Step 2: Grant Service Account Access to GTM

Follow Step 5 from Option 1 to grant the service account access to your GTM account.

Step 3: Configure Your Application

Set the environment variable:

GTM_AUTH_METHOD=service_account

No GOOGLE_APPLICATION_CREDENTIALS is needed - ADC will automatically discover the attached service account.


Restart and Test

  1. Restart Claude Desktop completely
  2. Ask Claude: "List my GTM accounts"
  3. ADC will authenticate automatically - no browser popup needed!

→ Skip to Available Tools


OAuth 2.0 Setup (Method B)

⬆ top

Step 1: Create a Google Cloud Project

  1. Go to Google Cloud Console
  2. Click on the project dropdown (top left)
  3. Click "New Project"
  4. Enter a project name (e.g., "My GTM MCP Server")
  5. Click "Create"
  6. Wait for the project to be created and select it

Step 2: Enable Tag Manager API

  1. In your project, go to "APIs & Services""Library"
  2. Search for "Tag Manager API"
  3. Click on it and click "Enable"
  4. Wait for it to enable (may take a minute)

Step 3: Configure OAuth Consent Screen

  1. Go to "APIs & Services""OAuth consent screen"
  2. Select "External" (unless you have a Google Workspace)
  3. Click "Create"
  4. Fill in required fields:
    • App name: My GTM MCP (or whatever you like)
    • User support email: Your email
    • Developer contact email: Your email
  5. Click "Save and Continue"
  6. Click "Update" then "Save and Continue"
  7. Add your email as a test user
  8. Click "Save and Continue"

Step 4: Create OAuth Credentials

  1. Go to "APIs & Services""Credentials"
  2. Click "Create Credentials""OAuth client ID"
  3. Select "Desktop app" as the application type
  4. Enter a name: "GTM MCP Desktop Client"
  5. Click "Create"
  6. IMPORTANT: A dialog appears with your credentials - DO NOT CLOSE IT YET

Step 5: Save Your Credentials

From the dialog that appeared:

  1. Copy the Client ID (looks like: 123456789-abc123.apps.googleusercontent.com)
  2. Copy the Client secret (looks like: GOCSPX-...)
  3. Note your Project ID from the Google Cloud Console (top bar, next to project name)
  4. Save these somewhere safe - you'll need them in the next step

You can also download the JSON file, but you only need the three values above.


Step 6: Configure Claude Desktop (OAuth 2.0)

⬆ top Edit your Claude Desktop config file:

  • Linux: ~/.config/Claude/claude_desktop_config.json
  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
  • Windows: %APPDATA%\Claude\claude_desktop_config.json
  • Claude Code: ~/.claude.json

Add your credentials:

{
  "mcpServers": {
    "gtm-mcp": {
      "command": "gtm-mcp",
      "env": {
        "GTM_AUTH_METHOD": "oauth",
        "GTM_CLIENT_ID": "your-client-id.apps.googleusercontent.com",
        "GTM_CLIENT_SECRET": "GOCSPX-your-client-secret",
        "GTM_PROJECT_ID": "your-project-id"
      }
    }
  }
}

Replace the values with your actual credentials from Step 5.

Note:

  • If you have other MCP servers configured, just add the "gtm-mcp" entry to the existing "mcpServers" object.
  • The GTM_AUTH_METHOD can be omitted as oauth is the default for backward compatibility.

Step 7: Restart and Authorize (OAuth 2.0)

⬆ top

  1. Restart Claude Desktop completely (close and reopen)

  2. Ask Claude to use a GTM tool (e.g., "List my GTM accounts")

  3. First-time authorization - a browser window will open automatically:

    • Sign in with your Google account
    • You'll see "Google hasn't verified this app" warning
    • Click "Advanced""Go to [Your App Name] (unsafe)"
    • This is safe because you created the app yourself
    • Grant the requested permissions
    • You'll see "The authentication flow has completed"
    • Return to Claude Desktop
  4. Your authorization is saved locally - you won't need to do this again!


🛠️ Available Tools

⬆ top
Once configured, Claude will have access to these GTM tools:

Tool Description
gtm_list_accounts List all your GTM accounts
gtm_list_containers List containers in an account
gtm_list_tags List tags in a workspace
gtm_get_tag Get detailed configuration of a specific tag
gtm_create_tag Create a new tag
gtm_update_tag Update an existing tag
gtm_list_triggers List triggers in a workspace
gtm_create_trigger Create a new trigger
gtm_list_variables List variables in a workspace
gtm_create_variable Create a new variable (constant, data layer, cookie, URL, etc.)
gtm_publish_container Create and publish a new container version

🔐 How Authentication Works

⬆ top
This MCP server supports two authentication methods:

Method A: Application Default Credentials (ADC)

With Application Default Credentials (ADC) authentication:

  1. ADC automatically discovers credentials from multiple sources:
    • GOOGLE_APPLICATION_CREDENTIALS environment variable pointing to a service account JSON file
    • Google Cloud SDK credentials (gcloud auth application-default login)
    • Service account attached to GCE/GKE/Cloud Run instances
    • Other Google Cloud environments
  2. Grant the account access to your GTM account
  3. Automatic authentication - no browser interaction required
  4. Tokens are managed automatically by Google's auth library

Benefits:

  • ✅ No browser interaction required
  • ✅ Perfect for server environments
  • ✅ Automatic token refresh
  • ✅ Flexible credential sources
  • ✅ Works seamlessly in Google Cloud environments
  • ✅ Simple for local development with gcloud CLI

How it works: When you set GTM_AUTH_METHOD=service_account, the library uses google.auth.default() to automatically discover credentials. This is the same approach used by the official Google Analytics MCP server and is Google's recommended authentication method.

Method B: OAuth 2.0 Authentication (Default)

With OAuth 2.0 authentication:

  1. You create OAuth credentials in your own Google Cloud project
  2. You configure those credentials in Claude Desktop
  3. First use: Browser opens to authorize access to your GTM account
  4. Your tokens are saved locally on your machine (~/.gtm-mcp/token.json) for future use

Benefits:

  • ✅ You maintain full control over the OAuth app
  • ✅ No shared credentials between users
  • ✅ You can revoke access anytime
  • ✅ Your credentials stay private
  • ✅ Compliant with Google's OAuth policies

Upgrade

Run pip install --upgrade gtm-mcp


❓ Troubleshooting

⬆ top

Application Default Credentials (ADC) Issues

"Failed to load Application Default Credentials" Error

Problem: The MCP server can't discover any credentials using ADC.

Solution: Choose one of these credential sources:

  1. Using a service account file:

    • Set GOOGLE_APPLICATION_CREDENTIALS to the path of your service account JSON file
    • Ensure the path is absolute and points to a valid JSON file
    • On Windows, use double backslashes: C:\\Users\\...
    • Restart Claude Desktop after editing the config
  2. Using gcloud CLI:

    • Run gcloud auth application-default login
    • Follow the browser flow to authenticate
    • Ensure your Google account has GTM access
    • Restart Claude Desktop
  3. Running on Google Cloud:

    • Ensure your GCE/GKE/Cloud Run resource has a service account attached
    • Grant the service account access to your GTM account

"Service account has no access" Error

Problem: The service account or user account can't access your GTM account.

Solution:

  1. Go to GTM → Admin → User Management
  2. Verify the account email is listed with appropriate permissions
  3. For service accounts: add the service account email from your JSON file
  4. For gcloud CLI: add your Google account email
  5. Grant at least "Read" permission (or higher based on your needs)

Which Authentication Method Am I Using?

Check your claude_desktop_config.json:

  • If you see GTM_AUTH_METHOD: "service_account"Application Default Credentials (ADC)
  • If you see GTM_CLIENT_ID, GTM_CLIENT_SECRETOAuth 2.0
  • If no GTM_AUTH_METHOD is set → OAuth 2.0 (default)

OAuth 2.0 Issues

"Missing required OAuth credentials" Error

Problem: The MCP server can't find your OAuth credentials.

Solution: Make sure you:

  • Set GTM_AUTH_METHOD=oauth (or omit it, as oauth is default)
  • Set GTM_CLIENT_ID, GTM_CLIENT_SECRET, and GTM_PROJECT_ID correctly in claude_desktop_config.json
  • Restarted Claude Desktop after editing the config
  • Used the correct format (no extra quotes in JSON)
  • The config file is valid JSON (use a JSON validator if unsure)

"Google hasn't verified this app" Warning

Problem: Google shows a security warning during first authorization.

Solution: This is completely normal for personal OAuth apps. Since you created the OAuth app yourself, Google shows this warning.

To proceed: Click "Advanced""Go to [App Name] (unsafe)"

This is safe because you control the app.

General Issues

Can't Access GTM Accounts

Possible causes:

  • Your Google account doesn't have access to any GTM accounts (OAuth)
  • Service account doesn't have GTM permissions (Service Account)
  • You didn't grant all requested permissions during authorization (OAuth)
  • Tag Manager API isn't enabled in your Google Cloud project

Solution:

  1. Verify your Google account or service account has GTM access
  2. For OAuth: Re-authorize by deleting ~/.gtm-mcp/token.json and trying again
  3. For ADC/Service Account: Check GTM user management for the service account or user account email
  4. Check that Tag Manager API is enabled in Google Cloud Console

Connection Issues

Debugging steps:

  1. Verify Claude Desktop is completely restarted
  2. Check Claude Desktop logs for MCP server errors
  3. Verify gtm-mcp command works: run gtm-mcp in terminal
  4. Check your config file is valid JSON
  5. Ensure all required environment variables are set correctly

Package Not Found After Install

Problem: gtm-mcp command not found after installation.

Solution:

# Ensure pip install location is in PATH
pip install --user gtm-mcp

# Or use pipx for isolated installation
pipx install gtm-mcp

Revoking Access

For OAuth 2.0:

  1. Go to Google Account Permissions
  2. Find your app name in the list
  3. Click "Remove access"
  4. Delete the local token file: rm ~/.gtm-mcp/token.json

You can re-authorize anytime by using any GTM tool in Claude again.

For Application Default Credentials (ADC):

If using a service account file:

  1. Go to GTM → Admin → User Management
  2. Find the service account email
  3. Click the remove button to revoke access
  4. Optionally delete the service account key from Google Cloud Console

If using gcloud CLI:

  1. Go to GTM → Admin → User Management
  2. Find your Google account email
  3. Click the remove button to revoke access
  4. Optionally revoke ADC locally: gcloud auth application-default revoke

🔒 Security Notes

⬆ top

For OAuth 2.0:

  • Your OAuth credentials are yours alone - keep them private
  • Never share your Client Secret - treat it like a password
  • Your access tokens are stored locally: ~/.gtm-mcp/token.json
  • You can regenerate credentials anytime in Google Cloud Console
  • You can revoke access anytime from your Google account settings

For Application Default Credentials (ADC):

  • Keep your JSON key file secure - it provides full access
  • Never commit the JSON file to version control - add it to .gitignore
  • Use restrictive file permissions - chmod 600 service-account.json on Unix/Linux
  • Store the file in a secure location with limited access
  • You can create new keys and delete old ones in Google Cloud Console
  • Regularly rotate service account keys for better security

General:

  • This server only accesses GTM - no other Google services
  • Credentials are only used for Google Tag Manager API authentication
  • No credentials are sent to any third-party services

💻 Development

Running Tests

pip install -e ".[dev]"
pytest

📝 License

⬆ top see LICENSE file for details


🤝 Contributing

⬆ top Contributions are welcome! Please:

  1. Fork the repository
  2. Create a feature branch
  3. Make your changes
  4. Submit a pull request

For bugs and feature requests, please open an issue.


🆘 Getting Help

⬆ top If you encounter issues:

  1. Check the Troubleshooting section above
  2. Review Claude Desktop logs for error messages
  3. Verify your Google Cloud project has Tag Manager API enabled
  4. Ensure environment variables are set correctly in the config
  5. Open an issue on GitHub with:
    • Your operating system
    • Python version (python --version)
    • Error messages from Claude Desktop logs
    • Steps to reproduce the issue

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

unboundai_gtm_mcp-0.0.6.tar.gz (50.6 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

unboundai_gtm_mcp-0.0.6-py3-none-any.whl (51.1 kB view details)

Uploaded Python 3

File details

Details for the file unboundai_gtm_mcp-0.0.6.tar.gz.

File metadata

  • Download URL: unboundai_gtm_mcp-0.0.6.tar.gz
  • Upload date:
  • Size: 50.6 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/2.2.1 CPython/3.11.5 Windows/10

File hashes

Hashes for unboundai_gtm_mcp-0.0.6.tar.gz
Algorithm Hash digest
SHA256 b00b5cc098fa519603e955591cafc30e4aea95bc049704da3f1a524e691082f4
MD5 28ccfaab2cadbf962ff603a9d40d3f97
BLAKE2b-256 36acb224103976ad82fccbe9d49e89818b3b0ae3d705e643916ea2e8547cfb55

See more details on using hashes here.

File details

Details for the file unboundai_gtm_mcp-0.0.6-py3-none-any.whl.

File metadata

  • Download URL: unboundai_gtm_mcp-0.0.6-py3-none-any.whl
  • Upload date:
  • Size: 51.1 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/2.2.1 CPython/3.11.5 Windows/10

File hashes

Hashes for unboundai_gtm_mcp-0.0.6-py3-none-any.whl
Algorithm Hash digest
SHA256 5a7c6755bbb45eb3dec3e7a54aefab5555c3b78648a5354377b9e49c5dbec2cd
MD5 a9536a0563da6ca0e56a5290f7b31dfb
BLAKE2b-256 ece834ef63dc9ea210e3b4d3b23971df25b2c1163f773afa9ab03a6008a16296

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page