WAS GTM MCP

Manage Google Tag Manager from Claude Desktop, Cursor, Google Antigravity, or any MCP-compatible client using natural language. 19 tools cover the full GTM v2 API: accounts, containers, workspaces, tags, triggers, variables, versions, environments, destinations, and more.

100 percent local. Your credentials stay on your machine. Built by Abdullah Al Masum — Web Analytics Solution (WAS). MIT licensed.

Step 1 — Create Google Cloud OAuth Credentials

1. Go to the Google Cloud Console Credentials Page.

2. Create a Project (or select an existing one) and enable the Google Tag Manager API.

3. Go to OAuth consent screen:

   • User Type: External -> Create

   • App name: anything (e.g. My GTM MCP)

   • User support email: your email

   • Developer contact email: your email -> Save

   • Under Test users, click Add Users and add the Google account you will sign in with.

4. Open Google Credentials Page:

   • Create credentials -> OAuth client ID

   • Application type: Desktop app

   • Name: anything -> Create

   • Copy the Client ID and Client secret from the dialog.

Step 2 — Connect with one terminal command

Run the interactive setup command in your terminal:

npx -y github:was-member-keramat/was-gtm-mcp auth

The tool will:

1. Ask you to paste your Client ID and Client secret

2. Open your browser to Google sign-in

3. After you click Allow, save everything to `~/.was-gtm-mcp/config.jsona on your machine

That's it for setup. No more typing.

Step 3 — Add 4 lines to your AI client config

Open your AI tool config file:

• Mac: ~/Library/Application Support/Claude/claude_desktop_config.json

• Windows: %APPDATA%\Claude\claude_desktop_config.json

Paste this block (merge with any existing mcpServers):

{
  "mcpServers: {
    "WAS GTM MCP": {
      "command": "npx",
      "args": ["-y", "github:was-member-keramat/was-gtm-mcp"]
    }
  }
}

What you can ask the AI

Here are example prompts you can use once connected:

• Discovery: "List all my GTM accounts and containers."

• Tags: "List all tags in my main workspace."

• Create Tag: "Create a GA4 Event tag named 'GA4 - Purchase Event' with event name 'purchase'."

• Triggers: "Create a Custom Event trigger for event name 'generate_lead'."

• Variables: "Create a Data Layer Variable named 'dlv - user_id' for variable 'user_id'."

• Raw API: "Use ngtm_api to GET container version headers."

All tools (19 total)

CLI commands

npx -y github:was-member-keramat/was-gtm-mcp          # Start MCP server (stdio)
npx -y github:was-member-keramat/was-gtm-mcp auth     # Interactive OAuth login
npx -y github:was-member-keramat/was-gtm-mcp status   # View current config status
npx -y github:was-member-keramat/was-gtm-mcp logout   # Delete saved credentials
npx -y github:was-member-keramat/was-gtm-mcp help     # Display usage instructions

Multi-account setup

You can run multiple GTM setups by specifying environment variables per entry in your AI client config:

{
  "mcpServers: {
    "GTM Account A": {
      "command": "npx",
      "args": ["-y", "github:was-member-keramat/was-gtm-mcp"],
      "env": {
        "GTM_CLIENT_ID": "xxxx.apps.googleusercontent.com",
        "GTM_CLIENT_SECRET": "GOCSPX-xxxx",
        "GTM_REFRESH_TOKEN": "1//xxxx-token-a"
      }
    }
  }
}

Troubleshooting

Windows Script Execution Policy Error

If PowerShell blocks execution (npx.ps1 cannot be loaded), run:

Set-ExecutionPolicy -Scope CurrentUser -ExecutionPolicy RemoteSigned
``a

### Stale `npx` Cache on Windows
If Windows caches an old version after updates are pushed to GitHub, clear the cache:
```powershell
Remove-Item -Recurse -Force "$env:LOCALAPPDATA\npm-cache" -ErrorAction SilentlyContinue
npm cache clean --force

Security & Privacy

• Stores refresh token locally in ~/.was-gtm-mcp/config.json at POSIX mode 0600.

• Runs over standard stdio transport. No remote servers, tracking, or proxy middleman.

License

MIT License. Built by Abdullah Al Masum — Web Analytics Solution (WAS).