Google Tag Manager MCP

Google Tag Manager MCP

MCP server for the Google Tag Manager API v2, enabling natural-language management of containers, workspaces, tags, triggers, variables, and publishing. It handles OAuth authentication and built-in rate limiting for the strict GTM API quota.

Category
Visit Server

README

Google Tag Manager MCP

npm CI Glama License: MIT

MCP server for the Google Tag Manager API v2: manage containers, workspaces, tags, triggers, variables and publishing from Claude, Cursor, Codex and other AI clients in natural language.

Ask the assistant to audit a container, wire up a new GA4 tag with its trigger, enable built-in variables, compile a version and push it live — the full GTM workflow without clicking through the web UI.

Quick start

  1. Get OAuth credentials for a Google Cloud project with the Tag Manager API enabled.

  2. Add the server — for example in Claude Code (other clients):

    claude mcp add google-tagmanager \
      -e GOOGLE_TAGMANAGER_CLIENT_ID=your_client_id \
      -e GOOGLE_TAGMANAGER_CLIENT_SECRET=your_client_secret \
      -e GOOGLE_TAGMANAGER_REFRESH_TOKEN=your_refresh_token \
      -- npx -y mcp-google-tagmanager@latest
    
  3. Ask the assistant: "List my GTM containers and show which tags fire on page view."

Tools

Tool Description
list_accounts List all GTM accounts the user can access
get_account Get one account
list_containers List containers of an account (with GTM-XXXXXX public ids)
get_container Get one container
create_container Create a container (web, server, ...)
list_workspaces List workspaces of a container
get_workspace Get one workspace
create_workspace Create a workspace (draft)
list_tags List tags of a workspace
list_triggers List triggers of a workspace
list_variables List user-defined variables of a workspace
get_resource Get any resource by its API path (tag, trigger, variable, version, ...)
create_entity Create a tag, trigger or variable
update_entity Update a tag/trigger/variable (PUT full replace, fingerprint-guarded)
delete_entity Delete a tag/trigger/variable
manage_built_in_variables List / enable / disable built-in variables (full enum from the discovery doc)
create_version Compile a workspace into a container version (⚠️ deletes the workspace)
publish_version Publish a version, get one version, or fetch the live version
raw_request Escape hatch: call any Tag Manager API v2 path directly

Built-in rate limiting

The Tag Manager API quota is unusually strict: 0.25 QPS per project (25 requests per 100-second sliding window) and 10,000 requests per day. The server handles this for you:

  • every API request goes through a serialized queue with a minimum spacing of 4.2 s between requests (tunable via GOOGLE_TAGMANAGER_MIN_INTERVAL_MS);
  • 429 and quota-403 (rateLimitExceeded / userRateLimitExceeded / quotaExceeded) responses are retried with exponential backoff honoring Retry-After;
  • 5xx and network errors are retried for reads only — a write that may have committed is never replayed.

Big fan-out requests ("list everything in every container") will therefore be slow by design — that is the quota, not the server.

Example prompts

  • "Which tags in container GTM-ABC123 fire on the page-view trigger?"
  • "Create a Custom HTML tag in the default workspace that logs to the console, firing on all pages."
  • "Enable the clickText and clickClasses built-in variables in my workspace."
  • "Compile my workspace into a version named 'March release' and publish it."

Installation

<details open> <summary><b>Claude Code</b></summary>

claude mcp add google-tagmanager \
  -e GOOGLE_TAGMANAGER_CLIENT_ID=your_client_id \
  -e GOOGLE_TAGMANAGER_CLIENT_SECRET=your_client_secret \
  -e GOOGLE_TAGMANAGER_REFRESH_TOKEN=your_refresh_token \
  -- npx -y mcp-google-tagmanager@latest

</details>

<details> <summary><b>Claude Desktop</b></summary>

claude_desktop_config.json — macOS ~/Library/Application Support/Claude/, Windows %APPDATA%\Claude\

{
  "mcpServers": {
    "google-tagmanager": {
      "command": "npx",
      "args": ["-y", "mcp-google-tagmanager@latest"],
      "env": {
        "GOOGLE_TAGMANAGER_CLIENT_ID": "your_client_id",
        "GOOGLE_TAGMANAGER_CLIENT_SECRET": "your_client_secret",
        "GOOGLE_TAGMANAGER_REFRESH_TOKEN": "your_refresh_token"
      }
    }
  }
}

</details>

<details> <summary><b>Cursor</b></summary>

~/.cursor/mcp.json (or .cursor/mcp.json in the project)

{
  "mcpServers": {
    "google-tagmanager": {
      "command": "npx",
      "args": ["-y", "mcp-google-tagmanager@latest"],
      "env": {
        "GOOGLE_TAGMANAGER_CLIENT_ID": "your_client_id",
        "GOOGLE_TAGMANAGER_CLIENT_SECRET": "your_client_secret",
        "GOOGLE_TAGMANAGER_REFRESH_TOKEN": "your_refresh_token"
      }
    }
  }
}

</details>

For a quick one-off session you can skip the trio and pass a short-lived token directly: GOOGLE_TAGMANAGER_ACCESS_TOKEN=ya29.... (Google access tokens expire after about an hour and are not refreshed automatically).

Getting credentials

The Tag Manager API only supports OAuth 2.0 — there are no API keys for user data. One-time setup:

  1. Create/pick a Google Cloud project at console.cloud.google.com and enable the Tag Manager API (direct link). Without a registered project the API grants zero quota — this step is mandatory.
  2. Configure the OAuth consent screen (APIs & Services → OAuth consent screen). For personal use, External + your account as a test user is enough.
  3. Create an OAuth client (APIs & Services → Credentials → Create credentials → OAuth client ID → Desktop app or Web application). Save the client id and client secret.
  4. Mint a refresh token. The easiest path is the OAuth 2.0 Playground:
    • gear icon → check Use your own OAuth credentials → paste the client id/secret (for a Web client also add https://developers.google.com/oauthplayground to its authorized redirect URIs);

    • in Step 1 authorize these scopes (space-separated):

      https://www.googleapis.com/auth/tagmanager.readonly https://www.googleapis.com/auth/tagmanager.edit.containers https://www.googleapis.com/auth/tagmanager.edit.containerversions https://www.googleapis.com/auth/tagmanager.publish
      
    • in Step 2 click Exchange authorization code for tokens and copy the refresh token.

  5. Put the three values into the environment variables above. The server exchanges the refresh token for access tokens automatically and caches them until just before expiry.

⚠️ The credentials are stored as plain text in your client's MCP config. Scope the OAuth consent to the four Tag Manager scopes above and nothing else.

Note the scope split: reading needs readonly, editing needs edit.containers, create_version needs edit.containerversions, and publishing needs publish. Authorize all four at once or re-consent mid-flow.

Configuration

Variable Required Default Description
GOOGLE_TAGMANAGER_CLIENT_ID yes* OAuth client id
GOOGLE_TAGMANAGER_CLIENT_SECRET yes* OAuth client secret
GOOGLE_TAGMANAGER_REFRESH_TOKEN yes* OAuth refresh token
GOOGLE_TAGMANAGER_ACCESS_TOKEN no Ready-made access token; replaces the trio for quick sessions
GOOGLE_TAGMANAGER_API_BASE no https://tagmanager.googleapis.com API root override
GOOGLE_TAGMANAGER_TIMEOUT_MS no 60000 Per-request timeout
GOOGLE_TAGMANAGER_MAX_RETRIES no 3 Retries on transient errors
GOOGLE_TAGMANAGER_MIN_INTERVAL_MS no 4200 Minimum spacing between API requests (0.25 QPS quota)

* the trio is required unless GOOGLE_TAGMANAGER_ACCESS_TOKEN is set.

Good to know

  • create_version deletes the source workspace. The response's newWorkspacePath points to the automatically created replacement — the server surfaces it and the tool description warns the model, but keep it in mind when scripting.
  • compilerError: true can arrive with HTTP 200 on create_version and publish. The server converts it into a tool error so it is never mistaken for success.
  • Updates are full replacements (PUT, not PATCH): update_entity expects the complete resource. Fetch with get_resource, edit, send back, and pass the fingerprint for optimistic-concurrency safety.
  • All ids are strings, and every resource carries its own path field — echo it back rather than assembling paths by hand.

Requirements

  • Node.js >= 20
  • A Google account with access to at least one GTM container

Documentation

Support

Questions and issues → GitHub Issues or Telegram @gistrec.

License

MIT

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured
E2B

E2B

Using MCP to run code via e2b.

Official
Featured