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
访问服务器

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

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选