Listed from the MCP registry; nobody on Odel publishes it.
MCP server for the Google Tag Manager API v2: containers, workspaces, tags, triggers, publishing.
Runs on your machine from its package; the steps are below.

At most: moves money, runs code or commands, deletes data, or sends messages for you.
Odel charges nothing per call. The publisher may charge for its own service.
Runs locally over stdio
This server isn't hosted — your MCP client launches it from a package registry. Use one of the commands below, or drop the config into your client (e.g. Claude Desktop).
Run
npx -y mcp-google-tagmanager
MCP client config
{
"mcpServers": {
"google-tag-manager-mcp": {
"command": "npx",
"args": [
"-y",
"mcp-google-tagmanager"
]
}
}
}English | Русский
A1 Google Tag Manager MCP lets an AI app inspect and manage Google Tag Manager containers in plain language. See what fires on a page, work with tags, triggers and variables in a draft workspace, then deliberately compile and publish a version when you are ready.
It connects to the Google Tag Manager API v2 through your Google account. The difference from asking an AI to guess a GTM setup is that it works with the actual container, workspace and version you choose.
127.0.0.1 with PKCE and keeps the tokens itself — no config files, no restart.Start with a read-only question:
Which tags in my containers fire on the page-view trigger?
Connect the server · Explore use cases · Open technical documentation
You: List my GTM containers and show which tags fire on page view.
Assistant: Lists the containers, their workspaces, relevant triggers and the tags attached to them. Nothing changes.
You: In the Default Workspace of
GTM-ABC123, prepare a GA4 configuration tag for measurement IDG-XXXXXXXon all pages.Assistant: Shows the workspace, proposed tag and trigger configuration, then asks for confirmation before creating the draft.
You: Confirm the draft.
Assistant: Creates the tag in the workspace. It does not publish the container; compiling and publishing a version remains a separate step.
You need Node.js 20+ and a Google account. Credentials are not required at install time — the server connects from the conversation.
In the app:
npx -y mcp-google-tagmanager@latest and the three environment variables below.| Variable | Value |
|---|---|
GOOGLE_TAGMANAGER_CLIENT_ID | Your Google OAuth client ID |
GOOGLE_TAGMANAGER_CLIENT_SECRET | Your Google OAuth client secret |
GOOGLE_TAGMANAGER_REFRESH_TOKEN | Your Google OAuth refresh token |
From the command line:
codex mcp add google-tagmanager \
-- npx -y mcp-google-tagmanager@latest
codex mcp list
claude mcp add \
--transport stdio \
--scope user \
google-tagmanager \
-- npx -y mcp-google-tagmanager@latest
claude mcp list
The current official path is Settings → Extensions. For a custom desktop extension, open Advanced settings → Extension Developer → Install Extension…, select a .mcpb file and follow the prompts.
This repository currently publishes an npm stdio package and does not contain a .mcpb bundle. For Claude Desktop builds that still support local configuration, use the following JSON stdio configuration as a fallback:
{
"mcpServers": {
"google-tagmanager": {
"command": "npx",
"args": ["-y", "mcp-google-tagmanager@latest"]
}
}
}
In those builds, save it to ~/Library/Application Support/Claude/claude_desktop_config.json on macOS or %APPDATA%\Claude\claude_desktop_config.json on Windows.
Add a user-level server to ~/.cursor/mcp.json on macOS/Linux or %USERPROFILE%\.cursor\mcp.json on Windows:
{
"mcpServers": {
"google-tagmanager": {
"type": "stdio",
"command": "npx",
"args": ["-y", "mcp-google-tagmanager@latest"]
}
}
}
Run MCP: Open User Configuration from the Command Palette and add:
{
"servers": {
"google-tagmanager": {
"type": "stdio",
"command": "npx",
"args": ["-y", "mcp-google-tagmanager@latest"]
}
}
}
Check it with MCP: List Servers.
April release.42 after I confirm the version and its changes.GTM has a clear release path:
This server can inspect each step. It does not treat a draft as a release: version creation and publishing are separate operations.
| Operation | What happens | Confirmation boundary |
|---|---|---|
| List accounts, containers, workspaces, tags, triggers, variables and versions | Reads GTM configuration | No change |
| Create a container or workspace | Adds a new GTM object | Changes GTM |
| Create a tag, trigger or variable | Adds a draft object to a workspace | Changes a draft workspace |
| Enable or disable built-in variables | Changes the workspace configuration | Changes a draft workspace |
| Update a tag, trigger or variable | Replaces the complete resource, protected by its fingerprint | Potentially destructive |
| Delete a tag, trigger or variable | Removes the selected object | Destructive |
| Compile a workspace | Creates a version and deletes the source workspace | Destructive |
| Publish a version | Makes a selected version live | Destructive |
| Raw API request | Can call API methods without a dedicated tool | Potentially destructive |
The AI client decides how it asks for confirmation. The server marks read-only, write and destructive operations so the client can distinguish inspection from a real change.
Google Tag Manager requires OAuth 2.0; an API key is not enough. There are two ways in, and the first one needs no configuration files.
Say "connect Google Tag Manager" and the assistant runs the flow with you:
setup_instructions prints the checklist: create or select a Google Cloud project, enable Tag Manager API, configure the consent screen and create a Desktop app OAuth client.set_client stores it owner-only. The secret never goes through the conversation.start_login returns a Google consent link. Open it on this machine and approve; the code comes back to a one-shot listener on 127.0.0.1 (PKCE), never through the chat.finish_login exchanges the code and saves the tokens to ~/.config/mcp-google-tagmanager/credentials.json (mode 0600) and verifies them with a real Tag Manager API call — so an API that is still switched off is caught right there.The tokens are re-read on every call, so the connection works immediately — no restart of the AI app. auth_status shows what is connected, logout revokes and deletes it.
Create or select a Google Cloud project and enable the Tag Manager API. A project without that API enabled receives no quota.
Configure the OAuth consent screen and create an OAuth client. A Desktop app client is suitable for local use.
Authorize your Google account and obtain a refresh token. The OAuth 2.0 Playground can do this if you enable Use your own OAuth credentials.
Request these scopes together:
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
The scopes are separate: reading, editing, compiling versions and publishing each need their corresponding permission. Treat the client secret and refresh token as passwords.
Every variable is optional — with none of them the server connects from the chat.
| Variable | Required | Description |
|---|---|---|
GOOGLE_TAGMANAGER_CLIENT_ID | No* | OAuth client ID. |
GOOGLE_TAGMANAGER_CLIENT_SECRET | No* | OAuth client secret. |
GOOGLE_TAGMANAGER_REFRESH_TOKEN | No* | OAuth refresh token. |
GOOGLE_TAGMANAGER_ACCESS_TOKEN | No* | Short-lived alternative to the OAuth trio. |
GOOGLE_TAGMANAGER_OAUTH_PORT | No | Fixed loopback port for the in-chat login; useful over SSH port forwarding. |
GOOGLE_TAGMANAGER_API_BASE | No | Tag Manager API base URL override. |
GOOGLE_TAGMANAGER_TIMEOUT_MS | No | Per-request timeout; default 60000 ms. |
GOOGLE_TAGMANAGER_MAX_RETRIES | No | Maximum retries on temporary failures; default 3. |
GOOGLE_TAGMANAGER_MIN_INTERVAL_MS | No | Minimum request spacing; default 4200 ms. |
* Provide either the OAuth trio or an access token. Access tokens expire in about an hour and are not refreshed automatically.
The server runs locally and sends GTM API requests and OAuth refresh requests to Google. Its anonymous telemetry contains a random installation ID, package version, AI client and Node.js/operating-system versions, and tool names. It does not send OAuth tokens, GTM data, tool arguments or prompts. For each tool call it also sends the outcome and, on failure, an excerpt of the error text with e-mails, addresses, tokens, numbers of 4+ digits and quoted text replaced by #; a value from your request can appear in it if the error text contains it.
Disable telemetry for A1 MCP servers with:
ASKADS_TELEMETRY=0
429 and Google quota 403 responses use exponential backoff and Retry-After. Reads retry after network and 5xx failures; writes are not replayed after an uncertain failure.create_version, save anything you need from the workspace and inspect the returned replacement workspace path.Found a bug or need a scenario? Create an issue or write in Telegram.
You made it to the end!