google-analytics-mcp-server

A stdio MCP server for Google Analytics with full coverage of the Admin and Data APIs. Empower your AI agent to explore data, run reports, manage configuration and control access in Google Analytics. 🚀

✨ Highlights

• Full API surface, compact tool set. 28 tools cover the pinned Admin and Data alpha/beta APIs, with 26 enabled by default.
• Built for context windows. Returns Markdown tables by default, with json and json_full available.
• Controlled writes. Writable leaf patches, confirmation for sensitive operations and a strict read-only mode.
• Runs as you. Your own OAuth client through Application Default Credentials (ADC).

Compared with Google's official MCP, this server adds configuration and access management, pivot/batch reports and async jobs.

🧰 Tools

Chat and raw API access are optional. For feature settings and strict read-only mode, see Appendix A: Environment Variables.

Available operations depend on your Google Analytics permissions and property eligibility. See Coverage and known limitations for current limitations.

🔑 Setup

You need a Google Cloud project and the gcloud CLI.

1. Enable the APIs

Enable both APIs on the project that will carry the quota:

gcloud services enable analyticsadmin.googleapis.com analyticsdata.googleapis.com --project=YOUR_PROJECT

2. Create an OAuth client

Create a Desktop app OAuth client in your Google Cloud project and download its JSON file. See Manage OAuth Clients.

• If your OAuth app's user type is set to External, check its publishing status before logging in.
• With the Analytics scopes used here, refresh tokens issued for an External app in Testing expire after seven days. For ongoing use, switch to In production. See Google's OAuth guidance.

3. Authorize access

Log in with the scopes needed for your work. Remove the scopes you don't need. A limited grant produces an error when a tool requires an additional scope:

gcloud auth application-default login \
  --client-id-file=YOUR_DESKTOP_CLIENT.json \
  --scopes=https://www.googleapis.com/auth/analytics.readonly,\
https://www.googleapis.com/auth/analytics.edit,\
https://www.googleapis.com/auth/analytics.manage.users,\
https://www.googleapis.com/auth/analytics.chatbot.read,\
https://www.googleapis.com/auth/cloud-platform

For limited read credentials, use:

• https://www.googleapis.com/auth/analytics.readonly
• https://www.googleapis.com/auth/analytics.manage.users.readonly

Change-history reads still require analytics.edit.

Already using google-tag-manager-mcp? If both servers read the same ADC file, include both sets in one login:

gcloud auth application-default login \
  --client-id-file=YOUR_DESKTOP_CLIENT.json \
  --scopes=https://www.googleapis.com/auth/analytics.readonly,\
https://www.googleapis.com/auth/analytics.edit,\
https://www.googleapis.com/auth/analytics.manage.users,\
https://www.googleapis.com/auth/analytics.chatbot.read,\
https://www.googleapis.com/auth/tagmanager.readonly,\
https://www.googleapis.com/auth/tagmanager.edit.containers,\
https://www.googleapis.com/auth/tagmanager.delete.containers,\
https://www.googleapis.com/auth/tagmanager.edit.containerversions,\
https://www.googleapis.com/auth/tagmanager.publish,\
https://www.googleapis.com/auth/tagmanager.manage.users,\
https://www.googleapis.com/auth/tagmanager.manage.accounts,\
https://www.googleapis.com/auth/cloud-platform

Enable chat with GA_MCP_ENABLE_CHAT=1 after authorization. See Appendix A: Environment Variables for the full list.

To use different identities or grants, set GOOGLE_APPLICATION_CREDENTIALS for each MCP process to its ADC credentials file.

🔌 Connect an MCP client

Install from PyPI with uv (recommended) or pipx:

uv tool install --python 3.14 google-analytics-mcp-server

The client examples below pin an explicit quota project. The identity needs permission to consume services on it.

The quota project is selected in this order:

1. GOOGLE_CLOUD_QUOTA_PROJECT overrides credential configuration.
2. Otherwise, a configured ADC quota_project_id is used.
3. Without an explicit quota project, attribution depends on the credentials and API, commonly the OAuth client's project.

Claude

Claude Code:

claude mcp add --scope user google-analytics-mcp-server \
  -e GOOGLE_CLOUD_QUOTA_PROJECT=YOUR_PROJECT \
  -- google-analytics-mcp-server

Claude Desktop:

Open Settings > Developer > Edit Config and add:

{
  "mcpServers": {
    "google-analytics-mcp-server": {
      "command": "google-analytics-mcp-server",
      "env": { "GOOGLE_CLOUD_QUOTA_PROJECT": "YOUR_PROJECT" }
    }
  }
}

If Claude Desktop cannot find the command, use its absolute path instead.

ChatGPT / Codex

In the desktop app, go to Settings > MCP servers > Add server and choose STDIO. Use google-analytics-mcp-server as the command.

Or add the server with the Codex CLI:

codex mcp add google-analytics-mcp-server \
  --env GOOGLE_CLOUD_QUOTA_PROJECT=YOUR_PROJECT \
  -- google-analytics-mcp-server

Try it out

After connecting the server, try asking your AI agent:

• “List my Google Analytics accounts and properties.”
• “Compare purchase revenue by country over the last 28 days.”
• “Create an event-scoped custom dimension for membership_level.”

🛡️ Safety model

• Your approval. Sensitive actions require you to approve what will change and where.
• Read-only mode. Limits your agent to viewing data and running reports; changes, new background jobs and chat are disabled.
• Access management. Specify every role a user should have when changing their access. Removing all roles deletes that direct access assignment.
• Sensitive output. Detailed results may include Measurement Protocol API secrets. Summary lists hide those values.

🧪 Coverage and known limitations

This server is in alpha. It supports the bundled Admin/Data API versions. Features depend on your Analytics permissions and property eligibility.

• Chat: Experimental and disabled by default. Successful sessions remain unverified.
• Account provisioning and Analytics 360: These workflows remain unverified; see known limitations.
• Outside scope: Universal Analytics, Measurement Protocol event collection and BigQuery export queries.

Development

For development from a source checkout:

uv sync --locked
uv run pytest

Tests run offline by default. See the test guide for packaging checks and opt-in live tests.

License

MIT, see LICENSE.

Appendix A: Environment Variables

Flags accept 1, true or yes. Restart after environment changes.