Delmain GA4 MCP
Enables querying Google Analytics 4 properties using natural language through MCP clients. Supports customizable reports with any dimensions and metrics, listing properties, and real-time data.
README
:Delmain GA4 MCP
A small MCP server that lets any MCP client (Claude Desktop, Claude Code, Cursor, ...) query Google Analytics 4.
It is intentionally neutral: tools take a property ID plus free-form dimensions and metrics, so it works for PPC, SEO, content, leadership, anyone. Each person authenticates with their own Google account (per-user OAuth), so everyone only sees the GA4 properties they already have access to.
Tools
| Tool | What it does |
|---|---|
list_properties |
List every GA4 account + property your account can read |
get_property |
Metadata for one property (name, timezone, currency) |
run_report |
Report with any dimensions/metrics over a date range |
run_realtime_report |
Active users / events in the last ~30 minutes |
run_report is the workhorse. Example dimensions: sessionSource,
sessionMedium, landingPage, date, country, deviceCategory,
eventName. Example metrics: sessions, totalUsers, newUsers,
screenPageViews, bounceRate, averageSessionDuration, conversions,
eventCount.
Installing with Claude Code
The fastest path. Open Claude Code (or the Code tab in the Claude Desktop App) and paste:
Install the MCP server from https://github.com/aline-delmain/delmain-ga4-mcp —
pip installit and add it to my MCP config as "delmain-ga4".
Claude will clone the repo, install the package, and wire up your MCP client config. Then you finish the two steps Claude can't do for you (by design, since every person authenticates as themselves):
- Paste the OAuth
client_id/client_secret(from the team vault) into~/.delmain-ga4-mcp/.env. - Run
delmain-ga4-mcp-setupand authorize in the browser with your own Google account.
Restart your MCP client and the GA4 tools show up.
Note: a plain chat at claude.ai cannot install software, only Claude Code or the Desktop Code tab can. And the admin setup below must be done once first, otherwise authorization fails with
org_internal/SERVICE_DISABLED.
One-time setup by an admin (do this once for the whole team)
- OAuth client — in the Google Cloud project that owns the ":Delmain GA4" app, create (or reuse) an OAuth client of type Desktop app. Note its client ID and client secret.
- Consent screen — so teammates don't each need to be added as test users:
- If everyone authorizes with an @delmain.co account, set the consent screen to Internal (org-only) and you're done.
- If people will use other Google accounts, set it to External and Publish the app (status: In production). While in Testing, only listed test users can authorize.
- Enable APIs in that project:
- Google Analytics Data API (
analyticsdata.googleapis.com) - Google Analytics Admin API (
analyticsadmin.googleapis.com)
- Google Analytics Data API (
- Put the client ID + secret in the team vault (1Password / Bitwarden). They are shared by everyone; only each person's refresh token is personal.
Per-user install
Requires Python 3.10+.
# 1. Install (from the repo, or once published, from GitHub)
pip install git+https://github.com/<org>/delmain-ga4-mcp.git
# or, working in a clone:
pip install -e .
# 2. Configure credentials
mkdir -p ~/.delmain-ga4-mcp
cp .env.example ~/.delmain-ga4-mcp/.env
# edit it and paste DELMAIN_GA4_CLIENT_ID + DELMAIN_GA4_CLIENT_SECRET
# (from the team vault). Leave the refresh token blank.
# 3. Generate YOUR personal refresh token (opens the browser)
delmain-ga4-mcp-setup
# sign in with the Google account that has your GA4 access, then authorize.
That writes your refresh token to ~/.delmain-ga4-mcp/.env. Done.
Connect it to your MCP client
The server runs over stdio via the delmain-ga4-mcp command.
Claude Desktop / Claude Code
Add to your MCP config (claude_desktop_config.json, or .mcp.json /
~/.claude.json for Claude Code):
{
"mcpServers": {
"delmain-ga4": {
"command": "delmain-ga4-mcp"
}
}
}
Cursor
~/.cursor/mcp.json:
{
"mcpServers": {
"delmain-ga4": { "command": "delmain-ga4-mcp" }
}
}
Restart the client. You should see the GA4 tools available.
Example prompts
- "List my GA4 properties."
- "For property 463688888, sessions and conversions by sessionSource/sessionMedium, last 30 days."
- "Top 10 landing pages by sessions for property 463688888 this month."
- "How many active users are on property 463688888 right now?"
Notes
- Read-only. The server only requests
analytics.readonly; it cannot change anything in GA4. - No secrets in git.
.env,*.jsoncredentials are gitignored. Share the OAuth client via the vault, never by committing it. - Access scope. You can only query properties your own Google account can
see.
list_propertiesshows exactly that set.
Recommended Servers
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.
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.
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.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
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.
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.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
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.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.