Skip to main content

Introduction to GA MCP Server

The GameAnalytics Model Context Protocol (GA MCP) Server is an open-source bridge that connects your GameAnalytics data with AI assistants supporting the MCP standard, such as Claude Desktop, Cursor, and Antigravity. This integration allows you to query your game metrics using natural language and receive data-driven answers without leaving your preferred workflow. We designed this tool to streamline data analysis by reducing the need to navigate complex dashboards or write manual SQL queries.

Key Benefits:

  • Instant Insights: Ask questions in plain language to get immediate answers about KPIs like retention and ARPU.
  • Workflow Integration: Check events and game statistics directly within your IDE (VS Code or Cursor) or AI assistant.
  • Accelerated Discovery: Iterate on data questions and explore new dimensions before building permanent dashboards.

Concepts and Context​

Before you begin, it is helpful to understand the core technology powering this integration.

Model Context Protocol (MCP)​

MCP is a standard that enables AI models to interact with external tools and APIs. In this context, the GA MCP Server acts as a "universal translator," allowing your AI assistant to "speak" directly to the GA Metrics API. Instead of manually filtering dates and dimensions, the server translates your natural language request into precise API calls, retrieves the JSON data, and summarizes the findings for you.

Use Cases by Role​

We support distinct workflows for different team members:

RoleBenefitExample Query
Product ManagerQuick clarity on KPIs"What was the D7 retention for our iOS build last week?"
DeveloperData validation in IDE"Verify if the new custom event 'Level_5_Complete' is appearing in the metrics."
AnalystComplex query drafting"Compare ARPU between users from the USA versus users from Germany."

Installation and Setup​

The GA MCP Server is fully open-source. You can run it locally or integrate it into your studio's workflow.

Prerequisites

To use the GA MCP Server, you must have:

  • A GameAnalytics account with active game data.
  • Access to your GameAnalytics Metrics API Key (found in PipelineIQ → Metrics API).
  • An MCP-compatible AI client (e.g., Claude Desktop, Cursor).

Retrieve GA Credentials​

You need to authenticate the server to access your game data.

  1. Log in to the GameAnalytics dashboard.
  2. Navigate to Pipeline IQ for the specific Organization you wish to query.
  3. Generate your Metrics API. You will need this to configure the environment.

Configure the Server​

Detailed installation instructions and the latest source code are available in the GA MCP GitHub Repository.

Generally, you will need to add the GA MCP Server configuration to your AI client's config file (e.g., claude_desktop_config.json).

{
  "mcpServers": {
    "GameAnalytics": {
      "command": "python3",
      "args": ["-m", "ga_mcp.server"],
      "env": {
        "GAMEANALYTICS_API_KEY": "your_api_key_here"
      }
    }
  }
}
note

The exact package name and arguments may vary. Please refer to the GA MCP GitHub Repository for the most up-to-date info.

Usage Examples​

Once the server is running, you can open your AI client and start chatting with your data.

Example: Analyzing Revenue Decline​

You can ask complex, multi-part questions to diagnose issues.

User Query​

Fetch the daily revenue for 'Galaxy Quest' for the last 30 days. Is there a specific country driving the decline?

AI Response (Summary)​

I have analyzed the daily revenue for 'Galaxy Quest' over the last 30 days (2025-11-08 to 2025-12-08). Summary: Yes, Germany (DE) appears to be the primary driver of the revenue decline.

Findings:

  • Top Markets: Revenue is driven by Germany (~81%) and Denmark (~19%).
  • Germany Trend: Shows a noticeable downward trend from ~$2,407 to ~$1,600.
  • Denmark Trend: Remains relatively stable around ~$548.

Since Germany accounts for the vast majority of revenue, its decline has a direct impact on overall metrics.

Best Practices​

Data Privacy

When using LLMs, ensure you understand how your AI provider handles data. While the GA MCP Server fetches data securely from our API, the conversation history is processed by the AI client (e.g., Anthropic, OpenAI).

Be Specific

For the best results, include specific date ranges and metric names in your prompt. For example, instead of asking "How is the game doing?", ask "What is the daily active user count for the last 7 days?"