GrowthBook MCP Server: A/B Testing Without Context Switching

Stop jumping between your code editor and the GrowthBook dashboard. This official MCP server brings your entire experimentation workflow directly into your AI coding assistant.

The Context Switching Problem

You're deep in a feature branch when you need to create a flag, check experiment results, or update targeting rules. Instead of staying in flow, you tab over to GrowthBook's web interface, hunt through projects, make your changes, then try to remember where you left off in your code.

The GrowthBook MCP server eliminates this friction by putting your entire A/B testing toolkit right in your editor's AI assistant.

What You Get

Complete Feature Flag Management

• Create flags with targeting conditions without leaving your editor
• Generate TypeScript types for your flags automatically
• Check which safe rollouts need attention
• Update flag values and targeting rules on the fly

Experiment Insights at Your Fingertips

• Pull experiment results while reviewing related code
• Get user attributes for targeting decisions
• Check experiment status across environments
• Review historical experiment data for context

Workflow Integration

• Create SDK connections for new projects instantly
• Search GrowthBook docs when you need implementation help
• Manage environments and project settings
• All through natural language commands to your AI assistant

Real-World Scenarios

During Code Reviews: "Show me the current state of the checkout-flow experiment while I review this PR"

Feature Development: "Create a feature flag for the new recommendation engine with 5% rollout to production users"

Debugging: "What experiments are running on the user profile page right now?"

Planning: "Generate TypeScript types for all our current feature flags so I can update the frontend"

Zero-Friction Setup

The beauty of MCP: no installation, no build process, no configuration files to manage.

Add this to your Cursor, VS Code, or Claude Desktop config:

{
  "mcpServers": {
    "growthbook": {
      "command": "npx",
      "args": ["-y", "@growthbook/mcp"],
      "env": {
        "GB_API_KEY": "your_api_key",
        "GB_USER": "your_name"
      }
    }
  }
}

That's it. Your AI assistant now has direct access to your GrowthBook instance.

Why This Changes Everything

Instead of context switching between tools, you can:

• Create feature flags while writing the code that uses them
• Check experiment results while debugging related issues
• Update targeting rules based on code review feedback
• Generate flag types that sync with your actual flags

This isn't just convenience—it's about maintaining flow state while building experimentation into your development process. When A/B testing becomes as easy as asking your AI assistant, you'll find yourself experimenting more thoughtfully and consistently.

The server comes from GrowthBook themselves, so you get the full API surface area with proper authentication and all the features you expect from their platform.

Your experimentation workflow just got a lot smoother.