Install Optimizely Analytics MCP server

Add the Optimizely Analytics MCP server to each supported AI client with the following steps. The server URL is the same for every client, and authentication is handled by an OAuth 2.0 flow that opens in your browser the first time you use an oa_ tool.

Before you start, make sure you have the following:

In Claude Desktop, go to Settings > Developer > Edit Config. This opens claude_desktop_config.json in your default text editor. You can also open it directly at the following locations:

Add the following to your configuration file. If you already have other MCP servers configured, add the "optimizely-analytics" entry inside the existing "mcpServers" object.

Save the file and fully restart the app. The Analytics server appears in your list of connected MCP servers.

The first time you call an Analytics tool, Claude Desktop prompts you to sign in. Complete the OAuth flow with your Optimizely credentials and connect to your Opal instance.

Ask Claude, List my Optimizely Analytics apps. If you see your apps listed, the install is complete.

In Cursor, go to Settings > Tools & MCPs > Add a Custom MCP Server. You can also edit the configuration file directly at .cursor/mcp.json in your project's root directory.

Cursor prompts the OAuth flow the first time you call an Analytics tool. Sign in with your Optimizely credentials and connect to your Opal instance.

In Composer or Chat, ask, List my Optimizely Analytics apps. If you see your apps listed, the install is complete.

Use one of the following two options to add the MCP server:

Option 1: CLI command

Run the following command in your terminal:

Option 2: Configuration file

Create or edit .mcp.json in your project root (for project-scoped access) or ~/.claude/.mcp.json (for global access) and add the following:

Claude Code opens an OAuth window in your browser the first time you call an Optimizely Analytics tool. Sign in with your Optimizely credentials and connect to your Opal instance.

Ask Claude, List my Optimizely Analytics apps. If you see your apps listed, the install is complete.

In Claude.ai, go to Settings > Connectors.

Click Add custom connector and enter the MCP server URL.

Complete the OAuth flow when prompted, signing in with your Optimizely credentials and connecting to your Opal instance.

Start a new conversation and ask, List my Optimizely Analytics apps. If you see your apps listed, the install is complete.

In Visual Studio Code, open Settings and search for "MCP" under the GitHub Copilot section, or edit your settings.json directly.

Add the following to your Visual Studio Code settings.json:

Visual Studio Code opens a browser window for OAuth the first time you call an Optimizely Analytics tool from Copilot Chat. Sign in with your Optimizely credentials.

In Copilot Chat, ask, List my Optimizely Analytics apps. If you see your apps listed, the install is complete..

Go to Settings > MCP Servers, or edit .windsurf/mcp.json in your project root.

Add the following configuration to the file:

When prompted, complete the OAuth flow by connecting to your Opal instance with your Optimizely credentials.

Ask Windsurf, List my Optimizely Analytics apps. If apps are returned, the install is complete.

Any AI client that supports remote MCP can connect to Analytics MCP server using the same URL and OAuth flow:

Refer to your client's documentation for how to add a remote MCP server.

To disconnect the Analytics MCP server from your AI client, remove the entry you added during configuration.

Same server, every platform – All platforms connect to the same MCP server URL and use the same OAuth 2.0 flow. You can connect from more than one client at the same time.

Your permissions carry over – The MCP server uses your Optimizely identity. You see the same apps, events, dashboards, and experiments you can access in the Optimizely Analytics UI. You do not need additional permissions.

Network access – Make sure analytics.mcp.opal.optimizely.com is reachable from your network. If you are behind a corporate proxy or VPN and have trouble connecting, contact your network administrator.

Check that your configuration file is valid JSON. A missing comma or bracket is the most common cause. Validate the file with a JSON validator, save, and fully restart your AI client.

Some clients only start the OAuth flow when you call an Analytics tool. Try asking List my Optimizely Analytics apps to trigger it. If that does not work, check that your default browser is set correctly in your operating system.

This usually means authentication did not finish. Call any Analytics tool to retrigger the OAuth flow. If you are already authenticated but the tools still do not appear, remove the server, add it again, and re-authenticate.

Confirm your Optimizely account has access to at least one Optimizely Analytics app. The MCP server only returns data for apps you can see in the Optimizely Analytics UI. If access looks correct and the issue persists, contact Optimizely Support.