The Rover init Command

Rover enables you to create a graph or MCP server with a single interactive command.

Initializing a graph

rover init

Running rover init starts a short wizard that helps you create a new graph.

rover init

The wizard walks you through a set of questions to help you choose the best option for your use case. When you're done, you'll have:

• A new graph in GraphOS

• A set of local configuration files and boilerplate code that represent the graph

• Credentials to interact with GraphOS

  • These credentials are necessary for actions like developing your graph locally or publishing graph updates.

Options

You can also pass options directly to rover init instead of answering prompts in the wizard. The wizard still asks for any information not provided via an option and will always ask for confirmation before creating any files locally.

rover init \
  --project-type <PROJECT_TYPE> \
  --org-id <ORGANIZATION_ID> \
  --project-use-case <PROJECT_USE_CASE> \
  --project-name <PROJECT_NAME> \
  --graph-id <GRAPH_ID>

For rover init --mcp options, see the Initializing an MCP Server section.

Available options

Choosing your use case

Once you run rover init, the wizard prompts you to select your use case.

• Start a graph with one or more REST APIs

  • Select this option to integrate REST APIs into your graph using Apollo Connectors.

• Start a graph with recommended libraries

  • Select this option if you want to integrate data that's not accessible via a REST API. This option helps you set up a graph using the Apollo Server library.

Created credentials

After you've chosen your use case, the wizard prompts you for a project name. It generates the following credentials based on the name you enter:

• Graph ID- A unique identifier for your graph. It represents your graph across all of Apollo.

• Graph ref- A reference for a specific variant of your graph. It's formatted graph-id@variant. To start, GraphOS automatically creates a variant titled current for you.

• Graph API key- Once it's generated, store it securely—you won't be able to access it later.

  • See the configuration docs to learn how to set your API key as an environment variable

Next steps

Once you've completed the wizard, it provides a rover dev command with prepopulated credentials. Run this command to start developing your graph.

rover dev --supergraph-config supergraph.yaml

rover dev --supergraph-config supergraph.yaml

rover dev --supergraph-config supergraph.yaml

1npm ci
2npm start

Additional resources

• After running rover init, open the generated GETTING_STARTED.md file for next steps.

• To go further, check out the Getting started guide to learn what else you can do with Apollo Connectors.

• If you learn best with videos and exercises, this interactive course teaches you how to integrate a demo REST API into a graph using Apollo Connectors.

Initializing an MCP server

rover init --mcp

Rover's init --mcp command provides a helpful starting point for working with the Apollo MCP Server.

rover init --mcp

The init CLI wizard walks you through the process of configuring and creating an MCP server. When you're done, you'll have starter files for an MCP server and a graph to run locally.

The command creates the following files:

• MCP config YAML for local and staging scenarios

• Starter MCP tools, based on your GraphQL API operations

• .env files for running the GraphQL API and MCP Server

• A set of local configuration files and boilerplate code that represent the graph

• mcp.Dockerfile (optional Docker container)

Options

You can also pass options directly to rover init --mcp instead of answering prompts in the wizard. The wizard still asks for any information not provided via an option and always asks for confirmation before creating any files locally.

rover init --mcp \
  --project-type <PROJECT_TYPE> \
  --org-id <ORGANIZATION_ID> \
  --project-name <PROJECT_NAME> \
  --graph-id <GRAPH_ID>

Available options

Next steps

Run your MCP server

When the rover init wizard is complete, you can use rover dev to start everything together. See Running the Apollo MCP Server.

Connect your MCP server to Claude Desktop

1# Copy and provide the configuration to Claude Desktop
2# macOS:
3cp claude_desktop_config.json ~/Library/Application\ Support/Claude/claude_desktop_config.json
4
5# Windows:
6copy claude_desktop_config.json "%APPDATA%\Claude\claude_desktop_config.json"
7
8# Linux:
9cp claude_desktop_config.json ~/.config/Claude/claude_desktop_config.json
10
11# Restart Claude Desktop

For instructions on adding your MCP server to other AI clients, see the MCP Quickstart guide.