​
Prerequisites

Node.js (>= 22.11.0 LTS)

Docker

​
Project setup

Clone the example project, which includes two subgraphs (Posts and Users) as well as a Router. The example project is available here: https://github.com/wundergraph/cosmo-demo, or you can clone it with the following command: 
git clone https://github.com/wundergraph/cosmo-demo.git
cd cosmo-demo

​
Step 1: Install the Cosmo CLI tool

Install the latest version of the Cosmo CLI. 
npm i -g wgc@latest

​
Step 2: Set Up and Run the Subgraphs

The cloned project contains preconfigured subgraphs and a script to run them.

​
Ensure the script is executable

In the project’s root directory, run this to ensure the start-subgraphs.sh script is executable: 
chmod +x start-subgraphs.sh

​
Start the subgraphs

Run the following command to install dependencies and start the subgraphs in the background: 
./start-subgraphs.sh

​
Verify the subgraphs are running

Once the script finishes, check the subgraphs are running by visiting:

http://localhost:4001/graphql

(for subgraph-posts)

http://localhost:4002/graphql

(for subgraph-users)
If both load correctly, the servers are running.

​
Test Queries

You can run the following queries to confirm the subgraphs are functioning correctly.

​
Posts Subgraph (localhost:4001)

To retrieve all posts, run the following query: 
query {
  posts {
    id
    content
    authorId
  }
}
Expected Output: 
{
  "data": {
    "posts": [
      {
        "id": "1",
        "content": "Sample content",
        "authorId": "user-1"
      },
      {
        "id": "2",
        "content": "Some more sample content",
        "authorId": "user-1"
      },
      {
        "id": "3",
        "content": "Quality content",
        "authorId": "user-2"
      }
    ]
  }
}
To retrieve a single post using its id, run the following query: 
query {
  post(id: "1") {
    id
    content
    authorId
  }
}
Expected Output: 
{
  "data": {
    "post": {
      "id": "1",
      "content": "Sample content",
      "authorId": "user-1"
    }
  }
}

​
Users Subgraph (localhost:4002)

To retrieve all users, run the following query: 
query {
  users {
    id
    name
  }
}
Expected Output: 
{
  "data": {
    "users": [
      {
        "id": "user-1",
        "name": "Jane Doe"
      },
      {
        "id": "user-2",
        "name": "John Doe"
      }
    ]
  }
}
To retrieve a single user with their id, run the following query: 
query {
  user(id: "user-1") {
    id
    name
  }
}
Expected Output: 
{
  "data": {
    "user": {
      "id": "user-1",
      "name": "Jane Doe"
    }
  }
}

​
Stop the Subgraphs

You will need to keep the script running in the background for this tutorial, but if it is necessary to stop it, run: 
pkill -f "npm run dev"

​
Step 3: Generate the Router Configuration

The repository includes a graph.localhost.yaml configuration file for the wgc CLI tool to introspect and generate a router configuration.

​
Setup Considerations

  • For Docker-based subgraphs, use graph.yaml instead of graph.localhost.yaml.
  • For custom subgraphs, update the configuration file to match your setup.

​
Generate the Router Configuration

Open a new terminal and run the following command to navigate to the router directory. 
cd router
Then, generate the router configuration by running: 
wgc router compose --input graph.localhost.yaml --out config.json
Expected Output: 
Router config successfully written to config.json

​
Step 4: Run the router

While still in the router directory, start the router using Docker. Make sure Docker is running on your machine before proceeding. We recommend using Docker to prevent dependency conflicts and maintain a consistent environment. 
docker run \
  --name cosmo-router \
  --rm \
  -p 3002:3002 \
  --add-host=host.docker.internal:host-gateway \
  --platform=linux/amd64 \
  -e pull=always \
  -e DEV_MODE=true \
  -e LISTEN_ADDR=0.0.0.0:3002 \
  -e EXECUTION_CONFIG_FILE_PATH="/config/config.json" \
  -v "$(pwd)/config.json:/config/config.json" \
  ghcr.io/wundergraph/cosmo/router:latest
The router will now be running in a Docker container.

​
Step 5: Query Against Your Router

Now that the router is running, you can test it by running queries on your federated graph at http://localhost:3002. Get All Posts with Author Details 
query {
  posts {
    id
    content
    author {
      id
      name
    }
  }
}
Expected Output: 
{
  "data": {
    "posts": [
      {
        "id": "1",
        "content": "Sample content",
        "author": {
          "id": "user-1",
          "name": "Jane Doe"
        }
      },
      {
        "id": "2",
        "content": "Some more sample content",
        "author": {
          "id": "user-1",
          "name": "Jane Doe"
        }
      },
      {
        "id": "3",
        "content": "Quality content",
        "author": {
          "id": "user-2",
          "name": "John Doe"
        }
      }
    ]
  },
  "extensions": { /* omitted for brevity */ }
Get a Single Post with Author Details 
query {
  post(id: "1") {
    id
    content
    author {
      id
      name
    }
  }
}
Expected Output: 
{
  "data": {
    "post": {
      "id": "1",
      "content": "Sample content",
      "author": {
        "id": "user-1",
        "name": "Jane Doe"
      }
    }
  },
  "extensions": { /* omitted for brevity */ }

​
All done!

Congrats! You’ve deployed two subgraphs, composed them into a federated Graph, and deployed a Router. Next, consider deploying to the cloud to enable features like metrics, tracing, and monitoring in Cosmo Studio.

​
Next Steps: Cosmo Cloud and deployments

Cosmo Cloud Onboarding