Configuring Apollo projects

Apollo projects are configured using an apollo.config.js file at the root of your project. Many Apollo tools leverage your Apollo config, reducing the net amount of configuration you need to do in your project in the end.

If you're using one of our workflow tools like the Apollo CLI or the Apollo VS Code extension, you'll need to have an apollo.config.js project to get the features those tools bring.

There are two types of projects, client and service. When a client key is found in the config, a project is treated as a client project, and only client:* commands may be run against it. The same is true for service projects. The apollo.config.js file is meant to describe configuration for a single project only. If used in a monorepo with multiple projects, there should be a separate config file for each project.

This document describes all the options available in the Apollo config and defines which are required vs. optional.

Client projects

Client projects are configured through a top level client key in the config.

1
module.exports = {
2
  client: { ... }, 
3
};

client.service

Required –– the CLI and VS Code extension rely on knowledge of your schema to show you "intellisense" (eg. autocomplete on fields, metrics annotations, query validation).

There are a few different ways you can link your client to a schema:

1. Use the Apollo schema registry
2. With a remote endpoint (from a running server)
3. With a local schema file

Option 1: Use the Apollo schema registry

To link your client to a schema through the Apollo schema registry, you'll need to have at least one version of your schema published to the registry.

With Apollo Studio set up, you can point your client directly to your graph's schema by specifying your graph's name in your Apollo config, like so:

1
module.exports = {
2
  client: {
3
    service: "my-apollo-service", // the name of your graph in Studio
4
  },
5
};

If you're tracking different versions of your schema in the registry using graph variants, you can link your client to a specific variant like so:

1
module.exports = {
2
  client: {
3
    service: "my-apollo-service@staging", // "staging" is the graph variant we're using
4
  },
5
};

If a graph variant is not specified, Apollo tools will fall back to the default value of current.

Option 2: Link a schema from a remote endpoint

Remote endpoints can be used to pull down a schema from a running service. This can be configured like so:

1
module.exports = {
2
  client: {
3
    service: {
4
      name: "github",
5
      url: "https://api.github.com/graphql",
6
      // optional headers
7
      headers: {
8
        authorization: "Bearer lkjfalkfjadkfjeopknavadf",
9
      },
10
      // optional disable SSL validation check
11
      skipSSLValidation: true,
12
    },
13
  },
14
};

Option 3: Link a schema from a local file

In some cases you may have a locally generated file with your schema that you want to link. This can be either a .graphql file with the schema in SDL form or a saved introspection result in .json. To link your client project to a local schema file, configure it like so:

1
module.exports = {
2
  client: {
3
    service: {
4
      name: "my-service-name",
5
      localSchemaFile: "./path/to/schema.graphql",
6
    },
7
  },
8
};

client.includes

Optional –– by default, Apollo tools will look under a ./src directory to find all operations and SDL to extract.

Client projects often contain client-side schema definitions for local state with Apollo Client. To make sure the Apollo CLI and VS Code extension can find these files and read them correctly, you may need to tell Apollo which folders to look for your schema and queries in like so:

1
module.exports = {
2
  client: {
3
    includes: ['./imports/**/*.js'], // array of glob patterns
4
    service: ...
5
  },
6
};

client.excludes

Optional –– by default, Apollo tools will exclude **/node_modules and **/__tests___ when looking for your queries and schema files.

If you want Apollo to ignore any of your other folders when looking for queries and schema definitions, adjust your config like so:

1
module.exports = {
2
  client: {
3
    excludes: ['**/__tests__/**/*'], // array of glob patterns
4
    service: ...
5
  },
6
};

client.tagName

Optional –– custom tagged template literal.

When using GraphQL with JavaScript or TypeScript projects, it is common to use the gql tagged template literal to write out operations. Apollo tools will be looking through your files for the gql tag to extract your queries, so if you use a different template literal, you can configure it like so:

1
module.exports = {
2
  client: {
3
    tagName: "graphql", 
4
    service: ...
5
  }
6
};

client.addTypename

Optional –– Apollo will by default add the __typename field to all your operations automatically and to all your generated types during codegen.

GraphQL clients like Apollo Client often add the __typename field to operations automatically when they're sent over the wire. This can come in really handy for things like caching, but it can be turned off by adding addTypename: false to the client config:

1
module.exports = {
2
  client: {
3
    addTypename: false, 
4
    service: ...
5
  }
6
};

client.clientOnlyDirectives, client.clientSchemaDirectives

Optional –– By default, Apollo projects support the following client-side directives:

• @client for local state
• @rest for using apollo-link-rest
• @connection for custom pagination with Apollo Client
• @type for dynamic type names with apollo-link-rest

Client side applications can use custom directives on their queries that aren't meant to be sent to the server. Configuration of client side directives beyond the defaults listed above can be set up like so:

1
module.exports = {
2
  client: {
3
    clientOnlyDirectives: ["connection", "type"],
4
    clientSchemaDirectives: ["client", "rest"],
5
    service: ...
6
  }
7
};

clientOnlyDirectives are directives that should be stripped out of the operation before being sent to the server. An example of this is the @connection directive.

clientSchemaDirectives are directives that indicate a portion of the operation that is not meant to be sent to the server. These directives are removed as well as the fields they are placed on. An example of this type of directive is the @client directive.

Server projects

Server projects are configured through a top level service key in the config.

1
module.exports = {
2
  service: { ... }, 
3
};

Defining a service key in your Apollo config will provide the CLI with the information it needs to perform commands like apollo service:push and apollo service:check. You can set up the schema for your service to load in one of two ways:

1. Using a remote endpoint
2. Using a local schema file

Option 1: Remote endpoint

Remote endpoints can be used to pull down a schema from a running service. This can be configured like so:

1
module.exports = {
2
  service: {
3
    endpoint: {
4
      url: "https://api.github.com/graphql", // defaults to http://localhost:4000
5
      headers: {
6
        // optional
7
        authorization: "Bearer lkjfalkfjadkfjeopknavadf",
8
      },
9
      skipSSLValidation: true, // optional, disables SSL validation check
10
    },
11
  },
12
};

Option 2: Local schema

In some cases you may have a locally generated file with your schema that you want to link. This can be either a .graphql file with the schema in SDL form or a saved introspection result in .json. To link your client project to a local schema file, configure it like so:

1
module.exports = {
2
  service: {
3
    localSchemaFile: "./path/to/schema.graphql",
4
  },
5
};