Run GraphQL API queries and mutations

Tier: Free, Premium, Ultimate Offering: GitLab.com, Self-managed, GitLab Dedicated

This guide demonstrates basic usage of the GitLab GraphQL API.

Read the GraphQL API style guide for implementation details aimed at developers who wish to work on developing the API itself.

Running examples

The examples documented here can be run using:

GraphiQL.
Command line.
Rails console.

GraphiQL

GraphiQL (pronounced “graphical”) allows you to run real GraphQL queries against the API interactively. It makes exploring the schema easier by providing a UI with syntax highlighting and autocompletion.

For most people, using GraphiQL will be the easiest way to explore the GitLab GraphQL API.

You can either use GraphiQL:

On GitLab.com.
On your self-managed GitLab instance on https://<your-gitlab-site.com>/-/graphql-explorer.

To get started, refer to the example queries and mutations.

Command line

You can run GraphQL queries in a curl request on the command line on your local computer. The requests POST to /api/graphql with the query as the payload. You can authorize your request by generating a personal access token to use as a bearer token. Read more about GraphQL Authentication.

Example:

GRAPHQL_TOKEN=<your-token>
curl "https://gitlab.com/api/graphql" --header "Authorization: Bearer $GRAPHQL_TOKEN" \
     --header "Content-Type: application/json" --request POST \
     --data "{\"query\": \"query {currentUser {name}}\"}"

To nest strings in the query string, wrap the data in single quotes or escape the strings with \\:

curl "https://gitlab.com/api/graphql" --header "Authorization: Bearer $GRAPHQL_TOKEN" \
    --header "Content-Type: application/json" --request POST \
    --data '{"query": "query {project(fullPath: \"<group>/<subgroup>/<project>\") {jobs {nodes {id duration}}}}"}'
      # or "{\"query\": \"query {project(fullPath: \\\"<group>/<subgroup>/<project>\\\") {jobs {nodes {id duration}}}}\"}"

Rails console

Tier: Free, Premium, Ultimate Offering: Self-managed, GitLab Dedicated

GraphQL queries can be run in a Rails console session. For example, to search projects:

current_user = User.find_by_id(1)
query = <<~EOQ
query securityGetProjects($search: String!) {
  projects(search: $search) {
    nodes {
      path
    }
  }
}
EOQ

variables = { "search": "gitlab" }

result = GitlabSchema.execute(query, variables: variables, context: { current_user: current_user })
result.to_h

Queries and mutations

The GitLab GraphQL API can be used to perform:

Queries for data retrieval.
Mutations for creating, updating, and deleting data.

In the GitLab GraphQL API, id refers to a Global ID, which is an object identifier in the format of "gid://gitlab/Issue/123". For more information, see Global IDs.

GitLab GraphQL Schema outlines which objects and fields are available for clients to query and their corresponding data types.

Example: Get only the names of all the projects the currently authenticated user can access (up to a limit) in the group gitlab-org.

query {
  group(fullPath: "gitlab-org") {
    id
    name
    projects {
      nodes {
        name
      }
    }
  }
}

Example: Get a specific project and the title of Issue #2.

query {
  project(fullPath: "gitlab-org/graphql-sandbox") {
    name
    issue(iid: "2") {
      title
    }
  }
}

Graph traversal

When retrieving child nodes use:

The edges { node { } } syntax.
The short form nodes { } syntax.

Underneath it all is a graph we are traversing, hence the name GraphQL.

Example: Get the name of a project, and the titles of all its issues.

query {
  project(fullPath: "gitlab-org/graphql-sandbox") {
    name
    issues {
      nodes {
        title
        description
      }
    }
  }
}

More about queries: GraphQL documentation

Authorization

If you’ve signed in to GitLab and use GraphiQL, all queries are performed as you, the authenticated user. For more information, read about GraphQL Authentication.

Mutations

Mutations make changes to data. We can update, delete, or create new records. Mutations generally use InputTypes and variables, neither of which appear here.

Mutations have:

Inputs. For example, arguments, such as which emoji reaction you’d like to add, and to which object.
Return statements. That is, what you’d like to get back when it’s successful.
Errors. Always ask for what went wrong, just in case.

Creation mutations

Example: Let’s have some tea - add a :tea: reaction emoji to an issue.

mutation {
  awardEmojiAdd(input: { awardableId: "gid://gitlab/Issue/27039960",
      name: "tea"
    }) {
    awardEmoji {
      name
      description
      unicode
      emoji
      unicodeVersion
      user {
        name
      }
    }
    errors
  }
}

Example: Add a comment to the issue. In this example, we use the ID of the GitLab.com issue. If you’re using a local instance, you must get the ID of an issue you can write to.

mutation {
  createNote(input: { noteableId: "gid://gitlab/Issue/27039960",
      body: "*sips tea*"
    }) {
    note {
      id
      body
      discussion {
        id
      }
    }
    errors
  }
}

Update mutations

When you see the result id of the note you created, take a note of it. Let’s edit it to sip faster.

mutation {
  updateNote(input: { id: "gid://gitlab/Note/<note ID>",
      body: "*SIPS TEA*"
    }) {
    note {
      id
      body
    }
    errors
  }
}

Deletion mutations

Let’s delete the comment, because our tea is all gone.

mutation {
  destroyNote(input: { id: "gid://gitlab/Note/<note ID>" }) {
    note {
      id
      body
    }
    errors
  }
}

You should get something like the following output:

{
  "data": {
    "destroyNote": {
      "errors": [],
      "note": null
    }
  }
}

We’ve asked for the note details, but it doesn’t exist anymore, so we get null.

More about mutations: GraphQL Documentation.

Update project settings

You can update multiple project settings in a single GraphQL mutation. This example is a workaround for the major change in CI_JOB_TOKEN scoping behavior.

mutation DisableCI_JOB_TOKENscope {
  projectCiCdSettingsUpdate(input:{fullPath: "<namespace>/<project-name>", inboundJobTokenScopeEnabled: false}) {
    ciCdSettings {
      inboundJobTokenScopeEnabled
    }
    errors
  }
}

Introspection queries

Clients can query the GraphQL endpoint for information about its schema by making an introspection query.

The GraphiQL Query Explorer uses an introspection query to:

Gain knowledge about the GitLab GraphQL schema.
Do autocompletion.
Provide its interactive Docs tab.

Example: Get all the type names in the schema.

{
  __schema {
    types {
      name
    }
  }
}

Example: Get all the fields associated with Issue. kind tells us the enum value for the type, like OBJECT, SCALAR or INTERFACE.

query IssueTypes {
  __type(name: "Issue") {
    kind
    name
    fields {
      name
      description
      type {
        name
      }
    }
  }
}

More about introspection: GraphQL documentation

Query complexity

The calculated complexity score and limit for a query can be revealed to clients by querying for queryComplexity.

query {
  queryComplexity {
    score
    limit
  }

  project(fullPath: "gitlab-org/graphql-sandbox") {
    name
  }
}

Sorting

Some of the GitLab GraphQL endpoints allow you to specify how to sort a collection of objects. You can only sort by what the schema allows you to.

Example: Issues can be sorted by creation date:

query {
  project(fullPath: "gitlab-org/graphql-sandbox") {
   name
    issues(sort: created_asc) {
      nodes {
        title
        createdAt
      }
    }
  }
}

Pagination

Pagination is a way of only asking for a subset of the records, such as the first ten. If we want more of them, we can make another request for the next ten from the server in the form of something like please give me the next ten records.

By default, the GitLab GraphQL API returns 100 records per page. To change this behavior, use first or last arguments. Both arguments take a value, so first: 10 returns the first ten records, and last: 10 the last ten records. There is a limit on how many records are returned per page, which is generally 100.

Example: Retrieve only the first two issues (slicing). The cursor field gives us a position from which we can retrieve further records relative to that one.

query {
  project(fullPath: "gitlab-org/graphql-sandbox") {
    name
    issues(first: 2) {
      edges {
        node {
          title
        }
      }
      pageInfo {
        endCursor
        hasNextPage
      }
    }
  }
}

Example: Retrieve the next three. (The cursor value eyJpZCI6IjI3MDM4OTMzIiwiY3JlYXRlZF9hdCI6IjIwMTktMTEtMTQgMDU6NTY6NDQgVVRDIn0 could be different, but it’s the cursor value returned for the second issue returned above.)

query {
  project(fullPath: "gitlab-org/graphql-sandbox") {
    name
    issues(first: 3, after: "eyJpZCI6IjI3MDM4OTMzIiwiY3JlYXRlZF9hdCI6IjIwMTktMTEtMTQgMDU6NTY6NDQgVVRDIn0") {
      edges {
        node {
          title
        }
        cursor
      }
      pageInfo {
        endCursor
        hasNextPage
      }
    }
  }
}

More about pagination and cursors: GraphQL documentation

Run GraphQL API queries and mutations

Running examples

GraphiQL

Command line

Rails console

Queries and mutations

Graph traversal

Authorization

Mutations

Creation mutations

Update mutations

Deletion mutations

Update project settings

Introspection queries

Query complexity

Sorting

Help & feedback

Docs

Product

Feature availability and product trials

Get Help