Skip to main content

​
Schema Definition

Prism’s GraphQL engine lets you define schemas entirely in Swift using a typed DSL. Every field, argument, and type is a first-class Swift value — you get compiler checks, autocomplete, and zero code generation.

​
Type System

GraphQL types map directly to PrismGraphQLType:
Available Types
.string       // GraphQL String
.int          // GraphQL Int
.float        // GraphQL Float
.boolean      // GraphQL Boolean
.id           // GraphQL ID
.nonNull(.string)          // String!
.list(.string)             // [String]
.list(.nonNull(.string))   // [String!]
.object("User")            // Custom object type
.enum("Status", ["ACTIVE", "INACTIVE"])  // Enum type
.input("CreateUserInput")  // Input type
Use .nonNull() to mark fields as required in your schema. Clients will get a validation error if a non-null field returns nil.

​
Defining Fields

Each field has a name, type, optional arguments, and a resolver function:
Field with Resolver
let nameField = PrismGraphQLField(
    name: "name",
    type: .nonNull(.string),
    args: [],
    description: "The user's full name",
    resolver: { info in
        let user = info.parentValue as? [String: Any]
        return user?["name"] as? String
    }
)

​
Fields with Arguments

Field with Arguments
let userField = PrismGraphQLField(
    name: "user",
    type: .object("User"),
    args: [
        PrismGraphQLArgument(name: "id", type: .nonNull(.id), description: "User ID")
    ],
    description: "Fetch a single user by ID",
    resolver: { info in
        let id = info.arguments["id"] as? String ?? ""
        return await UserService.find(id: id)
    }
)

​
Object Types

Group fields into object types:
User Type
let userType = PrismGraphQLObjectType(
    name: "User",
    fields: [
        "id": PrismGraphQLField(name: "id", type: .nonNull(.id), args: [], resolver: { info in
            (info.parentValue as? [String: Any])?["id"]
        }),
        "name": PrismGraphQLField(name: "name", type: .nonNull(.string), args: [], resolver: { info in
            (info.parentValue as? [String: Any])?["name"]
        }),
        "email": PrismGraphQLField(name: "email", type: .string, args: [], resolver: { info in
            (info.parentValue as? [String: Any])?["email"]
        })
    ],
    description: "A registered user"
)

​
Building a Complete Schema

1

Define your types

Create object types for each entity in your domain.
2

Define the Query type

The Query type is the entry point for all read operations.
3

Define Mutation type (optional)

Mutations handle create, update, and delete operations.
4

Assemble the schema

Combine everything into a PrismGraphQLSchema.

​
Blog Schema Example

Complete Blog Schema
// Post type
let postType = PrismGraphQLObjectType(
    name: "Post",
    fields: [
        "id": PrismGraphQLField(name: "id", type: .nonNull(.id), args: [], resolver: { info in
            (info.parentValue as? [String: Any])?["id"]
        }),
        "title": PrismGraphQLField(name: "title", type: .nonNull(.string), args: [], resolver: { info in
            (info.parentValue as? [String: Any])?["title"]
        }),
        "body": PrismGraphQLField(name: "body", type: .string, args: [], resolver: { info in
            (info.parentValue as? [String: Any])?["body"]
        }),
        "authorId": PrismGraphQLField(name: "authorId", type: .nonNull(.id), args: [], resolver: { info in
            (info.parentValue as? [String: Any])?["authorId"]
        })
    ]
)

// Query type — entry point
let queryType = PrismGraphQLObjectType(
    name: "Query",
    fields: [
        "posts": PrismGraphQLField(
            name: "posts",
            type: .list(.object("Post")),
            args: [],
            description: "List all posts",
            resolver: { _ in
                return [
                    ["id": "1", "title": "Hello Prism", "body": "Getting started...", "authorId": "1"],
                    ["id": "2", "title": "GraphQL in Swift", "body": "Native and fast...", "authorId": "1"]
                ]
            }
        ),
        "post": PrismGraphQLField(
            name: "post",
            type: .object("Post"),
            args: [PrismGraphQLArgument(name: "id", type: .nonNull(.id))],
            description: "Get a post by ID",
            resolver: { info in
                let id = info.arguments["id"] as? String
                // In reality, query your database here
                return ["id": id, "title": "Hello Prism", "body": "Content...", "authorId": "1"]
            }
        )
    ]
)

// Mutation type
let mutationType = PrismGraphQLObjectType(
    name: "Mutation",
    fields: [
        "createPost": PrismGraphQLField(
            name: "createPost",
            type: .object("Post"),
            args: [
                PrismGraphQLArgument(name: "title", type: .nonNull(.string)),
                PrismGraphQLArgument(name: "body", type: .string)
            ],
            resolver: { info in
                let title = info.arguments["title"] as? String ?? ""
                let body = info.arguments["body"] as? String
                // Save to database, return created post
                return ["id": UUID().uuidString, "title": title, "body": body, "authorId": "1"]
            }
        )
    ]
)

// Assemble schema
let schema = PrismGraphQLSchema(
    query: queryType,
    mutation: mutationType
)
The schema object is Sendable and can be shared across concurrent requests safely. Define it once at startup and pass it to your middleware.

​
Schema with Type Registry

For schemas with nested object types (e.g., a User that has Posts), register types in the schema so the executor can resolve nested selections:
Registering Types
var schema = PrismGraphQLSchema(query: queryType, mutation: mutationType)
schema.types["User"] = userType
schema.types["Post"] = postType
schema.types["Comment"] = commentType

​
What’s Next

Queries

Learn how to write resolvers and handle nested data

Mutations

Create, update, and delete data through GraphQL