In some cases, it might be useful to extend the GraphQL CRUD API that is being generated, using custom Types and Resolvers.
To illustrate this, let's assume we have an existing model Post with a field views. We want to add a custom mutation incrementPostsViews and build a custom resolver to increment post views on every call.

👉 1. Extending schema.gql

Let’s start by creating a new custom-schema.gql file at the same location as the schema.prisma file:
extend type Mutation { """ Increment post views by +1 """ incrementPostsViews(postId: Int!): Post }
Next time we run npx prisma generate, we want Prisma-AppSync to merge our custom-schema.gql with the default schema output. To do so, we edit the generator config as following:
generator appsync { // ... extendSchema = "./custom-schema.gql" }

👉 2. Extending resolvers.yaml

We want AWS AppSync to make the link between our new incrementPostsViews mutation (added at previous step) and our Lambda function resolver. To do so, we create a new custom-resolvers.yaml file at the same location as our schema.prisma file:
- typeName: Mutation fieldName: incrementPostsViews dataSource: prisma-appsync
Next time we run npx prisma generate, we want Prisma-AppSync to merge our custom-resolvers.yaml with the default resolvers config output. To do so, we edit our generator config as per the following:
generator appsync { // ... extendResolvers = "./custom-resolvers.yaml" }

👉 3. Lambda function handler

For our Lambda function handler (= API resolver function) to process our newly created incrementPostsViews mutation, we need to declare a new resolver:
return await prismaAppSync.resolve<'incrementPostsViews'>({ event, resolvers: { incrementPostsViews: async ({ args, prismaClient }: QueryParamsCustom) => { return await{ data: { views: { increment: 1 } }, where: { id: args.postId } }) }, } })
See full list of options in the Reference section.

👉 4. AppSyncStack function bundling

This step only applies if using the AWS CDK boilerplate provided with Prisma-AppSync.
Finally, we want to make sure that custom-schema.gql and custom-resolvers.yaml are both part of the bundle uploaded on AWS Lambda. To do so, we update the beforeBundling function inside cdk/index.ts with the below (adjust paths as necessary):
function: { bundling: { commandHooks: { beforeBundling(inputDir: string, outputDir: string): string[] { const schema = path.join(inputDir, 'prisma/schema.prisma') const gql = path.join(inputDir, 'prisma/custom-schema.gql') const yaml = path.join(inputDir, 'prisma/custom-resolvers.yaml') return [ `cp ${schema} ${outputDir}`, `cp ${gql} ${outputDir}`, `cp ${yaml} ${outputDir}`, ] }, }, } }
🚀🚀🚀 Done! Next time we deploy your API on AWS AppSync, we should be able to use the newly created incrementPostsViews mutation.