S3 AppSync Resolver

This article will walk through adding an AppSync Resolver to fetch objects from S3


  • General knowledge of GraphQL
  • General knowledge of AWS AppSync and Apache Velocity

S3 Data Source

Since S3 is not available as a Data Source out of the box, we have to specify it as an HTTP Data Source

The easiest way to do this is with cloudformation. You can view a cdk implementation of this here.

To add S3 as a Data Source manually, use the following values:

  • Data source name: <something>
  • Data source type: HTTP Endpoint
  • HTTP endpoint: <bucket_domain>

Your bucket domain is going to look something like the following:


Request / Response Templates

Request Template

This template calls the GetObject API and return the contents.

Response Template

This template returns the response from GetObject.

Error Handling

The response template includes basic error handling. If you need more fine-grained control, you can read about the errors that S3 returns here.

Sample Query

Finally if we wanted to test our resolver, you can use this example query:

If this object existed it would return the contents. In our case since the object doesn't exist, we get a response like this:

Data source CDK Construct

This snippet contains a CDK Construct that implements the `appsync.BackedDataSource` class. You can use this with CDK to specify S3 as a data source.

You can see this in the Sample Stack

Sample CDK Stack

This is a sample stack that creates a new graphql api, specifies S3 as a data source, and creates resolvers with the templates that we just walked through.

To use this you will just need to specify a `schema.graphql` and download the `request.vtl` and `response.vtl` templates into a place that the stack can find.

Warning: This stack may create resources that will cost you money