Skip to content

Uploading Local Artifacts added in v1.21.0

Copilot supports uploading local files referenced from your addon templates to S3, and replacing the relevant resource properties with the uploaded S3 location. On copilot svc deploy or copilot svc package --upload-assets, certain fields on supported resources will be updated with a S3 location before the addons template is sent to CloudFormation. Your templates on disk will not be modified. To see the full list of resources that are supported, take a look at the AWS CLI documentation.

This feature can be used to deploy local Lambda functions stored in the same repo as another Copilot service. For example, to deploy a javascript Lambda function alongside a copilot service, you can add this resource to your addon template:

Example Lambda Function
    Type: AWS::Lambda::Function
      Code: lambdas/example/
      Handler: "index.handler"
      Timeout: 900
      MemorySize: 512
      Role: !GetAtt "ExampleFunctionRole.Arn"
      Runtime: nodejs20.x
exports.handler = function (event, context) {
    console.log('example event:', event);

On copilot svc deploy, the lambdas/example directory will be zipped and uploaded to S3, and the Code property will be updated to:

  S3Bucket: copilotBucket
  S3Key: hashOfLambdasExampleZip
before the addon template is uploaded and deployed by Copilot. If you specify a file, the file is directly uploaded to S3. If you specify a folder, the folder will be zipped before being uploaded to S3. For some resources that require a zip (e.g., AWS::Serverless::Function), a file will be zipped before upload as well.

File paths are considered relative to the parent of the copilot/ directory in your repo. For the above example, the folder structure would look like:

├── copilot
│   └── example-service
│       ├── addons
│       │   └── lambda.yml
│       └── manifest.yml
└── lambdas
    └── example
        └── index.js
Absolute paths are supported as well, though they may not work as well across multiple machines.

Example: DynamoDB Stream Processing Lambda

This example will walk through creating an Amazon Dynamo DB table with a lambda function connected to process events from the table's stream. This architecture could be useful if you have a service that needs to minimize latency on storing data, but can kick off a separate process that takes longer to process the data.



  1. Generate a DynamoDB table addon for your service by running copilot storage init (More info here!)
  2. Add the StreamSpecification property to the generated AWS::DynamoDB::Table resource:
      StreamViewType: NEW_AND_OLD_IMAGES
  3. Add a Lambda function, IAM Role, and Lambda event stream mapping resource, making sure to give access to the DynamoDB table stream in the IAM Role:
        Type: AWS::Lambda::Function
          Code: lambdas/record-processor/ # local path to the record processor lambda
          Handler: "index.handler"
          Timeout: 60
          MemorySize: 512
          Role: !GetAtt "recordProcessorRole.Arn"
          Runtime: nodejs20.x
        Type: AWS::IAM::Role
            Version: 2012-10-17
              - Effect: Allow
                  - sts:AssumeRole
          Path: /
            - !Sub arn:${AWS::Partition}:iam::aws:policy/service-role/AWSLambdaBasicExecutionRole
            - PolicyDocument:
                Version: 2012-10-17
                  - Effect: Allow
                      - dynamodb:DescribeStream
                      - dynamodb:GetRecords
                      - dynamodb:GetShardIterator
                      - dynamodb:ListStreams
                    # replace <table> with the generated table's resource name
                    Resource: !Sub ${<table>.Arn}/stream/*
        Type: AWS::Lambda::EventSourceMapping
          FunctionName: !Ref recordProcessor
          EventSourceArn: !GetAtt <table>.StreamArn # replace <table> here too
          BatchSize: 1
          StartingPosition: LATEST
  4. Write your lambda function:
    "use strict";
    const { unmarshall } = require('@aws-sdk/util-dynamodb');
    exports.handler = async function (event, context) {
      for (const record of event?.Records) {
        if (record?.eventName != "INSERT") {
        // process new records
        const item = unmarshall(record?.dynamodb?.NewImage);
        console.log("processing item", item);
  5. Run copilot svc deploy to deploy your lambda function!🎉 As your service adds records to the table, the lambda function will be triggered and can process new records.