Amazon S3 + Transloadit

Build the integration

This Template imports one JPEG, resizes it to fit within 1200×1200, and stores the result under processed/.

1. Create an S3 bucket and a least-privilege IAM principal for that bucket.
2. Save the AWS values in Transloadit as Template Credentials named my_s3_credentials.
3. Paste the Instructions into a Template and start an Assembly from that Template.

{
  "steps": {
    "imported": {
      "robot": "/s3/import",
      "credentials": "my_s3_credentials",
      "bucket_region": "us-east-1",
      "path": "incoming/photo.jpg"
    },
    "resized": {
      "robot": "/image/resize",
      "use": "imported",
      "width": 1200,
      "height": 1200,
      "resize_strategy": "fit"
    },
    "exported": {
      "robot": "/s3/store",
      "use": "resized",
      "credentials": "my_s3_credentials",
      "bucket_region": "us-east-1",
      "acl": "bucket-default",
      "path": "processed/${file.url_name}"
    }
  }
}

Authentication

• Create a dedicated IAM principal and save its bucket, access key ID, secret access key, and bucket region as Transloadit Template Credentials.
• Grant only the bucket and object actions needed by the import and store Steps. The Robot documentation contains starting policies for each direction.
• Keep the AWS secret out of Assembly Instructions and application code by referring to the saved credential name.

Limitations and operational notes

• Directory imports can return many files and may require pagination; use a narrow path prefix for predictable batches.
• Server-side encryption with AWS KMS requires additional KMS permissions beyond ordinary S3 object access.
• Object ownership and bucket policies can reject ACL writes. The recipe uses bucket-default so PutObjectAcl is not required.