shrimp is a small program that can reliably upload large files to Amazon S3. My personal use case is to upload large files to S3 over a slow residential connection, and shrimp is optimized for this use case.


  • shrimp supports most of the arguments used for aws s3 cp. In many cases you can simply replace aws s3 cp with shrimp and everything will work.
  • shrimp has interactive keyboard controls that lets you limit the bandwidth used for the upload (you can also specify an initial limit with -bwlimit, e.g. -bwlimit=2.5m for 2.5 MB/s). While the upload is in progress, press ? to see the available keyboard controls.
  • shrimp can resume the upload in case it fails for whatever reason (just re-run the command). Unlike the aws cli, shrimp will never abort the multipart upload in case of failures (please set up a lifecycle policy for this!).
  • shrimp supports automatically attaching SHA256 checksums to the object metadata if a SHA256SUMS file present in the working directory. You can use s3sha256sum to verify the object after it has been uploaded. See here for more information.


  • MFA support.
  • Automatically change bandwidth limit on a schedule.

Keep in mind:

  • shrimp will always use a multipart upload, so do not use it for small files.
  • shrimp uploads a single part at a time.

Current status: testing phase. Please do not use it for important files just yet. Please report any bugs.


Precompiled binaries will be provided at a later date. For now you can install using go install:

$ go install[email protected]


$ shrimp -help
Usage: shrimp [parameters] <LocalPath> <S3Uri>
LocalPath must be a local file.
S3Uri must have the format s3://<bucketname>/<key>.

  -bwlimit string
    	Bandwidth limit. (e.g. "2.5m")
  -cache-control string
    	Specifies caching behavior for the object.
  -content-disposition string
    	Specifies presentational information for the object.
  -content-encoding string
    	Specifies what content encodings have been applied to the object.
  -content-language string
    	Specifies the language the content is in.
  -content-type string
    	A standard MIME type describing the format of the object data.
  -expected-bucket-owner string
    	The account ID of the expected bucket owner.
  -metadata string
    	A map of metadata to store with the object in S3. (JSON syntax is not supported)
  -profile string
    	Use a specific profile from your credential file.
  -storage-class string
    	Storage class. (e.g. "STANDARD" or "DEEP_ARCHIVE")
  -tagging string
    	The tag-set for the object. The tag-set must be encoded as URL Query parameters.
    	Print version number.


View Github