Is it possible to download Docker image without installing Docker on the machine (download image from Docker Hub)? (Edited)

You can use the download-frozen-image-v2.sh script from the Moby project, which allows you to download an image from the Docker Hub without without having a Docker client installed. The script downloads all the image layers (blobs) and saves them in a specified directory, preserving the necessary structure. After that, you can combine the downloaded layers into a single tar archive, which can then be transferred to a host where Docker is installed and loaded using the docker load command.

Usage example:

1. Downloading the image:

Run the following bash command to download the debian:stable image into the /data/debian_image directory:

		

			
				
			
		
		bash download-frozen-image-v2.sh /data/debian_image debian:stable
	

In this step, the script reaches out to Docker Hub, retrieves the image manifest (which contains metadata about the image and its layers), and downloads each layer into the specified directory. The directory /data/debian_image will then hold all the files needed to reconstruct the image, maintaining the structure required by Docker.

2. Combining layers into an archive:

Once all layers are successfully downloaded, create a single tar archive with this command:

		

			
				
			
		
		tar -cC '/data/debian_image' . > debian.tar
	

This command packages the entire contents of /data/debian_image into the file debian.tar. The archive will contain all the layers and metadata needed next.

3. Transferring and importing:

Next, transfer the debian.tar archive to the target machine where Docker is installed. On the target machine, run:

		

			
				
			
		
		docker load < debian.tar
	

The docker load command imports the image from the tar archive, restoring it in Docker's local storage. The archive will contain all the layers and metadata needed to import the image into Docker.

In my experience, this approach really comes in handy when Docker installation isn't an option—whether because of strict security policies, limited resources, or just because the server doesn't have Docker. I've used the download-frozen-image-v2.sh script to whip up a portable image archive that I can easily move around and load into a Docker environment later. It's been a reliable solution when things are a bit tricky.

I wrote a PowerShell script to download Docker images. Here’s a snippet:

		

			
				
			
		
		$image = "ubuntu"
$tag = "latest"
$repository = "library/$image"

# Get access token for the repository
$token = (Invoke-RestMethod "https://auth.docker.io/token?service=registry.docker.io&scope=repository:$repository:pull").token

# Request the image manifest
$manifestUrl = "https://registry-1.docker.io/v2/$repository/manifests/$tag"
$headers = @{
    "Authorization" = "Bearer $token"
    "Accept"        = "application/vnd.docker.distribution.manifest.v2+json"
}
$manifest = Invoke-RestMethod -Uri $manifestUrl -Headers $headers

# Determine layer list (supporting both schema v2 and older manifests)
$layers = if ($manifest.layers) { $manifest.layers } else { $manifest.fsLayers }

# Download each layer
foreach ($layer in $layers) {
    $digest = if ($layer.digest) { $layer.digest } else { $layer.blobSum }
    $filename = ($digest -replace "[:/]", "_") + ".gz"
    $blobUrl = "https://registry-1.docker.io/v2/$repository/blobs/$digest"
    Invoke-WebRequest -Uri $blobUrl -Headers @{ "Authorization" = "Bearer $token" } -OutFile $filename
    Write-Host "Downloaded layer: $filename"
}
	

Frankly, if you have internet access to DockerHub without limits, it's easier to temporarily install Docker or Podman and use the standard Docker tools.

Workarounds like skopeo or the API method add complexity and potential compatibility issues.

I don't see why you would load the image if you can't use it.

Downloading a Docker image without Docker on a machine using HTTP API V2.

1. Run a request to retrieve the image manifest (metadata):

		

			
				
			
		
		bash

curl -sSL -H "Accept: application/vnd.docker.distribution.manifest.v2+json" \
https://registry.hub.docker.com/v2/library/ubuntu/manifests/latest
	

This returns a JSON manifest listing the image layers.

2. Parse the JSON to get a list of layers with their digest.

3. Download the layers:

For each digest, download the corresponding blob using a URL like:

		

			
				
			
		
		bash

curl -sSL https://registry.hub.docker.com/v2/library/ubuntu/blobs/ -o .tar.gz
	

4. Assemble the archive:

Once all layers are downloaded, manually bundle them into a tar archive preserving the Docker image structure.

This method involves some manual parsing and assembly, but it offers flexibility if you need to retrieve an image for specialized purposes.

Using Skopeo to download a docker image without using docker:

I rely on skopeo for this task because it allows me to seamlessly copy container images between registries and save them as either OCI images or Docker archives — all without needing a running Docker daemon. This is especially beneficial in environments where installing Docker is not an option due to security policies or resource constraints.

The process is straightforward. For example, you can download the Ubuntu image with the tag latest as a tar file by running:

		

			
				
			
		
		bash

skopeo copy docker://docker.io/library/ubuntu:latest docker-archive:ubuntu_latest.tar:latest
	

In this command, Skopeo connects to the Docker registry, fetches the specified image, and converts it into a tar archive. This tarball can then be used for later import into a local registry or further analysis.

It's important to ensure that you’re using an up-to-date version of Skopeo and that your operating system meets its dependency requirements. This not only guarantees compatibility but also takes advantage of the latest features and security updates provided by the tool.

Overall, Skopeo offers a robust and efficient solution for managing Docker containers and container images in environments where traditional Docker installation is not feasible.