Tracto registry
TractoAI is able to serve as a Docker registry for workloads running on the platform as well as for external uses. It is compatible with the standard Docker API.
It is available on the "cr" subdomain of your tenant's proxy address: cr.your-tenant.trtr.ai.
Images are stored in the Cypress distributed filesystem, which significantly speeds up their retrieval in the workloads running on TractoAI.
Cypress representation
Images in the registry correspond to the nodes in Cypress distributed filesystem. This means that you must adhere to a strict naming convention:
cr.<your-tenant>.trtr.ai/home/<your-tenant>/path/to/image-name:tag
Here /home/<your-tenant>/path/to must be an arbitrary existing directory in your tenant's home directory, and image-name and tag may be arbitrary.
Web interface
As any other Cypress objects, images may be viewed in the web interface. Navigate to any image in your tenant's home directory and click on it to view its contents. There are two image-specific tabs:
- Image - shows information about tags, their digests, sizes and other metadata
- Inspector - allows to inspect the image contents, including individual layers and their filesystems

Authentication
Authentication is done on a basis of your token.
For example, to push an image to the registry, you can use the following command:
docker login cr.your-tenant.trtr.ai --username tracto --password $YT_TOKEN
docker push cr.your-tenant.trtr.ai/home/your-tenant/ubuntu:2025-04-30-0a59
Make sure you have read/write permissions for the //home/your-tenant/ubuntu path. Without proper permissions, push and pull operations will fail.
Using images in operations
When running a MapReduce operation, it is possible to specify Tracto registry images as docker_image parameter without the need to pass authentication credentials. The system will automatically authenticate requests to Tracto registry on your behalf.
yt vanilla --tasks '{
task={
command="ls";
docker_image="cr.\<proxy_address\>/home/registry/ubuntu:2025-04-30-0a59"
};
job_count=1;
}'