Six Open-Source Skills for SageMaker
Updated
AWS shows six open-source skills from Hugging Face Skills that deploy Hugging Face models on Amazon SageMaker AI.
On September 18, 2026, AWS described six open-source agent skills for deploying Hugging Face models on Amazon SageMaker AI. You install the six skills from the Hugging Face Skills GitHub repo, point a coding agent at a Hugging Face model, and AWS says you get a real-time endpoint with autoscaling, Amazon CloudWatch alarms, the serving container from the AWS Deep Learning Containers catalog, and a verified teardown path.
The planner, hf-cloud-sagemaker-deployment-planner, orchestrates five others: local AWS context, an isolated Python environment, a SageMaker execution-role check, serving-image selection, and production defaults that attach autoscaling, alarms, and tags. AWS says the skills use only Python and the AWS CLI and work unchanged on macOS, Linux, and Windows. A real-time endpoint is the default. The same skills also support real-time with scale-to-zero, serverless inference, asynchronous inference, batch transform, and Amazon Bedrock Custom Model Import.
Key facts
- Posted September 18, 2026: AWS describes installing six skills from the Hugging Face Skills GitHub repo to deploy Hugging Face models on Amazon SageMaker AI.
- AWS says you point a coding agent at a model and get a real-time endpoint with autoscaling, Amazon CloudWatch alarms, a serving container from the AWS Deep Learning Containers catalog, and a verified teardown path.
- The six skills are open source, use only Python and the AWS CLI, and work unchanged on macOS, Linux, and Windows. A planner skill orchestrates the other five.
- AWS says a real-time endpoint is the default. The skills also support real-time with scale-to-zero, serverless inference, asynchronous inference, batch transform, and Amazon Bedrock Custom Model Import.
Source: Deploy Hugging Face models on Amazon SageMaker AI with coding agents · Amazon Web Services