Menu
AWS Identity and Access Management
User Guide

Actions, Resources, and Condition Keys for Amazon SageMaker

Amazon SageMaker (service prefix: sagemaker) provides the following service-specific resources, actions, and condition context keys for use in IAM permission policies.

References:

Actions Defined by Amazon SageMaker

You can specify the following actions in the Action element of an IAM policy statement. By using policies, you define the permissions for anyone performing an operation in AWS. When you use an action in a policy, you usually allow or deny access to the API operation or CLI command with the same name. However, in some cases, a single action controls access to more than one operation. Alternatively, some operations require several different actions. For details about the columns in the following table, see The Actions Table.

Actions Description Access Level Resource Types (*required) Condition Keys Dependent Actions
AddTags Adds or overwrites one or more tags for the specified Amazon SageMaker resource.

Tagging

endpoint*

endpoint-config*

model*

notebook-instance*

training-job*

CreateEndpoint Creates an endpoint using the endpoint configuration specified in the request.

Write

endpoint*

CreateEndpointConfig Creates an endpoint configuration that can be deployed using Amazon SageMaker hosting services.

Write

endpoint-config*

CreateModel Creates a model in Amazon SageMaker. In the request, you specify a name for the model and describe one or more containers.

Write

model*

CreateNotebookInstance Creates an Amazon SageMaker notebook instance. A notebook instance is an Amazon EC2 instance running on a Jupyter Notebook.

Write

notebook-instance*

CreatePresignedNotebookInstanceUrl Returns a URL that you can use from your browser to connect to the Notebook Instance.

Read

notebook-instance*

CreateTrainingJob Starts a model training job. After training completes, Amazon SageMaker saves the resulting model artifacts and other optional output to an Amazon S3 location that you specify.

Write

training-job*

DeleteEndpoint Deletes an endpoint. Amazon SageMaker frees up all the resources that were deployed when the endpoint was created.

Write

endpoint*

DeleteEndpointConfig Deletes the endpoint configuration created using the CreateEndpointConfig API. The DeleteEndpointConfig API deletes only the specified configuration. It does not delete any endpoints created using the configuration.

Write

endpoint-config*

DeleteModel Deletes a model created using the CreateModel API. The DeleteModel API deletes only the model entry in Amazon SageMaker that you created by calling the CreateModel API. It does not delete model artifacts, inference code, or the IAM role that you specified when creating the model.

Write

model*

DeleteNotebookInstance Deletes an Amazon SageMaker notebook instance. Before you can delete a notebook instance, you must call the StopNotebookInstance API.

Write

notebook-instance*

DeleteTags Deletes the specified set of tags from an Amazon SageMaker resource.

Tagging

endpoint*

endpoint-config*

model*

notebook-instance*

training-job*

DescribeEndpoint Returns the description of an endpoint.

Read

endpoint*

DescribeEndpointConfig Returns the description of an endpoint configuration, which was created using the CreateEndpointConfig API.

Read

endpoint-config*

DescribeModel Describes a model that you created using the CreateModel API.

Read

model*

DescribeNotebookInstance Returns information about a notebook instance.

Read

notebook-instance*

DescribeTrainingJob Returns information about a training job.

Read

training-job*

InvokeEndpoint After you deploy a model into production using Amazon SageMaker hosting services, your client applications use this API to get inferences from the model hosted at the specified endpoint.

Read

endpoint*

ListEndpointConfigs Lists endpoint configurations.

List

ListEndpoints Lists endpoints.

List

ListModels Lists the models created with the CreateModel API.

List

ListNotebookInstances Returns a list of the Amazon SageMaker notebook instances in the requester's account in an AWS Region.

List

ListTags Returns the tag set associated with the specified resource.

List

ListTrainingJobs Lists training jobs.

List

StartNotebookInstance Launches an EC2 instance with the latest version of the libraries and attaches your EBS volume.

Write

notebook-instance*

StopNotebookInstance Terminates the EC2 instance. Before terminating the instance, Amazon SageMaker disconnects the EBS volume from it. Amazon SageMaker preserves the EBS volume.

Write

notebook-instance*

StopTrainingJob Stops a training job. To stop a job, Amazon SageMaker sends the algorithm the SIGTERM signal, which delays job termination for 120 seconds.

Write

training-job*

UpdateEndpoint Updates an endpoint to use the endpoint configuration specified in the request.

Write

endpoint*

UpdateEndpointWeightsAndCapacities Updates variant weight, capacity, or both of one or more variants associated with an endpoint.

Write

endpoint*

UpdateNotebookInstance Updates a notebook instance. Notebook instance updates include upgrading or downgrading the EC2 instance used for your notebook instance to accommodate changes in your workload requirements. You can also update the VPC security groups.

Write

notebook-instance*

Resources Defined by SageMaker

The following resource types are defined by this service and can be used in the Resource element of IAM permission policy statements. Each action in the Actions table identifies the resource types that can be specified with that action. A resource type can also define which condition keys you can include in a policy. These keys are displayed in the last column of the table. For details about the columns in the following table, see The Resource Types Table.

Resource Types ARN Condition Keys
endpoint arn:${Partition}:sagemaker:${Region}:${Account}:endpoint/${EndpointName}
endpoint-config arn:${Partition}:sagemaker:${Region}:${Account}:endpoint-config/${EndpointConfigName}
model arn:${Partition}:sagemaker:${Region}:${Account}:model/${ModelName}
notebook-instance arn:${Partition}:sagemaker:${Region}:${Account}:notebook-instance/${NotebookInstanceName}
training-job arn:${Partition}:sagemaker:${Region}:${Account}:training-job/${TrainingJobName}

Condition Keys for Amazon SageMaker

SageMaker has no service-specific context keys that can be used in the Condition element of policy statements. For the list of the global context keys that are available to all services, see Available Keys for Conditions in the IAM Policy Reference.