Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

Using a coding assistant to expedite your machine learning workflows

Overview

JupyterLab in Amazon SageMaker AI includes integrated coding assistant support through the Agent Context Protocol (ACP). By default, Kiro coding assistant is pre-configured in the chat panel, providing AI-powered code completion, debugging assistance, and interactive coding support directly within your JupyterLab environment.

When you use coding assistants in Amazon SageMaker AI JupyterLab, the space automatically loads relevant Amazon SageMaker AI Skills into your assistant's context. These Skills are loaded from the AWSLabs GitHub repository and provide specialized knowledge about SageMaker APIs, ML workflows, best practices, and common patterns, enabling your coding assistant to give more accurate, SageMaker-specific guidance.

Additionally, you can configure other ACP-compatible coding assistants of your choice, giving you flexibility to work with the tools that best fit your workflow. ACP-compatible assistants can benefit from the same Amazon SageMaker AI Skills integration when used within Amazon SageMaker AI JupyterLab.

What is Agent Context Protocol (ACP)?

The Agent Context Protocol (ACP) is an open protocol that standardizes communication between code editors and AI coding agents. This means you can switch between different coding assistants without learning new interfaces or workflows.

Minimum requirements

Getting Started

Step 1: Open or Create a SageMaker Space with JupyterLab

Step 2: Start Using Kiro in the Chat Panel:

Kiro requires authentication before you can use it as your coding assistant. The chat panel will guide you through the authentication process.

Note that the first time you use Kiro in a space, it will ask you to log in. In order to log in, follow the instructions provided by the chat, or follow here:

Select one of the 3 login options in the terminal:

Follow the instructions and screens for the selected option.

Example prompts:

Accessing Amazon SageMaker AI Skills in Kiro

Amazon SageMaker AI Skills are automatically available when you use Kiro in SageMaker JupyterLab. These Skills are loaded from the AWSLabs GitHub repository and stored in the .kiro/skills and .agent/skills folders within your JupyterLab environment, making them compatible with any agent that loads from these directories.

Skills are update-able, allowing you to benefit from the latest SageMaker best practices and API patterns as they evolve. To update your Skills, you can pull the latest versions from the AWSLabs repository. You can run the following command to update your skills for use with Kiro:

npx skills add https://github.com/awslabs/agent-plugins/tree/main/plugins/sagemaker-ai/skills --all --agent kiro-cli --copy

For other agents, please see the SageMaker AI Skills README for more information. To view the available Skills in your environment, navigate to the .kiro/skills folder in the JupyterLab file browser.

As part of new SageMaker Distribution (SMD) releases, we provide updated versions of the skills. We will update skills automatically as long as they have not been modified or deleted by the user within a space. If you manually update or modify your skills, please use the npx commands above to update or reset your skills.

Configuring Other Coding Assistants with JupyterLab AI

Amazon SageMaker AI JupyterLab supports any coding assistant that implements the Agent Context Protocol (ACP). Example assistants that support ACP include:

To use a different ACP-compatible coding assistant:

Please note that for Claude Code specifically, you can configure it to use Amazon Bedrock as the backend. Follow the prerequisites in the Claude code guide, specifically enabling Bedrock model access and providing your execution role access to bedrock:InvokeModel and bedrock:InvokeModelWithResponseStream. Then, create the following file to configure Claude Code to use Bedrock.

~/.claude/settings.json:

{
  "env": {
    "CLAUDE_CODE_USE_BEDROCK": "1"
  }
}

Switching Between Assistants

You can switch between different coding assistants at any time:

Each assistant maintains its own conversation context, so you can switch back and forth as needed for different tasks.

Switching Between Kiro Profiles

Kiro in Amazon SageMaker AI JupyterLab supports multiple profiles that are optimized for different workflows and use cases. You can switch between profiles to access different sets of capabilities and behaviors tailored to your current task. Amazon SageMaker AI JupyterLab comes with the following Kiro profiles:

To switch between Kiro profiles in JupyterLab:

Additional Resources