This toolkit empowers AI coding agents to build, deploy, and manage applications on AWS by providing essential tools, knowledge, and guardrails. It integrates seamlessly with popular coding agents like Claude Code, Codex, and Kiro, enabling them to interact with AWS services effectively and securely.

How It Works

The toolkit leverages the AWS MCP Server, a managed service enabling agents to interact with AWS via the Model Context Protocol. It offers plugins that bundle MCP Server configurations and agent skills for simplified installation with specific agents (e.g., Claude Code, Codex), or allows direct configuration and skill installation for other agent frameworks. Skills are curated packages loaded on demand, ensuring agents only retrieve relevant resources for a given task. This approach provides full AWS API coverage, sandboxed script execution, real-time documentation access, and enterprise controls like IAM condition keys, CloudWatch metrics, and CloudTrail logging, specifically designed to differentiate and secure agent actions.

Quick Start & Requirements

• Primary Install/Run:
  • Claude Code / Plugin Marketplace: /plugin install aws-core@agent-toolkit-for-aws
  • Codex (terminal): codex plugin marketplace add aws/agent-toolkit-for-aws, then launch Codex and run /plugins.
  • Kiro: Configure MCP Server in .kiro/settings/mcp.json, then npx skills add aws/agent-toolkit-for-aws/skills.
  • Other agents: Configure AWS MCP Server directly, then npx skills add aws/agent-toolkit-for-aws/skills.
• Prerequisites: uv installed; AWS account with locally configured credentials (required for API calls and script execution, not for documentation search or skill discovery).
• Links: User guide (setup, configuration, reference), AWS MCP Server documentation.

Highlighted Details

• Plugins (aws-core, aws-agents, aws-data-analytics) bundle MCP Server configuration and agent skills for simplified installation with Claude Code and Codex.
• The aws-core plugin covers foundational AWS services, infrastructure-as-code (CDK/CloudFormation), serverless, containers, storage, observability, billing, SDK usage, and deployment.
• The aws-agents plugin facilitates building AI agents on AWS using Amazon Bedrock and AgentCore.
• The aws-data-analytics plugin supports data lake, analytics, and ETL workflows with S3 Tables, AWS Glue, and Athena.
• The AWS MCP Server provides full AWS API coverage, sandboxed Python script execution, real-time documentation access, and enterprise controls like IAM condition keys, CloudWatch metrics, and CloudTrail audit logging.

Maintenance & Community

This toolkit is the successor to AWS Labs MCP servers, skills, and plugins. AWS Labs will continue to accept contributions, with successful elements transitioning to the Agent Toolkit. Specific community links or contributor details were not present in the provided README snippet.

Licensing & Compatibility

Licensed under the Apache-2.0 License. This license is permissive and generally compatible with commercial use and closed-source linking.

Limitations & Caveats

Requires the uv package manager and locally configured AWS credentials for execution. It is presented as the recommended path forward, succeeding earlier AWS Labs tools. No specific limitations, alpha status, or known bugs were detailed in the provided README snippet.