deep-dive

What Are Specialized Skills in the AWS Agent Toolkit? A Complete Guide to the 15 Core Capabilities

June 27, 2026 aws/agent-toolkit-for-aws ↗

Specialized skills in the AWS Agent Toolkit are self-contained markdown capabilities bundled in the aws-core plugin that enable AI coding agents to authenticate, deploy, and manage AWS infrastructure across 15 domains including serverless, containers, IAM, and Bedrock.

The AWS Agent Toolkit provides a structured framework for AI coding assistants to interact with AWS services through domain-specific expertise. These specialized skills are packaged as individual SKILL.md documents within the aws-core plugin in the aws/agent-toolkit-for-aws repository, and are loaded on demand by agents including Claude Code, Codex, Cursor, and Kiro.

Architecture of Specialized Skills

Each specialized skill follows a modular architecture designed for AI agent consumption. Located under plugins/aws-core/skills/, every skill contains a master SKILL.md file that defines capabilities, step-by-step instructions for the agent, and references to supplementary materials.

The integration layer is defined in plugins/aws-core/.mcp.json, which configures the MCP (Model Context Protocol) server endpoint, while plugins/aws-core/.cursor-plugin/plugin.json exposes the skill catalog to Cursor IDE users. Security is enforced through plugins/aws-core/hooks/secret-safety.py, which redacts sensitive credentials before they are transmitted back to the agent.

The 15 Specialized Skills in the AWS Agent Toolkit

Authentication and Security

Compute and Deployment

Infrastructure as Code

SDK Integration

Operations and Observability

Artificial Intelligence

Invoking Specialized Skills in Practice

Specialized skills are invoked through the agent interface after installing the aws-core plugin. The following examples demonstrate common usage patterns.

First, install the plugin from the aws/agent-toolkit-for-aws repository:


# Add the repository to the marketplace (Cursor, Kiro, etc.)

npx skills add aws/agent-toolkit-for-aws/skills

# Install the core plugin

/plugin install aws-core@claude-plugins-official

To generate a Lambda function using the AWS Serverless skill:

skill aws-serverless create-lambda \
  --runtime python3.11 \
  --handler handler.handle \
  --description "Process S3 upload events"

This command returns a ready-to-deploy SAM/CloudFormation snippet based on the guidance in aws-serverless/references/lambda.md.

For cost optimization using the Billing skill:

skill aws-billing-and-cost-management recommend-savings \
  --service ec2 \
  --region us-east-1

The skill reads references/service-optimization.md and returns Savings Plans versus Reserved Instances recommendations.

To generate Python SDK code with pagination handling:

skill aws-sdk-python-usage list-s3-objects \
  --bucket my-data-bucket \
  --prefix logs/

This produces a boto3 script that correctly implements pagination and error handling per references/s3.md.

For Bedrock model configuration:

skill amazon-bedrock create-model \
  --model-id anthropic.claude-v2 \
  --instance-type ml.g5.xlarge

The skill generates the required IAM policy, runtime configuration, and a sample invoke script from references/sdk-converse-api-python.md.

Summary

  • The AWS Agent Toolkit provides 15 specialized skills organized under the aws-core plugin, covering authentication, compute, infrastructure as code, SDKs, observability, and AI.
  • Each skill is a self-contained module with a SKILL.md master file, reference documentation in references/, and optional assets for complex operations.
  • Skills are loaded on demand by AI agents through the MCP server configuration in plugins/aws-core/.mcp.json, with IDE integration via plugins/aws-core/.cursor-plugin/plugin.json.
  • Security is enforced at the hook level by plugins/aws-core/hooks/secret-safety.py, which redacts credentials before transmission.
  • Practical invocation follows a consistent skill <skill-name> <action> pattern, returning vetted code snippets and deployment templates.

Frequently Asked Questions

How do I add a new specialized skill to the AWS Agent Toolkit?

To add a new specialized skill, create a directory under plugins/aws-core/skills/ containing a SKILL.md file, a references/ subdirectory for documentation, and an optional assets/ folder for scripts or templates. The skill automatically becomes available after reloading the plugin through the MCP server configuration.

What is the difference between core skills and the assets folder?

Core skills are defined by the SKILL.md file which contains instructions for the AI agent, while the assets/ folder contains executable resources like Python scripts (powertools-handler.py) or CloudFormation templates that the skill can deploy or reference during execution.

How does the AWS Agent Toolkit handle credential security?

The toolkit uses plugins/aws-core/hooks/secret-safety.py to intercept and redact AWS credentials, secrets, and sensitive configuration values before they are returned to the AI agent, ensuring that authentication tokens never appear in chat history or logs.

Can I use these skills with any AI coding agent?

Yes, the specialized skills are designed to work with any MCP-compatible agent including Claude Code, Codex, Cursor, and Kiro. The plugins/aws-core/.cursor-plugin/plugin.json provides native Cursor integration, while the .mcp.json configuration enables standard MCP protocol support for other agents.

Have a question about this repo?

These articles cover the highlights, but your codebase questions are specific. Give your agent direct access to the source. Share this with your agent to get started:

Share the following with your agent to get started:
curl -s "https://instagit.com/install.md"
Works with
Claude Codex Cursor VS Code OpenClaw Any MCP Client

Maintain an open-source project? Get it listed too →