Core Skills in the AWS Agent Toolkit: Complete Guide to the aws-core Plugin

The AWS Agent Toolkit provides 15 core skills through the aws-core plugin, each defined as a self-contained markdown document (SKILL.md) that enables AI agents to perform authenticated AWS operations across serverless, containers, observability, and cost optimization.

The aws/agent-toolkit-for-aws repository delivers these core skills as the primary interface between AI coding agents—such as Claude Code, Codex, and Cursor—and AWS services. Each skill is a modular package containing step-by-step instructions, reference documentation, and reusable assets that agents load on demand to execute specific AWS tasks.

What Are Core Skills in the AWS Agent Toolkit?

Core skills are self-contained capability modules packaged under the aws-core plugin in the aws/agent-toolkit-for-aws repository. Each skill follows a standardized structure centered on a SKILL.md file that contains concise descriptions, procedural instructions, and links to reference materials.

According to the source code, these skills reside in plugins/aws-core/skills/ with each subdirectory representing a distinct functional domain. The agent loads these markdown definitions dynamically, retrieving only the relevant guidance for the current task to minimize context noise and ensure up-to-date best practices.

Complete List of Core Skills by Domain

The aws-core plugin ships 15 distinct skills organized across infrastructure, security, SDKs, and operations:

Authentication and Security

Compute and Serverless

Infrastructure as Code

Observability and Operations

Cost and AI Services

SDK Support

Skill Architecture and Key Files

Each core skill follows a consistent directory structure within plugins/aws-core/skills/. The SKILL.md file serves as the entry point, while supporting materials reside in standardized subdirectories:

  • references/: In-depth markdown documentation for specific service implementations (e.g., lambda.md, s3.md).

  • assets/: Helper scripts, templates, and boilerplate code (e.g., powertools-handler.py for Lambda functions).

  • hooks/: Security and validation scripts such as secret-safety.py, which redacts sensitive credentials before agent transmission.

The plugin manifest at plugins/aws-core/.cursor-plugin/plugin.json exposes these skills to Cursor, while plugins/aws-core/.mcp.json configures the MCP server endpoint for agent communication.

Practical Usage: Invoking Core Skills

Agents consume core skills through natural language commands or structured skill invocations. Below are executable patterns for common workflows.

Installing the aws-core Plugin

Add the toolkit to your agent's marketplace and install the core plugin:


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

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

# Install the core plugin

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

Generating Serverless Resources

Use the AWS Serverless skill to scaffold Lambda functions with proper structure:

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

The skill retrieves templates from aws-serverless/references/lambda.md and returns a ready-to-deploy SAM template along with a minimal handler file.

Optimizing Costs with the Billing Skill

Query cost recommendations using the AWS Billing and Cost Management skill:

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

This invocation reads references/service-optimization.md and returns a comparison table of Savings Plans versus Reserved Instances.

Working with AWS SDKs

Generate boto3 code with proper pagination and error handling:

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

The skill generates Python code based on references/s3.md, ensuring correct pagination patterns and exception handling.

Configuring Amazon Bedrock

Provision Bedrock models with the necessary IAM policies:

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

The skill produces the IAM policy, runtime configuration, and sample invocation scripts from references/sdk-converse-api-python.md.

Security Controls and Secret Handling

The aws-core plugin implements safety guardrails through plugins/aws-core/hooks/secret-safety.py. This hook automatically redacts AWS credentials and sensitive tokens before they are transmitted back to the agent, ensuring that authentication materials never leak into chat logs or model contexts.

Summary

  • The AWS Agent Toolkit delivers 15 core skills via the aws-core plugin, each defined in plugins/aws-core/skills/<skill-name>/SKILL.md.

  • Skills cover authentication, serverless, containers, infrastructure as code, observability, cost optimization, and SDK usage across Python, JavaScript, and Swift.

  • Each skill packages reference documentation (references/) and reusable assets (assets/) for comprehensive guidance.

  • Security is enforced through pre-transmission hooks like secret-safety.py that prevent credential leakage.

  • Agents invoke skills through natural language commands, receiving vetted code templates and deployment configurations.

Frequently Asked Questions

How do I install the aws-core plugin in my coding agent?

Install the plugin by adding the repository to your agent's marketplace with npx skills add aws/agent-toolkit-for-aws/skills, then run /plugin install aws-core@claude-plugins-official. The plugin manifest at plugins/aws-core/.cursor-plugin/plugin.json handles the rest of the configuration.

What file format defines a core skill?

Each core skill is defined by a SKILL.md file located in its respective directory under plugins/aws-core/skills/. This markdown file contains the skill description, procedural instructions, and references to supporting documentation in the references/ subdirectory.

How do core skills handle AWS authentication?

The Signing‑in to AWS skill manages credential configuration and MCP server sessions. Additionally, the hooks/secret-safety.py script automatically redacts sensitive credentials before they are returned to the agent, preventing accidental exposure of access keys or tokens.

Can I use core skills with any AI coding agent?

Yes, the aws-core plugin is designed for compatibility with multiple agents including Claude Code, Codex, Cursor, and Kiro. The MCP server configuration in plugins/aws-core/.mcp.json standardizes the interface, allowing any MCP-compliant client to load and invoke the skills.

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 →