awslabs/agent-plugins

Agent Plugins for AWS equip AI coding agents with the skills to help you architect, deploy, and operate on AWS.

What it solves

It provides AI coding agents with specialized skills and knowledge to architect, deploy, and operate applications on AWS. Instead of developers manually pasting long AWS guides into prompts, these plugins encode best practices and workflows as reusable capabilities, reducing context bloat and improving the reliability of agent-generated AWS infrastructure and code.

How it works

Plugins act as containers that package several types of expertise artifacts for AI assistants (such as Claude Code, Cursor, and Codex):

  • Agent skills: Step-by-step playbooks and structured workflows for complex tasks like deployment or architecture planning.
  • MCP servers: Connections to live external data, such as AWS documentation and real-time pricing APIs.
  • Hooks: Automated guardrails that trigger on specific developer actions, such as validating a SAM template after it is edited.
  • References: Documentation and configuration defaults that the agent can consult.

Who it’s for

Developers using AI coding agents who want to build, migrate, or manage applications on AWS without having to manually provide extensive domain-specific guidance to the LLM.

Highlights

  • Broad AWS Coverage: Includes plugins for serverless (Lambda, API Gateway), Amplify Gen 2, SageMaker AI, Amazon Location Service, and database migration.
  • Deployment Automation: The deploy-on-aws plugin can analyze a codebase, recommend services, estimate costs, and generate Infrastructure as Code (CDK/CloudFormation).
  • Codebase Analysis: A dedicated documentor plugin that analyzes AWS-deployed services to generate technical documentation with source-of-truth citations.
  • Multi-Agent Support: Compatible with Claude Code, Cursor, and Codex.