aws-harness — skill for Claude Code: deploy an agent on AWS
aws-harness — a skill for Claude Code that creates a new AI agent or wraps an existing one and deploys it to AWS Bedrock AgentCore, with tools, cross-session memory, and sandboxed execution.
no license ★ 301 updated 2026-10-23
npx skills add hoodini/ai-agents-skills --skill aws-harness -a claude-codeWhat aws-harness does
The skill explains what an agent harness is (the runtime around a model: agent loop, tool execution, memory, identity, observability), then shows two fully-working approaches. Path A uses the CLI (agentcore create/dev/deploy/invoke) to build a new agent and deploy it immediately. Path B wraps an agent you already built—using Strands, LangGraph, or custom code—via SDK or FastAPI + Docker, then ships it to ECR.
Every command and import in the skill is verified against AWS and framework docs. You need an AWS account with configured credentials, Bedrock model access, and Python 3.10+. Path A also requires Node.js 20+ and AWS CDK. Docker with buildx support is needed for containerization.
What it helps with
- scaffold and deploy a new AI agent on AWS with AgentCore CLI
- wrap an existing agent (Strands, LangGraph, custom) and ship it to Bedrock
- add tools, cross-session memory, and user identity to an agent
- run an agent locally before deploying
- package an agent in a Docker container and push to ECR
Example prompts for the agent
- Create a new Strands agent on AgentCore:
agentcore create --name MyBot --framework Strands --defaults - Deploy an existing agent to AWS:
agentcore deploy - Run your agent locally before shipping:
agentcore dev
How to install aws-harness
1Skills CLI (Vercel)
npx skills add hoodini/ai-agents-skills --skill aws-harness -a claude-code2Manually: put the files in place
~/.claude/skills/aws-harness/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/aws-harness/
1Skills CLI (Vercel)
npx skills add hoodini/ai-agents-skills --skill aws-harness -a codex2Manually: put the files in place
.agents/skills/aws-harness/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add hoodini/ai-agents-skills --skill aws-harness -a cursor2Manually: put the files in place
.cursor/skills/aws-harness/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install hoodini/ai-agents-skills skills/aws-harness2Manually: put the files in place
.github/skills/aws-harness/SKILL.md1Skills CLI (Vercel)
npx skills add hoodini/ai-agents-skills --skill aws-harness -a gemini-cli2Platform CLI
gemini skills install https://github.com/hoodini/ai-agents-skills --path skills/aws-harnessManually: put the files in place
.devin/skills/aws-harness/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about aws-harness
How do I install aws-harness in Claude Code?
Run npx skills add hoodini/ai-agents-skills --skill aws-harness -a claude-code, then open ~/.claude/skills/aws-harness/SKILL.md for the full guide.
What does aws-harness do?
It shows how to create or wrap an AI agent and deploy it to AWS Bedrock AgentCore. The skill includes working examples with CLI and SDK for both paths: building a new agent from scratch and wrapping existing code.
What is the difference between Path A and Path B?
Path A is for new agents: the CLI (agentcore create) scaffolds a project and deploys via CDK. Path B is for existing agents: you wrap your code through the SDK or package it as a container and push to ECR manually.