aws-strands — Claude Code skill: Python agents with Strands SDK
aws-strands — a skill for building Python agents with Strands Agents SDK, an open-source framework.
no license ★ 301 updated 2026-10-23
npx skills add hoodini/ai-agents-skills --skill aws-strands -a claude-codeWhat aws-strands does
The skill contains examples and patterns for Strands—a framework that manages agent logic, tool binding, and agent-to-agent coordination. By default it targets AWS Bedrock with Claude Sonnet, but BedrockModel and other provider classes let you swap models and endpoints. Core features: the @tool decorator wraps functions so the model can call them (it reads docstrings to understand when), prebuilt tools from strands-agents-tools, context windowing via SlidingWindowConversationManager, and async streaming of results. For orchestrating multiple agents, choose one of three patterns—agents-as-tools (an agent becomes a tool for another), Swarm (agents hand off dynamically), or Graph (a deterministic DAG where output from one node feeds into the next).
Running locally requires Python and pip install strands-agents strands-agents-tools. To ship an agent to AWS as a serverless endpoint with managed memory and observability, pair this with the aws-harness skill (AgentCore runtime).
What it helps with
- write an agent that calls your own functions as tools
- coordinate multiple agents through delegation and handoffs
- build a deterministic pipeline with structured agent routing
- tune context windowing and message retention
- stream agent results asynchronously as they execute
Example prompts for the agent
- Build an agent that counts words in text by exposing a
word_countfunction as a tool - Create a researcher and writer agent where the writer calls the researcher as a tool
- Set up a Graph pipeline: research node outputs to a write node, which feeds into an edit node
How to install aws-strands
1Skills CLI (Vercel)
npx skills add hoodini/ai-agents-skills --skill aws-strands -a claude-code2Manually: put the files in place
~/.claude/skills/aws-strands/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/aws-strands/
1Skills CLI (Vercel)
npx skills add hoodini/ai-agents-skills --skill aws-strands -a codex2Manually: put the files in place
.agents/skills/aws-strands/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add hoodini/ai-agents-skills --skill aws-strands -a cursor2Manually: put the files in place
.cursor/skills/aws-strands/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install hoodini/ai-agents-skills skills/aws-strands2Manually: put the files in place
.github/skills/aws-strands/SKILL.md1Skills CLI (Vercel)
npx skills add hoodini/ai-agents-skills --skill aws-strands -a gemini-cli2Platform CLI
gemini skills install https://github.com/hoodini/ai-agents-skills --path skills/aws-strandsManually: put the files in place
.devin/skills/aws-strands/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-strands
How do I install aws-strands in Claude Code?
Run npx skills add hoodini/ai-agents-skills --skill aws-strands -a claude-code, then open ~/.claude/skills/aws-strands/SKILL.md. For other IDEs the path changes: Cursor uses .cursor/skills/, Codex uses .agents/skills/, GitHub Copilot uses gh skill install hoodini/ai-agents-skills aws-strands.
What does aws-strands do?
It's a skill with examples and patterns for Strands Agents SDK—a Python framework for writing agent logic, tool definitions, and multi-agent systems. It defaults to AWS Bedrock but plugs into other model providers. To deploy an agent onto AWS infrastructure, use aws-harness.
How does aws-strands relate to aws-harness?
aws-strands is the SDK skill (how you write agent logic). aws-harness is the deployment runtime (how you host agents on AWS with managed memory and identity). They work together: code with Strands, deploy with AgentCore.