Strands Agents

Strands Agents is AWS’s open-source SDK for building production AI agents, with a model-driven approach that lets the language model itself plan, call tools and recover from errors. The project has collected more than 8,800 stars on GitHub under the Apache-2.0 license and was actively developed through 2026.

Overview

Strands takes the harness idea seriously: instead of hand-crafting rigid workflows, you give the agent a model, a system prompt and a set of tools, and the SDK’s agent loop handles reasoning, tool execution and conversation state end to end. It works with models from Amazon Bedrock, Anthropic, OpenAI and local providers, in both Python and TypeScript, and speaks the Model Context Protocol so agents can consume the growing MCP tool ecosystem. The same SDK scales from a five-line prototype to deployed, multi-agent production systems on any cloud.

Key Features

• Model-driven agent loop: reasoning, tool use and self-correction built in
• Python and TypeScript SDKs with the same core concepts
• Works with Bedrock, Anthropic, OpenAI, Ollama and other model providers
• Native MCP support plus a library of built-in and community tools
• Multi-agent patterns including swarms, graphs and agents-as-tools
• Deploy anywhere, from local runs to AWS Lambda, Fargate or any cloud

💳 Pricing

Strands Agents is free and open source under the Apache-2.0 license. You pay only for the model inference and infrastructure you run it on.

Strands comes from AWS, and it shows in the defaults: guardrails, observability hooks and deployment patterns assume the agent will end up in production, not just in a demo. A growing set of community tools — from web browsing to code execution — can be attached to an agent without changing its core loop.

Who Is It For

It is for developers — especially those already on AWS — who want to build agents in a few lines of code without being locked to one model provider or a heavyweight framework.