Strands Agents is an open‑source monorepo that provides a Python and TypeScript SDK for constructing AI agents without an external control plane. The core offering, Strands harness, is instantiated with a single create_harness() (Python) or createHarness() (TypeScript) call and delivers a pre‑configured agent loop that includes lifecycle controls (turn limits, token budgets, cancellation, stop reasons), tool integration, structured output, Model Context Protocol (MCP), multi‑agent patterns, persistent memory and sessions, model‑agnostic backends, streaming, guardrails, tracing, and evaluation harnesses. First‑class model providers include Amazon Bedrock, Anthropic, OpenAI, Gemini, and Ollama, with support for custom endpoints. The repository splits functionality into harness‑py/ and harness‑ts/ for the ready‑to‑use agent, strands‑py/ and strands‑ts/ for the lower‑level SDK, strands‑cli/ for terminal prototyping, site/ for the Astro/Starlight documentation, and team/ for governance. Installation requires Python 3.10+ (pip install strands‑harness) or Node.js 22+ (npm install @strands‑agents/harness). Example usage shows agents invoking natural‑language requests such as “Find the slowest test in this repo and explain why it's slow” or mathematical queries via the calculator tool. Contributions follow the project’s Apache 2.0 license and contributing guide.