strands-agents/harness-sdk

Build an agent harness and control it end-to-end. Open-source SDK for production AI agents in Python & TypeScript - any model, any cloud.

What it solves

Strands Agents provides a standardized way to build and run AI agents without having to manually write complex agent loops. It eliminates the need for a hosted control plane, running directly in the user's process, while providing the essential infrastructure for production-ready agents, such as lifecycle controls, memory, and observability.

How it works

The project offers two paths for development: a high-level "harness" for rapid deployment and a low-level SDK for granular control. The harness (create_harness()) provides a pre-configured agent with benchmarked defaults for models, tools, and memory. The SDKs (available in Python and TypeScript) allow developers to manually wire up the agent loop, select model providers, and define custom tools.

Who it’s for

Developers who want to build AI agents in Python or TypeScript and prefer to own their agent loop and infrastructure rather than relying on a hosted service.

Highlights

  • Model Agnostic: First-class support for Amazon Bedrock, Anthropic, OpenAI, Gemini, and others.
  • Coded-in Controls: Built-in lifecycle management including turn limits, token budgets, and cancellation.
  • Extensibility: Support for Model Context Protocol (MCP), multi-agent patterns, and structured output.
  • Oversight: Integrated tracing by default and guardrails to catch errors before execution.
  • Cross-Language Support: Full SDKs for both Python and TypeScript.

Related

  • Project
    strands-agents/samples

    Strands Agents Samples provides Python and TypeScript example projects that demonstrate how to build, deploy, and integrate AI agents using the Strands model‑driven SDK. It includes tutorials ranging from a simple hello‑world agent to advanced patterns like multi‑agent systems, RAG, cloud deployment, and edge/robotics integration.

  • Project
    strands-agents/box

    Strands Box is an open‑source sandbox that isolates AI agents and enforces Dogwood‑written semantic policies on file, program, and network actions, with unified rule evaluation, credential injection, and audit logging.

  • Dispatch
    OpenAI Agents API Launch – Managed Harness for Durable Cloud Agents

    OpenAI introduced the Agents API, a managed Codex harness that lets developers run durable, tool‑enabled agents in hosted or self‑hosted sandboxes, with session persistence and multi‑agent support.

  • Project
    gotempsh/temps

    Temps is an open‑source, self‑hosted platform that bundles AI‑driven deployment, monitoring, analytics, error tracking, email, and secure sandbox execution into a single dashboard, letting you replace many SaaS tools with one binary.

  • Project
    livekit/agents

    LiveKit Agents is a Python SDK for building real‑time, multimodal voice agents (LLM‑driven assistants) that run on LiveKit servers. It provides unified adapters for STT/LLM/TTS, job scheduling, WebRTC/telephony integration, turn detection, MCP tool support, and a built‑in test framework.