AWS releases Strands Harness open-source AI agent tool

Amazon Web Services (AWS) has released Strands Harness, an open-source tool designed to simplify the creation of high-performance AI agents. The tool functions as a “harness” that provides the necessary environment for Large Language Models (LLMs) to operate autonomously.

Key features of the harness include long-term context memory, access to files, networks, and execution environments, and integration with external tools like databases and APIs. It also includes guardrails to ensure agents operate within defined policies and permissions. The tool supports a wide range of LLMs, including Amazon Bedrock, Anthropic Claude, OpenAI GPT, Google Gemini, and Ollama.

Strands Harness is designed to be model-agnostic and deployment-flexible, allowing users to swap LLMs or deploy the agents in any environment equipped with Linux containers. It includes built-in capabilities such as shell execution, file manipulation, and web searching. Additionally, the tool features automated context management, including prompt caching and context summarization to prevent overflows.