AWS releases Strands Harness open-source agent
AWS's Strands Agents team released Strands Harness on September 21, 2026, an open-source agent designed to run in any environment—locally or deployed across major cloud platforms.
The release positions Strands Harness as a fully assembled, state-of-the-art agent ready for immediate use, rather than a bare framework requiring extensive customization. "Today we're releasing Strands harness: a fully assembled state-of-the-art agent you can easily run locally or deploy to your favorite provider," AWS stated in the announcement.
Deployment Flexibility Across Clouds
Strands Harness supports deployment to Amazon ECS, Amazon Bedrock AgentCore, Google Cloud Run, Azure Container Apps, Modal, and Cloudflare Containers, giving users multi-cloud portability without vendor lock-in. The agent operates with models from Amazon Bedrock, Anthropic, OpenAI, and Google, allowing developers to select the LLM backbone that fits their use case and cost profile.
Open-Source Licensing and Developer Access
The project is available under the Apache 2.0 license, permitting free use, modification, and distribution in both open and commercial applications. This licensing choice lowers barriers to adoption for enterprises and independent developers building on top of the agent.
Market Context
The release arrives amid intensifying competition in the agent infrastructure space. By offering a pre-built, customizable agent rather than requiring teams to assemble components from scratch, AWS targets developers seeking to reduce time-to-production and operational overhead. The multi-cloud deployment support explicitly addresses cost concerns—a persistent friction point for organizations evaluating agent platforms.
Strands Harness is immediately available to developers via the official announcement and documentation channels. The combination of model flexibility, deployment optionality, and open-source accessibility positions this release as a developer-facing tool reducing barriers to agent deployment at scale.