Strands harness-sdk is an Apache-2.0 SDK for building in-process AI-agent harnesses that run model-agnostic across Bedrock, Anthropic, OpenAI and local Ollama.
Yes, Strands harness-sdk is open source under the Apache-2.0 license.
Strands harness-sdk is free to use.
Yes, Strands harness-sdk can be self-hosted (the source is available under the Apache-2.0 license).
Apache-2.0 SDK (Python and TypeScript) from the AWS-backed Strands Agents project for building AI-agent harnesses that run in-process, with no hosted control plane. It is model-agnostic — first-class Amazon Bedrock plus Anthropic, OpenAI and Gemini, and local models via Ollama — so an agent you build stays portable across providers and can run entirely on your own hardware instead of being tied to one vendor's runtime.
The quiet tax on agent infrastructure is the runtime lock-in: build on a hosted control plane and your agents assume that provider's models, quotas and availability. harness-sdk is the portable answer — an in-process SDK where the harness is yours, the model is a swap, and the same agent can point at Bedrock in production or Ollama on a laptop. For a team standardizing how it ships agents and wanting an escape hatch from any single vendor, that portability plus Apache-2.0 is worth the build cost. The honest read: it is AWS-backed (the strands-agents org, Bedrock first-class), not the official aws/ org, and "local model" means Ollama specifically. It is a builder's kit, not a turnkey product — the leverage is in what you assemble on it.