Harness

As local LLMs improve, harnesses are learning to pair them with frontier models. A look at the four collaboration patterns already shipping in open source.

Summarizes the six-component agent harness model: execution, tools, context, state, hooks, and evaluation.

Principles for designing AI harnesses: context, tools, verification, autonomy, observability, and composition.

Pattern where agents declare structured pre-tool intent for gating, auditability, and drift detection.