awesome-harness-engineering

Why we included this project

Building the scaffolding around an AI agent is a different job from training or picking the model, and this curated list maps that territory well. It gathers pointers to the pieces that often decide whether an agent works in practice: context delivery, tool interfaces, planning artifacts, verification loops, memory systems, and sandboxes. The sections are organized around the problems engineers actually hit, with coverage of evals, MCP, permissions, observability, and orchestration, so you can go straight to the area you are stuck on. Since it is a curated index rather than a tool itself, its real value is the editorial judgment behind the selection, which makes it a sensible first stop before you commit to any one framework.