Tag
This paper studies harness design for LLM agents, separating it into task decomposition and guided execution, and shows that more elaborate harnesses are not uniformly better; it reveals failure modes and proposes partial harnesses as effective.