Search papers, labs, and topics across Lattice.
The paper introduces Natural-Language Agent Harnesses (NLAHs), which externalize agent harness behavior in editable natural language, and the Intelligent Harness Runtime (IHR), a shared runtime for executing these harnesses. This approach aims to improve the transferability, comparability, and study of agent harnesses by decoupling the high-level control logic from the controller code. Experiments on coding and computer-use benchmarks demonstrate the operational viability of NLAHs, the impact of module ablation, and the feasibility of code-to-text harness migration.
Stop burying your agent harness logic in code: NLAHs let you express it in natural language, making it portable, editable, and analyzable.
Agent performance increasingly depends on \emph{harness engineering}, yet harness design is usually buried in controller code and runtime-specific conventions, making it hard to transfer, compare, and study as a scientific object. We ask whether the high-level control logic of an agent harness can instead be externalized as a portable executable artifact. We introduce \textbf{Natural-Language Agent Harnesses} (NLAHs), which express harness behavior in editable natural language, and \textbf{Intelligent Harness Runtime} (IHR), a shared runtime that executes these harnesses through explicit contracts, durable artifacts, and lightweight adapters. Across coding and computer-use benchmarks, we conduct controlled evaluations of operational viability, module ablation, and code-to-text harness migration.
Tsinghua AI