cs.SENov 15, 2026

HydroMind: A Checkable Agent for Autonomous Hydrological Research

Authors: Zheng Fang, Yiqun Sun, Simin Qu, Shijie Jiang

Organizations: College of Hydrology and Water Resources, Hohai University, Nanjing, China · Department of Biogeochemical Integration, Max Planck Institute for Biogeochemistry, Jena, Germany

Abstract

General-purpose coding agents have written code for hydrological and Earth-system models. In these studies the researchers decided in advance how the model would be evaluated, for example the test period, the scoring function, and the standard the model had to meet. No study has tested what happens when the agent makes these decisions itself: how many of its claims are supported by its own data, code, and results, and whether readers can check them. Both may depend on the agent's harness, that is, the tools it can call and the written instructions it follows. We compared three harnesses: the agent's own harness; this harness plus the modeling tools of HydroMind, a hydrological modeling system we developed for general-purpose agents, which run every model through fixed commands and store its settings, trained model, and predictions; and both plus HydroMind's research procedure, under which the agent writes a research design before running any model and records every change to it. Under each harness, the same agent studied four open questions in hydrological model development and was given only the data. Under the three harnesses, the experiments submitted 23, 20, and 41 claims, each stating that one model or approach is better than, worse than, or similar to another. A claim does not hold if the difference recomputed from the experiment's own code and predictions disagrees with it, or if the code did not compare the two models or approaches the claim names. Under the three harnesses, 30%, 20%, and 5% of the claims did not hold. A statistical test, confidence interval, or decision rule that lets readers judge the stated difference was reported for 48%, 53%, and 83% of the claims. With the language model unchanged, changing only the harness reduced the share of claims that do not hold and increased the share that readers can check.

Explore similar work

Nov 15, 2026

Anatomy of an Agent Harness: Decisions, Invariants and Corner Cases

An agent harness is the program around a language model that turns its replies into actions: it builds each request, runs the tools the model asks for, decides what the model may do without asking, stores the conversation, recovers from failures, and shows a person what happened. Most of these decisions are made in code and rarely written down, so each harness meets the same corner cases again. This paper writes them down as invariants to test. It gives a system model with trust boundaries; 30 decision cards in six groups, each with options and costs, an invariant, the corner cases that break it, a test and an evidence tag; six cause classes for failures; a test method built on scripted providers, recorded behaviour and planted changes; and a checklist. The cards come from building one desktop harness, Crewroom, the author's own, and from manually testing ten other harnesses and studying open-source harnesses as reference; some patterns in the cards were adopted from those, so the catalogue is not independent of them. Of the 83 invariant clauses, 56 are tagged Tested, 12 Believed and 15 Open. The checklist is applied to the author's harness as one worked example, by its builder with no second rater: of its 30 items, 11 hold, 14 do not and 5 were not tested. It is not a ranking. The paper makes no claims about answer quality or cost or other platforms, and does not evaluate or rank other harnesses.
Oct 23, 2026

When Agents Fail: Model or Harness--Who's to Blame?

LLM agents combine models with execution harnesses, so agentic benchmark performance reflects their joint behavior. We argue that harnesses should therefore be evaluated through their interactions with underlying models. We develop a framework combining five complementary measures: (i) Valid Completion Rate (VCR), how reliably agents finish normally and deliver evaluable results, irrespective of task correctness; (ii) Model-Harness Fit, how well a harness supports task achievement through valid completion for a given model; (iii) Cross-Model Fit, a harness's overall suitability across evaluated models; (iv) Interaction Residuals, pairing-specific execution strengths and weaknesses hidden by component-level summaries; and (v) Token Load, the token demands of agent execution, including failed attempts. We evaluate four general-purpose harnesses (OpenCode, Codex CLI, Pi CLI, and DeepAgents) with four core models, plus two fixed native model-harness systems, on six benchmarks covering context management, tool use, long-context planning, multimodal interaction, and artifact handoff. Including native references and stress conditions, 111 evaluation settings yield 52,168 trajectories. Harness rankings vary across models and benchmarks; task performance and valid completion can favor different pairings, and similar task scores can accompany distinct execution profiles. OpenCode ranks first in cross-model performance, while DeepSeek V4 Flash Preview maintains consistently high Mean VCR across harnesses on the shared four-benchmark panel. These findings support evaluating model-harness interactions, training transferable harness-use capabilities in models, and harness designs adapted to the model and task.
May 18, 2026

HydroAgent: Closing the Gap Between Frontier LLMs and Human Experts in Hydrologic Model Calibration via Simulator-Grounded RL

Calibrating distributed hydrologic models is a critical bottleneck across operational water resources management - streamflow prediction, reservoir operation, drought monitoring, infrastructure design, and flood forecasting all depend on it. Each basin demands an expert to translate hydrograph signatures into adjustments of a high-dimensional parameter vector, and the resulting workflow does not transfer between watersheds. We ask: can frontier large language model (LLM) agents replace the human hydrologic modeler, and if not, what would it take? We benchmark nine frontier LLM agents - Claude Opus 4.6/4.7, Sonnet 4.6, GPT-5/5.4/5.4-pro, and Gemini 2.5-pro/3.1-pro/3-flash - on the operational CREST distributed hydrologic model used by the U.S. National Weather Service for flash-flood forecasting. Best-of-twenty-rounds Nash-Sutcliffe Efficiency (NSE) across four held-out gauges spanning 329-40,792 km2 ranges from -0.16 (GPT-5.4) to 0.75 (Sonnet 4.6); the ceiling reproduces across all three vendors and capability tiers, with the strongest models concentrating in the 0.65-0.75 band, and no model reaches the human-expert reference except Opus-4.7 on one gauge. We argue this gap is not a parameter-count problem but a domain-grounding problem. We then propose HYDROAGENT, fine-tuning open-weight Qwen3-4B with supervised fine-tuning on 2,576 expert calibration trajectories and Group-Relative Policy Optimization using NSE as a verifiable reward from online CREST simulations - reinforcement learning with simulation feedback (RLSF). For Earth system science, a small domain-tuned policy with simulator-in-the-loop RL is a more compute-efficient and physically faithful path than scaling generic frontier models, and the multi-modal richness of Earth data - remote sensing, in-situ time series, and forecaster narrative - makes domain agents a leveraged direction for AI in physical science.