LongHorizon-Harness: Advancing Long-Horizon Agents for Real-World Tasks
2026-08-03 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors found that when large language models work on long, multi-step tasks, it’s hard for the models to keep track of what’s been done and decide what to do next. To fix this, they created LongHorizon-Harness, a system that keeps a clear and separate record of the task’s progress, checking facts carefully after each step. Their method uses three parts: one to plan, one to act, and one to check the results. This approach helped several language models perform better on different difficult benchmark tests. Overall, the authors show that carefully managing task state and verification improves long tasks for language model agents.
large language modelslong-horizon taskstask state managementtool useself-assessmentagent harnesssubtask executionenvironment verificationbenchmarkreasoning
Authors
Ziyu Ma, Hailang Huang, Shun Zou, Yong Wang, Shidong Yang, Yiming Hu, Fei Wei, XiangXiang Chu
Abstract
Large language model (LLM) agents increasingly undertake long-horizon tasks that require sustained reasoning, tool use, and revision across many interdependent steps. However, existing agent harnesses maintain task execution, task state, and completion assessment within a growing context, making the state difficult to track and allowing incorrect self-assessments to propagate into later decisions. We reformulate long-horizon execution as a task-state management problem and propose LongHorizon-Harness, which maintains the task state explicitly outside execution and updates it only with facts independently verified from the environment. Its Manage-Execute-Audit(MEA) loop uses a manager to maintain the task state and determine the next subtask, a fresh-context executor to perform it, and a read-only auditor to verify the resulting environment state before the next round. A lightweight AgentAdapter supports interchangeable model and harness backends without modifying their native agent loops. LongHorizon-Harness improves Qwen~3.7-Plus from 51.8% to 80.7% on WeaveBench, from 69.7% to 77.2% on Terminal-Bench~2.1, and from 2.8% to 8.3% on OSWorld~2.0. It also raises Claude Opus~4.7 from 20.0% to 34.3% on an OSWorld2.0 subset, demonstrating consistent gains across models, harnesses, and interaction domains.