HarnessBridge: Learnable Bidirectional Controller for LLM Agent Harness
- URL: http://arxiv.org/abs/2606.12882v1
- Date: Thu, 11 Jun 2026 04:18:37 GMT
- Title: HarnessBridge: Learnable Bidirectional Controller for LLM Agent Harness
- Abstract summary: We introduce HarnessBridge, a lightweight learnable harness controller that parameterizes the agent--environment interface as a bidirectional projection.<n>On Terminal-Bench2.0 and SWE-bench Verified, HarnessBridge matches or surpasses strong specialized harnesses while substantially reducing token usage and trajectory length.
- Score: 80.53160974102161
- License: http://creativecommons.org/licenses/by/4.0/
- Abstract: Large language models are increasingly deployed as agents for long-horizon tasks, yet their performance is shaped not only by model capability and environment design, but also by the harness that mediates agent--environment interaction. Existing harnesses are largely manually engineered, making them difficult to scale as trajectories grow longer and interactions become more complex. In this work, we ask whether harness can be generated by a learnable plug-in module that can be trained in an end-to-end fashion. We introduce HarnessBridge, a lightweight learnable harness controller that parameterizes the agent--environment interface as a bidirectional projection. HarnessBridge learns two bidirectional projections: observation projection, which distills raw trajectories into compact, decision-relevant states, and action projection, which converts proposed actions into executable transitions or trajectory-grounded rejections. We train HarnessBridge on a harness supervision dataset via unified instruction tuning. On Terminal-Bench~2.0 and SWE-bench Verified, HarnessBridge matches or surpasses strong specialized harnesses while substantially reducing token usage and trajectory length, and generalizes from smaller generators to larger commercial models.
Related papers
- HarnessBandit: Joint Learnability-Transferability Scheduling for Multi-Harness Agentic Reinforcement Learning [40.61476978001808]
We develop HarnessBandit, an online scheduler that selects one harness per step.<n>We train Qwen3.5-2B across six harnesses on ClawGym and evaluate on PinchBench and ClawEval.
arXiv Detail & Related papers (2026-09-12T06:22:45Z) - Guava: An Effective and Universal Harness for Embodied Manipulation [74.34187069605844]
We develop an end-to-end training pipeline that distills embodied manipulation capabilities into a 4B open-source model.<n>Results suggest that a well-designed harness can serve as a scalable, model-agnostic interface for embodied manipulation.
arXiv Detail & Related papers (2026-06-16T18:09:26Z) - A Switching Beamformer for Highly Non-Stationary Environments [6.752249951921619]
This paper introduces the Universal Switching Beamformer (USB), which integrates competitive sequential prediction into the beamforming architecture.<n>The USB achieves the agility of short-window estimators and the precision of long-term integration, providing a principled solution for tracking highly non-stationary scenes.
arXiv Detail & Related papers (2026-06-07T00:44:39Z) - Agentic Harness Engineering: Observability-Driven Automatic Evolution of Coding-Agent Harnesses [57.20181537213498]
Agentic Harness Engineering (AHE) is a closed loop that automates harness engineering.<n>AHE addresses challenges through three matched observability pillars.<n>Ten AHE lift pass@1 on Terminal-Bench 2 from 69.7% to 77.0%, surpassing the human-designed harness Codex-CLI.<n>On SWE-bench-verified it tops aggregate success at 12% fewer tokens than the seed, and on Terminal-Bench 2 it yields +5.1 to +10.1pp cross-family gains.
arXiv Detail & Related papers (2026-04-28T16:55:02Z) - DriveMamba: Task-Centric Scalable State Space Model for Efficient End-to-End Autonomous Driving [47.573692944838115]
DriveMamba is a Task-Centric Scalable paradigm for efficient E2E-AD.<n>It integrates sequential task relation modeling, implicit correspondence learning and long-term temporal fusion into a single-stage Unified Mamba decoder.<n>Extensive experiments conducted on nuScenes and Bench2Drive datasets demonstrate the superiority, generalizability and great efficiency of DriveMamba.
arXiv Detail & Related papers (2026-02-09T11:48:29Z) - Building a Foundational Guardrail for General Agentic Systems via Synthetic Data [76.18834864749606]
LLM agents can plan multi-step tasks, intervening at the planning stage-before any action is executed-is often the safest way to prevent harm.<n>Existing guardrails mostly operate post-execution, which is difficult to scale and leaves little room for controllable supervision at the plan level.<n>We introduce AuraGen, a controllable engine that synthesizes benign trajectories, injects category-labeled risks with difficulty, and filters outputs via an automated reward model.
arXiv Detail & Related papers (2025-10-10T18:42:32Z) - Activation Space Interventions Can Be Transferred Between Large Language Models [0.5541644538483947]
We show that safety interventions can be transferred between models through learned mappings of their shared activation spaces.<n>We demonstrate this approach on two well-established AI safety tasks: backdoor removal and refusal of harmful prompts.<n>We also propose a new task, textitcorrupted capabilities, where models are fine-tuned to embed knowledge tied to a backdoor.
arXiv Detail & Related papers (2025-03-06T13:38:44Z) - Detectors for Safe and Reliable LLMs: Implementations, Uses, and Limitations [76.19419888353586]
Large language models (LLMs) are susceptible to a variety of risks, from non-faithful output to biased and toxic generations.
We present our efforts to create and deploy a library of detectors: compact and easy-to-build classification models that provide labels for various harms.
arXiv Detail & Related papers (2024-03-09T21:07:16Z)
This list is automatically generated from the titles and abstracts of the papers in this site.
This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.