Harnesses for Inference-Time Alignment over Execution Trajectories
Harness engineering has emerged as an important inference-time technique for large language model (LLM) agents, aiming to improve long-term performance through task decomposition and guided execution. However, more elaborate harnesses are not uniformly better: increasing decomposition or guidance can sometimes improve execution, but can also reduce final task success. We study harness design through the lens of inference-time trajectory alignment. This perspective separates harness into two mech
Record details
Published: 15 May 2026
Source: arXiv
Category: Research
Topics: Safety & alignment · Agents & autonomy
Retrieved: 14 July 2026
Related evidence
These records share source-supplied organisations, an exact publisher byline, automatic topics or regions. The reason is shown on every link; related does not mean supporting, agreeing with or verifying this record.
SkiP: When to Skip and When to Refine for Efficient Robot Manipulation
arXiv · 15 May 2026
Body-Grounded Perspective Formation and Conative Attunement in Artificial Agents
arXiv · 16 May 2026
Ensemble Monitoring for AI Control: Diverse Signals Outweigh More Compute
arXiv · 14 May 2026
From I/O to Code with Discovery Agent
arXiv · 14 May 2026
Beyond Binary: Reframing GUI Critique as Continuous Semantic Alignment
arXiv · 14 May 2026
Event-Grounded Sparse Autoencoders for Vision-Language-Action Policies
arXiv · 17 May 2026
How to cite this record
ethics.ai (15 May 2026), “Harnesses for Inference-Time Alignment over Execution Trajectories,” evidence record 4276, https://ethics.ai/record/4276 (originally published by arXiv).
Use and limitations
This page is a stable index and citation surface for a source record. ethics.ai did not author the underlying report and has not independently verified every claim. Automatic topics may be imperfect. For consequential use, quote and cite the original publisher.