GUI-HARVEST: Self-Improving GUI Agents through Evidence-Driven Harness Evolution
cs.LG, cs.AI
Submitted: 2026-10-01
Updated: 2026-10-01
Comments: Preprint
Code: https://github.com/GaryYang12345/GUI-HARVEST
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Sources
- Agent S: An Open Agentic Framework that Uses Computers Like a Human
- GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning
- Qwen3-VL Technical Report
- Recovering Policy-Induced Errors: Benchmarking and Trajectory Synthesis for Robust GUI Agents
- Fantastic Adaptive Taxonomies and How to Use Them
- From Failed Trajectories to Reliable LLM Agents: Diagnosing and Repairing Harness Flaws
- On the Reliability of Computer Use Agents
- Scaling Agents for Computer Use
- VLAA-GUI: Knowing When to Stop, Recover, and Search, A Modular Framework for GUI Automation
- HarnessEvolve: Learning from Reference Trajectories for Reliable Agent Self-Evolution
- Feedback Descent: Open-Ended Text Optimization via Pairwise Comparison
- Meta-Harness: End-to-End Optimization of Model Harnesses
- Agentic Harness Engineering: Observability-Driven Automatic Evolution of Coding-Agent Harnesses
- HarnessBank: Semantic Gene-Bank Search with Gated Verification for Agent-Harness Self-Evolution
- UI-TARS: Pioneering Automated GUI Interaction with Native Agents
- Harness-R1: Learning to Edit Executable Runtime Harnesses from Agent Failure Trajectories
- Reflexion: Language Agents with Verbal Reinforcement Learning
- Learning from Failure: Inference-Time Self-Improvement for Computer-Use Agents
- Step-level Optimization for Efficient Computer-use Agents
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks