Paper Overview
Field: Machine Learning Authors: Jiaxi Li, Ke Deng, Yun Wang Published: 2025-06-01 arXiv: 2606.04391
Abstract
Language agents increasingly rely on reusable skills to improve multi-step web automation across related tasks. A growing line of work studies online skill learning, where agents continually induce skills from previous task trajectories and reuse them in future tasks on the fly. However, existing methods mainly reuse skills at the task-level: a fixed set of skills is retrieved based on the initial task instruction and then held fixed throughout execution. This static strategy is misaligned with web execution, where the appropriate next action depends not only on the task goal but also on the current webpage state, which often transitions into situations that the initial skills fail to cover.
To address this gap, the authors propose State-Grounded Dynamic Retrieval (SGDR), an online skill learning method that enables web agents to reuse skills progressively. SGDR consists of three components:
- Sliding-window extraction process: converts completed trajectories into reusable subroutines that can be invoked at intermediate execution states.
- Dual text-code representation: connects skill retrieval with executable actions.
- State-grounded dynamic retrieval mechanism: matches skills against both the task goal and the current webpage state.
Results
Experiments on WebArena across five domains show that SGDR consistently outperforms strong baselines, achieving average success rates of 37.5% with GPT-4.1 and 24.3% with Qwen3-4B — improvements of 10.6% and 10.0% respectively over the strongest baseline.
Paper: https://arxiv.org/abs/2606.04391