---
id: 20260910-T0-08
title: "SCAFFOLD：让网页Agent通过递归参数技能抽象实现自我改进"
title_en: "SCAFFOLD Enables Self-Improving Web Agents via Recursive Parametric Skill Abstraction"
url: https://ai.daily.yangsir.net/daily/20260910-T0-08
issue_date: 2026-09-10
publish_date: 2026-09-09T04:00:00.000Z
category: research
source_name: "arXiv cs.AI"
source_url: https://arxiv.org/abs/2609.05511
---

# SCAFFOLD：让网页Agent通过递归参数技能抽象实现自我改进

arXiv新论文提出SCAFFOLD框架，核心思路是让网页Agent不再孤立地学习每个任务并丢弃积累的程序性知识，而是通过递归参数技能抽象（recursive parametric skill abstraction）持续累积和复用技能。框架旨在应对跨网站视觉丰富、长时程且界面多变的任务场景。

## English Version

**SCAFFOLD Enables Self-Improving Web Agents via Recursive Parametric Skill Abstraction**

A new arXiv paper proposes SCAFFOLD, a framework that enables web agents to avoid learning tasks in isolation and discarding procedural knowledge. Instead, it uses recursive parametric skill abstraction to continuously accumulate and reuse skills, targeting visually rich, long-horizon interfaces that change across websites.

---

**来源**：[arXiv cs.AI](https://arxiv.org/abs/2609.05511)

**详情页**：https://ai.daily.yangsir.net/daily/20260910-T0-08

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*