# Yonsei team tests a unified web-access interface for agents

> Web$hell combines search, browser actions and revisiting earlier page states in one interface.

- Canonical URL: https://newsverum.com/en/news/yonsei-team-tests-unified-interface-for-web-agents
- Publisher: Newsverum
- Section: Technology
- Country: South Korea
- Published: 2026-10-13T00:51:41+00:00
- Updated: 2026-10-13T00:51:41+00:00
- Event / data period: 2026-10-09
- Other editions: [Türkçe](https://newsverum.com/tr/news/yonsei-team-tests-unified-interface-for-web-agents.md), [العربية](https://newsverum.com/ar/news/yonsei-team-tests-unified-interface-for-web-agents.md), [Français](https://newsverum.com/fr/news/yonsei-team-tests-unified-interface-for-web-agents.md), [Español](https://newsverum.com/es/news/yonsei-team-tests-unified-interface-for-web-agents.md), [Deutsch](https://newsverum.com/de/news/yonsei-team-tests-unified-interface-for-web-agents.md), [Português](https://newsverum.com/pt/news/yonsei-team-tests-unified-interface-for-web-agents.md)

Yonsei University researchers introduced Web$hell, an interface that brings web search, page interaction and revisiting previously gathered evidence into one system for AI agents. WebContract compiles pages into information and action objects; WebCache stores page states so agents can inspect earlier versions after an interaction.

In experiments, Web$hell used an average of 17.6 interaction turns per task on DeepWeb-Bench research tasks, compared with 52.4 for CLI-Any-Browser. The authors also report performance gains on web-search, navigation and multi-source research benchmarks. These measurements reflect specific agents, interfaces and benchmark settings; they do not establish success across general web use.

## Data

### Measure

DeepWeb-Bench multi-source research tasks; mean turns per task

Methodology: Values come from the paper's stated benchmark experiments; they are not field measurements or general performance guarantees.

| Measure | Reported value | Experimental scope |
|---|---|---|
| Mean turns per DeepWeb-Bench task with Web$hell | 17.6 | DeepWeb-Bench multi-source research tasks; mean turns per task |

Sources: <https://arxiv.org/html/2610.13416v1> · Download data: [JSON](https://newsverum.com/api/v1/articles/yonsei-team-tests-unified-interface-for-web-agents/data/reported-result-1?lang=en&format=json) · [CSV](https://newsverum.com/api/v1/articles/yonsei-team-tests-unified-interface-for-web-agents/data/reported-result-1?lang=en&format=csv)

### Measure

DeepWeb-Bench multi-source research tasks; CLI-Any-Browser comparison

Methodology: Values come from the paper's stated benchmark experiments; they are not field measurements or general performance guarantees.

| Measure | Reported value | Experimental scope |
|---|---|---|
| Mean turns per DeepWeb-Bench task with CLI-Any-Browser | 52.4 | DeepWeb-Bench multi-source research tasks; CLI-Any-Browser comparison |

Sources: <https://arxiv.org/html/2610.13416v1> · Download data: [JSON](https://newsverum.com/api/v1/articles/yonsei-team-tests-unified-interface-for-web-agents/data/reported-result-2?lang=en&format=json) · [CSV](https://newsverum.com/api/v1/articles/yonsei-team-tests-unified-interface-for-web-agents/data/reported-result-2?lang=en&format=csv)

## Sources

- [arXiv preprint by Yonsei University researchers](<https://arxiv.org/html/2610.13416v1>) (Source checked: 2026-10-13; Published: 2026-10-09)

---

Cite the canonical URL above. [Editorial standards and corrections](https://newsverum.com/en/editorial)