Web$hell: A Unified Interface for Agentic Web Access
WebShell unifies search, browsing, and evidence reuse so LLM agents can complete long web tasks more cheaply.
WebShell is a research interface that gives LLM agents one access model for gathering information, taking browser actions, and revisiting evidence. WebContract and WebCompiler turn live pages into addressable information and action objects stored in WebCache and handled by shared operations. Existing search, browser, and filesystem approaches mainly cover only part of that workflow or combine at higher cost. Experiments on web search, navigation, and deep-research benchmarks report strong task performance at lower cost.
- Unifies search, browsing, and filesystem-style web access in one model.
- WebCompiler turns live pages into persistent information and action objects.
- Benchmarks cover search, navigation, and deep research at lower cost.
- Argues interface design is central to capable web agents.
From the original article
1 min read at arxiv.orgLarge language model (LLM) agents are becoming direct users of the open web. As they take on increasingly complex and information-intensive tasks, effective web access is becoming a core capability for agentic problem solving. Existing approaches have advanced agentic web access through different access models, including search-augmented agents, browser-based agents, and, more recently, filesystem-based agents.
By Sunghwan Kim · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.