Web Agents Should Adopt the Plan-Then-Execute Paradigm

• URL: http://arxiv.org/abs/2605.14290v1
• Date: Thu, 14 May 2026 02:48:57 GMT
• Title: Web Agents Should Adopt the Plan-Then-Execute Paradigm
• Abstract summary: We argue that web agents should default to plan-then-execute instead of ReAct.<n>Plan-then-execute changes this boundary: untrusted data may influence values or branches inside a predefined execution graph.<n>We identify the main barrier to adopting plan-then-execute on the web.
• Score: 9.920367562132336
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: ReAct has become the default architecture across LLM agents, and many existing web agents follow this paradigm. We argue that it is the wrong default for web agents. Instead, web agents should default to plan-then-execute: commit to a task-specific program before observing runtime web content, then execute it. The reason is that web content mixes inputs from many parties. An e-commerce product page may combine a seller's listing, customer reviews and sponsored advertisements. Under ReAct, all of this content flows into the model when deciding on the next action, creating a direct path for prompt injections to steer the agent's control flow. Plan-then-execute changes this boundary: untrusted data may influence values or branches inside a predefined execution graph, but it cannot redefine the user task or cause the model to synthesize new actions at runtime. We analyze WebArena, a popular web agent benchmark, and find that all tasks are compatible with plan-then-execute, while 80% can be completed with a purely programmatic plan, without any runtime LLM subroutine. We identify the main barrier to adopting plan-then-execute on the web: For it to work well, tools must map cleanly to semantic actions, with effects known before execution, so agents have enough information to plan. The web does not naturally expose that interface. Browser tools such as click, type, and scroll have page-dependent meanings. Planning at this layer is near-sighted: the agent can only see actions on the current page, and later actions appear only after it acts. Closing this gap requires typed interfaces that turn website interactions from clicks and keystrokes to task-level operations. This is an infrastructure problem, not a modeling problem. Web tasks do not need reactivity by default; they need typed, complete, auditable website APIs.

Related papers

• GTA: Generating Long-Horizon Tasks for Web Agents at Scale [82.43869456830664]
  We introduce a scalable framework, GTA, that integrates crawling, retrieval-based seeding, in-context generation, and automated quality control.<n>We instantiate the pipeline on over 50 websites covering e-commerce, government, forums, and news, with multilingual and multi-hop coverage.<n>Our contributions are three-fold: (i) formalizing multi-hop web-agent task generation, (ii) proposing an efficient and validated pipeline for automatic data creation, and (iii) releasing a dynamic benchmark with reproducible evaluation.
  arXiv  Detail & Related papers  (2026-05-28T01:05:50Z)
• Skim: Speculative Execution for Fast and Efficient Web Agents [8.47432001892345]
  Skim is a speculative execution framework for web agents.<n>It exploits the predictable structure of purpose-built websites.<n>Skim reduces median per-task cost by 1.9x and latency by 33.4% with no accuracy loss.
  arXiv  Detail & Related papers  (2026-05-15T19:12:43Z)
• MolmoWeb: Open Visual Web Agent and Open Data for the Open Web [60.29597961827816]
  MolmoWebMix is a mixture of browser task demonstrations and web-GUI perception data.<n>MolmoWeb-8B is a family of fully open multimodal web agents.<n>We will release model checkpoints, training data, code, and a unified evaluation harness to enable open research on web agents.
  arXiv  Detail & Related papers  (2026-04-09T17:54:02Z)
• Atomicity for Agents: Exposing, Exploiting, and Mitigating TOCTOU Vulnerabilities in Browser-Use Agents [15.381306470663695]
  We present a large scale empirical study of TOCTOU vulnerabilities in browser-use agents.<n> Dynamic or adversarial web content can exploit this window to induce unintended actions.<n>We design a lightweight mitigation based on pre-execution validation.
  arXiv  Detail & Related papers  (2026-02-28T05:25:03Z)
• EmbeWebAgent: Embedding Web Agents into Any Customized UI [3.034887612600091]
  We present EmbeWebAgent, a framework for embedding agents directly into existing UIs.<n>It supports mixed-granularity actions ranging from primitives to higher-level composites.<n>Our demo shows minimal retrofitting effort and robust multi-step behaviors grounded in a live UI setting.
  arXiv  Detail & Related papers  (2026-02-16T15:59:56Z)
• SPILLage: Agentic Oversharing on the Web [13.930845226612767]
  LLMs are beginning to automate user's tasks across the open web, often with access to user resources such as emails and calendars.<n>We ask how do web agents handle user resources when accomplishing tasks on their behalf across live websites?<n>In this paper, we formalize Natural Agentic Oversharing -- the unintentional disclosure of task-irrelevant user information through an agent trace of actions on the web.
  arXiv  Detail & Related papers  (2026-02-13T23:02:50Z)
• WALT: Web Agents that Learn Tools [66.73502484310121]
  WALT is a framework that reverse-engineers latent website functionality into reusable invocable tools.<n>Rather than hypothesizing ad-hoc skills, WALT exposes robust implementations of automations already designed into websites.<n>On VisualWebArena and WebArena, WALT achieves higher success with fewer steps and less LLM-dependent reasoning.
  arXiv  Detail & Related papers  (2025-10-01T23:41:47Z)
• Beyond Browsing: API-Based Web Agents [58.39129004543844]
  API-Based Agents outperform web Browsing Agents in experiments on WebArena.<n>Hybrid Agents out-perform both others nearly uniformly across tasks.<n>Results strongly suggest that when APIs are available, they present an attractive alternative to relying on web browsing alone.
  arXiv  Detail & Related papers  (2024-10-21T19:46:06Z)
• AgentOccam: A Simple Yet Strong Baseline for LLM-Based Web Agents [52.13695464678006]
  This study enhances an LLM-based web agent by simply refining its observation and action space.<n>AgentOccam surpasses the previous state-of-the-art and concurrent work by 9.8 (+29.4%) and 5.9 (+15.8%) absolute points respectively.
  arXiv  Detail & Related papers  (2024-10-17T17:50:38Z)
• CAAP: Context-Aware Action Planning Prompting to Solve Computer Tasks with Front-End UI Only [21.054681757006385]
  We propose an agent that perceives its environment solely through screenshot images.<n>By leveraging the reasoning capability of the Large Language Models, we eliminate the need for large-scale human demonstration data.<n>Agent achieves an average success rate of 94.5% on MiniWoB++ and an average task score of 62.3 on WebShop.
  arXiv  Detail & Related papers  (2024-06-11T05:21:20Z)
• A Real-World WebAgent with Planning, Long Context Understanding, and Program Synthesis [69.15016747150868]
  We introduce WebAgent, an agent that learns from self-experience to complete tasks on real websites. WebAgent plans ahead by decomposing instructions into canonical sub-instructions, summarizes long HTML documents into task-relevant snippets, and acts on websites. We empirically demonstrate that our modular recipe improves the success on real websites by over 50%, and that HTML-T5 is the best model to solve various HTML understanding tasks.
  arXiv  Detail & Related papers  (2023-07-24T14:56:30Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.