REVIEW 14 cited by
AgentOccam: A Simple Yet Strong Baseline for LLM-Based Web Agents
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Autonomy via agents using large language models (LLMs) for personalized, standardized tasks boosts human efficiency. Automating web tasks (like booking hotels within a budget) is increasingly sought after. Fulfilling practical needs, the web agent also serves as an important proof-of-concept example for various agent grounding scenarios, with its success promising advancements in many future applications. Prior research often handcrafts web agent strategies (e.g., prompting templates, multi-agent systems, search methods, etc.) and the corresponding in-context examples, which may not generalize well across all real-world scenarios. On the other hand, there has been limited study on the misalignment between a web agent's observation/action representation and the pre-training data of the LLM it's based on. This discrepancy is especially notable when LLMs are primarily trained for language completion rather than tasks involving embodied navigation actions and symbolic web elements. Our study enhances an LLM-based web agent by simply refining its observation and action space to better align with the LLM's capabilities. This approach enables our base agent to significantly outperform previous methods on a wide variety of web tasks. Specifically, on WebArena, a benchmark featuring general-purpose web interaction tasks, our agent AgentOccam surpasses the previous state-of-the-art and concurrent work by 9.8 (+29.4%) and 5.9 (+15.8%) absolute points respectively, and boosts the success rate by 26.6 points (+161%) over similar plain web agents with its observation and action space alignment. We achieve this without using in-context examples, new agent roles, online feedback or search strategies. AgentOccam's simple design highlights LLMs' impressive zero-shot performance on web tasks, and underlines the critical role of carefully tuning observation and action spaces for LLM-based agents.
Forward citations
Cited by 14 Pith papers
-
Prismata: Confining Cross-Site Prompt Injection in Web Agents
Prismata cuts web-agent prompt-injection attack success from 85.5% to 0.7% via Biba-inspired DOM trust labeling and mechanical least-privilege confinement without site annotations.
-
Routing Is Least Learnable Where It Is Most Valuable: Bounds on Representation Routing for Web Agents
Per-task routing between text, image, and hybrid observations of a browser page does not currently beat one fixed choice, because the labels needed to learn routing exist only where the agent already succeeds; only a ...
-
From Bug Reports to Browser-Executable Procedures: An LLM-Driven Agent for Web GUI Bug Reproduction
ReBug reproduces 49.96% of 667 real web GUI bug reports by reconstructing prerequisites, executing state-aware browser actions, and validating outcomes, beating both browser-agent and script-generation baselines.
-
Agent Security Needs Redefinition through a Holistic Framework
Agent security should be redefined around four contextual authorization properties instead of the content of the action performed.
-
Persona2Web: Benchmarking Personalized Web Agents for Contextual Reasoning with User History
Persona2Web is a new open-web benchmark where agents must infer a user's preferences from synthetic browsing history to solve intentionally ambiguous queries; current best agents score 13% success.
-
Agent Learning via Early Experience
Training agents to predict consequences of their own alternative actions (implicit world modeling) or to reflect on why expert actions were better (self-reflection) consistently improves task success, OOD generalizati...
-
WebSynthesis: World-Model-Guided MCTS for Efficient WebUI-Trajectory Synthesis
A world-model-guided MCTS pipeline synthesizes 4k web navigation trajectories and yields a WebArena Pass@3 success rate of 20.15%, above OS-Genesis (18.66%) and AgentTrek (11.94%).
-
Context manipulation attacks : Web agents are susceptible to corrupted memory
Web agents with protected prompts can still be hijacked by injecting malicious steps into their stored task plans, reaching up to 63% success on privacy leaks.
-
Thinking vs. Doing: Agents that Reason by Scaling Test-Time Interaction
Scaling the number of interaction steps, trained via a curriculum over rollout horizon, improves web-agent task success and outperforms scaling per-step reasoning under fixed token budgets.
-
Better Harnesses, Smaller Models: Building 90% Cheaper Agents via Automated Harness Adaptation
Automatically adapted agent harnesses let capable SLMs recover most frontier-LLM accuracy on repetitive business tasks at a few percent of the inference cost.
-
Branch-and-Browse: Efficient and Controllable Web Exploration with Tree-Structured Reasoning and Action Memory
Branch-and-Browse, a tree-structured web agent with page-level action memory, reports 35.8% success and up to 40.4% less time than the cited Tree Search baseline on WebArena.
-
Coding Agents with Multimodal Browsing are Generalist Problem Solvers
OpenHands-Versa, a single agent with a small set of general tools, achieves state-of-the-art or competitive results on SWE-Bench Multimodal, GAIA, and The Agent Company.
-
LineRetriever: Planning-Aware Observation Reduction for Web Agents
LineRetriever uses a small LM to select relevant lines from web page observations, cutting context by up to 73% with only small success-rate drops on web agent benchmarks.
-
Reflection-Based Memory For Web navigation Agents
Reflection-Augmented Planning (ReAP) retrieves short self-reflections from past web navigation tasks and lifts held-out task success by 11 points on WebArena.
Discussion (0). Continue with ORCID to comment.