On 108 long-horizon real-world computer workflows, frontier agents complete at most 20.6% of tasks and fail mainly by losing hidden state, not by basic GUI control.
hub Canonical reference
Webvoyager: Building an end-to-end web agent with large multimodal models
Canonical reference. 80% of citing Pith papers cite this work as background.
18
Pith papers citing it
34
external citations · OpenAlex
Background
80% of classified citations
hub tools
citation-role summary
background 4
dataset 1
citation-polarity summary
representative citing papers
LiveBrowseComp shows search agents rely on intrinsic knowledge on standard benchmarks, with scores dropping 25-40 points and closed-book accuracy below 2% on questions about facts from the prior 90 days.
ShopGym introduces ShopArena to convert live storefronts into self-contained sandbox shops and ShopGuru to synthesize 224 benchmark tasks, with validation showing structural preservation and positive correlation of agent performance between synthetic and live shops.
MMSkills packages multimodal procedural knowledge into state-conditioned skills with text, state cards, and multi-view keyframes, generated from public trajectories via an agentic process and used at inference via branch-loaded inspection to improve visual agents on GUI and game benchmarks.
FlowEval evaluates generated UIs by measuring how closely their navigation flows match real websites via reference-based similarity metrics and shows strong correlation with human expert judgments.
Kamera stores a low-rank patch with each position-free KV chunk to restore cross-chunk conditioning lost in naive reuse, enabling cheap reordering, sliding windows, and recall across attention mechanisms.
MemVenom poisons multimodal memories in web agents via a two-stage trigger-conditioned retrieval and post-retrieval induction attack, achieving up to 99.15% success on GPT-5-family agents while preserving benign performance.
New benchmark Scammer4U finds 54-93% critical PII leakage from frontier web agents on scam sites versus 0% on benign twins, plus a 30-point gap between verbalized suspicion and actual submission.
PageGuide is a browser extension that grounds LLM responses in webpage DOM elements via visual overlays for Find, Guide, and Hide modes, reporting performance gains over unaided browsing in a 94-user study.
Mango raises web agent success rates to 63.6% on WebVoyager and 52.5% on WebWalkerQA by bandit-based starting-point selection and memory, beating baselines by 7.3% and 26.8%.
TEC is a new public dataset of detailed human trial-and-error trajectories and reflections on web tasks, with humans showing substantially higher accuracy than LLMs.
DynaWeb introduces a model-based RL framework that trains web agents via imagined rollouts in a learned web world model interleaved with real expert trajectories, yielding consistent gains on WebArena and WebVoyager benchmarks.
This survey categorizes agentic environments for LLMs by eight attributes and domains, introduces symbolic and neural synthesis paradigms with evaluation, and outlines four agent evolution pathways plus three environment evolution paradigms.
Cognitive Kernel-Pro provides an open-source agent framework with curated training data across web, file, code, and reasoning domains plus test-time reflection and voting, achieving SOTA results on GAIA among free agents.
A survey of 87 agents for computer use and 33 datasets that introduces a three-dimensional taxonomy across domain, interaction, and agent perspectives and identifies six research gaps.
WebUncertainty improves web agent performance on benchmarks by adaptively selecting planning modes based on task uncertainty and using confidence-induced action uncertainty in MCTS to quantify aleatoric and epistemic uncertainty for better decisions.
StepGuard framework with DDPO and CANR claims SOTA navigation and answer accuracy on web benchmarks by switching policies and triggering reflection on low-confidence steps.
citing papers explorer
-
OSWorld 2.0: Benchmarking Computer Use Agents on Long-Horizon Real-World Tasks
On 108 long-horizon real-world computer workflows, frontier agents complete at most 20.6% of tasks and fail mainly by losing hidden state, not by basic GUI control.
-
LiveBrowseComp: Are Search Agents Searching, or Just Verifying What They Already Know?
LiveBrowseComp shows search agents rely on intrinsic knowledge on standard benchmarks, with scores dropping 25-40 points and closed-book accuracy below 2% on questions about facts from the prior 90 days.
-
ShopGym: An Integrated Framework for Realistic Simulation and Scalable Benchmarking of E-Commerce Web Agents
ShopGym introduces ShopArena to convert live storefronts into self-contained sandbox shops and ShopGuru to synthesize 224 benchmark tasks, with validation showing structural preservation and positive correlation of agent performance between synthetic and live shops.
-
MMSkills: Towards Multimodal Skills for General Visual Agents
MMSkills packages multimodal procedural knowledge into state-conditioned skills with text, state cards, and multi-view keyframes, generated from public trajectories via an agentic process and used at inference via branch-loaded inspection to improve visual agents on GUI and game benchmarks.
-
FlowEval: Reference-based Evaluation of Generated User Interfaces
FlowEval evaluates generated UIs by measuring how closely their navigation flows match real websites via reference-based similarity metrics and shows strong correlation with human expert judgments.
-
Kamera: Unified Position-Invariant Multimodal KV Cache for Training-Free Reuse
Kamera stores a low-rank patch with each position-free KV chunk to restore cross-chunk conditioning lost in naive reuse, enabling cheap reordering, sliding windows, and recall across attention mechanisms.
-
MemVenom: Triggered Poisoning of Multimodal Memories in Web Agents
MemVenom poisons multimodal memories in web agents via a two-stage trigger-conditioned retrieval and post-retrieval induction attack, achieving up to 99.15% success on GPT-5-family agents while preserving benign performance.
-
"I Strongly Suspect This Website Is a Scam": Benchmarking PII Leakage and Detection without Defense in Autonomous Web Agents
New benchmark Scammer4U finds 54-93% critical PII leakage from frontier web agents on scam sites versus 0% on benign twins, plus a 30-point gap between verbalized suspicion and actual submission.
-
PageGuide: Browser extension to assist users in navigating a webpage and locating information
PageGuide is a browser extension that grounds LLM responses in webpage DOM elements via visual overlays for Find, Guide, and Hide modes, reporting performance gains over unaided browsing in a 94-user study.
-
Mango: Multi-Agent Web Navigation via Global-View Optimization
Mango raises web agent success rates to 63.6% on WebVoyager and 52.5% on WebWalkerQA by bandit-based starting-point selection and memory, beating baselines by 7.3% and 26.8%.
-
TEC: A Collection of Human Trial-and-error Trajectories for Problem Solving
TEC is a new public dataset of detailed human trial-and-error trajectories and reflections on web tasks, with humans showing substantially higher accuracy than LLMs.
-
DynaWeb: Model-Based Reinforcement Learning of Web Agents
DynaWeb introduces a model-based RL framework that trains web agents via imagined rollouts in a learned web world model interleaved with real expert trajectories, yielding consistent gains on WebArena and WebVoyager benchmarks.
-
Agentic Environment Engineering for Large Language Models: A Survey of Environment Modeling, Synthesis, Evaluation, and Application
This survey categorizes agentic environments for LLMs by eight attributes and domains, introduces symbolic and neural synthesis paradigms with evaluation, and outlines four agent evolution pathways plus three environment evolution paradigms.
-
Cognitive Kernel-Pro: A Framework for Deep Research Agents and Agent Foundation Models Training
Cognitive Kernel-Pro provides an open-source agent framework with curated training data across web, file, code, and reasoning domains plus test-time reflection and voting, achieving SOTA results on GAIA among free agents.
-
A Comprehensive Survey of Agents for Computer Use: Foundations, Challenges, and Future Directions
A survey of 87 agents for computer use and 33 datasets that introduces a three-dimensional taxonomy across domain, interaction, and agent perspectives and identifies six research gaps.
-
WebUncertainty: Dual-Level Uncertainty Driven Planning and Reasoning For Autonomous Web Agent
WebUncertainty improves web agent performance on benchmarks by adaptively selecting planning modes based on task uncertainty and using confidence-induced action uncertainty in MCTS to quantify aleatoric and epistemic uncertainty for better decisions.
-
StepGuard: Guarding Web Navigation via Single-Step Calibration
StepGuard framework with DDPO and CANR claims SOTA navigation and answer accuracy on web benchmarks by switching policies and triggering reflection on low-confidence steps.
- Designing for Doubt: The Case for Informed Abstention in Autonomous Agents