AI Agent Evaluation

As reinforcement learning continues to scale the training of large language model-based agents, reliably verifying agent behaviors in complex environments has become increasingly challenging. Existing approaches rely on rule-based verifiers or LLM-as-a-Judge models, which struggle to generalize beyond narrow domains. Agent-as-a-Judge addresses this limitation by actively interacting with environments and tools to acquire verifiable evidence, yet its capabilities remain underexplored. We introduce a benchmark AJ-Bench to systematically evaluate Agent-as-a-Judge across three domains-search, data systems, and graphical user interfaces-comprising 155 tasks and 516 annotated trajectories. The benchmark comprehensively assesses judge agents' abilities in information acquisition, state verification, and process verification. Experiments demonstrate consistent performance gains over LLM-as-a-Judge baselines, while also revealing substantial open challenges in agent-based verification. Our data and code are available at https://aj-bench.github.io/.

AI agents are reshaping the workspace, leading to drastic change of how humans work. Despite the considerable potential of human-agent collaboration both in preserving human agency and generating economic value, this paradigm remains largely absent from occupational task evaluation, hindered by the difficulty of gathering real human data and accounting for inter-human variability. We introduce CollabSkill, a framework for evaluating human-agent collaboration on real-world occupational tasks. CollabSkill pairs real human workers with AI agents on tasks matched to their occupational background, collecting data that capture the complexity of economically valuable tasks and the usage patterns of real workers. To account for inter-human variability, CollabSkill employs a Bayesian skill rating system to disentangle and quantify the skill contributions of both humans and AI agents. Drawing on over 1,500 prompts from 386 working sessions contributed by 93 human workers, our analysis yields insights on two fronts: on the agent side, rankings on CollabSkill diverge meaningfully from those of existing fully autonomous benchmarks where Codex leads, with Claude Code ranking first; on the human side, CollabSkill reveals that practical experience emerges as the primary driver of collaboration skill, with hands-on collaboration meaningfully shifting workers' AI literacy. Together, we hope CollabSkill enables the community to invest in systematic evaluation of human-agent collaboration and spurs development efforts aimed at building AI agents that genuinely augment human workers.

Recent advances in large language models have highlighted their potential to automate computational research, particularly reproducing experimental results. However, existing approaches still use fixed sequential agent pipelines with weak global coordination, which limits their robustness and overall performance. In this work, we propose Hierarchical Research Agent System (HiRAS), a hierarchical multi-agent framework for end-to-end experiment reproduction that employs supervisory manager agents to coordinate specialised agents across fine-grained stages. We also identify limitations in the reference-free evaluation of the Paper2Code benchmark and introduce Paper2Code-Extra (P2C-Ex), a refined protocol that incorporates repository-level information and better aligns with the original reference-based metric. We conduct extensive evaluation, validating the effectiveness and robustness of our proposed methods, and observing improvements, including >10% relative performance gain beyond the previous state-of-the-art using open-source backbone models and significantly reduced hallucination in evaluation. Our work is available on GitHub: https://github.com/KOU-199024/HiRAS.

Agentic AI systems plan, use tools, maintain state, and act across multi-step workflows with external effects, meaning trustworthy deployment can no longer be judged by task completion alone. The current literature remains fragmented across benchmark-centered evaluation, standards-based governance, orchestration architectures, and runtime assurance mechanisms. This paper contributes a bounded evidence synthesis across a manually coded corpus of twenty-four recent sources. The core finding is a governance-to-action closure gap: evaluation tells us whether outcomes were good, governance defines what should be allowed, but neither identifies where obligations bind to concrete actions or how compliance can later be proven. To close that gap, the paper introduces three linked artifacts: (1) a four-layer framework spanning evaluation, governance, orchestration, and assurance; (2) an ODTA runtime-placement test based on observability, decidability, timeliness, and attestability; and (3) a minimum action-evidence bundle for state-changing actions. Across sources, evaluation papers identify safety, robustness, and trajectory-level measurement as open gaps; governance frameworks define obligations but omit execution-time control logic; orchestration research positions the control plane as the locus of policy mediation, identity, and telemetry; runtime-governance work shows path-dependent behavior cannot be governed through prompts or static permissions alone; and action-safety studies show text alignment does not reliably transfer to tool actions. A worked enterprise procurement-agent scenario illustrates how these artifacts consolidate existing evidence without introducing new experimental data.

Instruction granularity is an important yet poorly controlled variable in language-guided embodied AI. Existing benchmarks typically pair each task with a single static instruction, making it difficult to study how agent behavior changes when the same task is described at different levels of detail. We introduce Mini-BEHAVIOR-Gran, a new benchmark for controlled studies of instruction granularity that extends Mini-BEHAVIOR with multiple instruction variants per task, ranging from high-level goal descriptions to step-by-step guidance. Using this benchmark, we compare four candidate metrics for cross-task granularity quantification: token count, entity count, action-verb count, and planning-width, and find that width correlates most consistently with agent performance. Using width to organize training and evaluation further reveals a non-monotonic U-shaped relationship between instruction granularity and performance, with peaks at both fine and coarse extremes. Further analysis suggests that the coarse-granularity performance rebound is associated with shallow grounding, where agents learn vision-dominant policies.

Agentic AI systems are rapidly advancing toward real-world applications, yet their readiness in complex and personalized environments remains insufficiently characterized. To address this gap, we introduce PersonalHomeBench, a benchmark for evaluating foundation models as agentic assistants in personalized smart home environments. The benchmark is constructed through an iterative process that progressively builds rich household states, which are then used to generate personalized, context-dependent tasks. To support realistic agent-environment interaction, we provide PersonalHomeTools, a comprehensive toolbox enabling household information retrieval, appliance control, and situational understanding. PersonalHomeBench evaluates both reactive and proactive agentic abilities under unimodal and multimodal observations. Thorough experimentation reveals a systematic performance reduction as task complexity increases, with pronounced failures in counterfactual reasoning and under partial observability, where effective tool-based information gathering is required. These results position PersonalHomeBench as a rigorous evaluation platform for analyzing the robustness and limitations of personalized agentic reasoning and planning.

Recent advances in agentic frameworks have enabled AI agents to perform complex reasoning and decision-making. However, evidence comparing their reasoning performance, efficiency, and practical suitability remains limited. To address this gap, we empirically evaluate 22 widely used agentic frameworks across three reasoning benchmarks: BBH, GSM8K, and ARC. The frameworks were selected from 1,200 GitHub repositories collected between January 2023 and July 2025 and organized into a taxonomy based on architectural design. We evaluated them under a unified setting, measuring reasoning accuracy, execution time, computational cost, and cross-benchmark consistency. Our results show that 19 of the 22 frameworks completed all three benchmarks. Among these, 12 showed stable performance, with mean accuracy of 74.6-75.9%, execution time of 4-6 seconds per task, and cost of 0.14-0.18 cents per task. Poorer results were mainly caused by orchestration problems rather than reasoning limits. For example, Camel failed to complete BBH after 11 days because of uncontrolled context growth, while Upsonic consumed USD 1,434 in one day because repeated extraction failures triggered costly retries. AutoGen and Mastra also exhausted API quotas through iterative interactions that increased prompt length without improving results. We also found a sharp drop in mathematical reasoning. Mean accuracy on GSM8K was 44.35%, compared with 89.80% on BBH and 89.56% on ARC. Overall, this study provides the first large-scale empirical comparison of agentic frameworks for reasoning-intensive software engineering tasks and shows that framework selection should prioritize orchestration quality, especially memory control, failure handling, and cost management.

Much of the focus in RL today is on evaluation design: building meaningful evals that serve simultaneously as benchmarks and as well-defined reward signals for post-training. Yet, many real-world tasks are governed by subjective, procedural, and domain-specific requirements that are difficult to encode as exact-match targets or open-ended preference judgments frequently used in RL pipelines today. In this work, we present AsymmetryZero, a framework for operationalizing human expert preferences as semantic evals. AsymmetryZero represents each task as a stable evaluation contract that makes grading criteria explicit: what is being graded, how each criterion is judged, and how criterion-level decisions are aggregated into a task outcome. The same contract can be executed using Inspect for model-only evaluations, as well as the Harbor Framework for agentic evaluations, enabling comparable scores and shared audit artifacts across both settings. We argue that the central challenge in post-training today is the faithful encoding of expert requirements into the evaluation itself. To that end, we present a study using Harbor that holds task contracts fixed and compares a five-model frontier jury against a five-model compact jury across four frontier-class solvers (Claude Opus 4.6, GPT-5.4, Grok-4.20, Gemini-3.1-Pro). We find that criterion-level frontier-vs-compact agreement ranges from $75.9\%$ to $89.6\%$ (strict common-subset agreement: $77.8\%$ to $92.1\%$), while compact juries exhibit substantially higher internal dissent (3--2 split rate $28.7\%$--$32.4\%$) than frontier juries ($6.1\%$--$11.5\%$). Verifier traces further show that compact juries reduce per-criterion judging cost to roughly $4.2\%$--$5.6\%$ of frontier and latency to roughly $21.7\%$--$27.1\%$, even as aggregated task-level outcomes often remain comparatively stable.

Modern information access ecosystems consist of mixtures of systems, such as retrieval systems and large language models, and increasingly rely on marketplaces to mediate access to models, tools, and data, making competition between systems inherent to deployment. In such settings, outcomes are shaped not only by benchmark quality but also by competitive pressure, including user switching, routing decisions, and operational constraints. Yet evaluation is still largely conducted on static benchmarks with accuracy-focused measures that assume systems operate in isolation. This mismatch makes it difficult to predict post-deployment success and obscures competitive effects such as early-adoption advantages and market dominance. We introduce Marketplace Evaluation, a simulation-based paradigm that evaluates information access systems as participants in a competitive marketplace. By simulating repeated interactions and evolving user and agent preferences, the framework enables longitudinal evaluation and marketplace-level metrics, such as retention and market share, that complement and can extend beyond traditional accuracy-based metrics. We formalize the framework and outline a research agenda, motivated by business and economics, around marketplace simulation, metrics, optimization, and adoption in evaluation campaigns like TREC.

Current LLM agent benchmarks, which predominantly focus on binary pass/fail tasks such as code generation or search-based question answering, often neglect the value of real-world engineering that is often captured through the iterative optimization of feasible designs. To this end, we introduce Frontier-Eng, a human-verified benchmark for generative optimization -- an iterative propose-execute-evaluate loop in which an agent generates candidate artifacts, receives executable verifier feedback, and revises them under a fixed interaction budget -- spanning $47$ tasks across five broad engineering categories. Unlike previous suites, Frontier-Eng tasks are grounded in industrial-grade simulators and verifiers that provide continuous reward signals and enforce hard feasibility constraints under constrained budgets. We evaluate eight frontier language models using representative search frameworks, finding that while GPT 5.4 achieves the most robust performance, the benchmark remains challenging for all models. Our analysis suggests a dual power-law decay in improvement frequency ($\sim$ 1/iteration) and magnitude ($\sim$ 1/improvement count). We further show that although width improves parallelism and diversity, depth remains crucial for hard-won improvements under a fixed budget. Frontier-Eng establishes a new standard for assessing the capacity of AI agents to integrate domain knowledge with executable feedback to solve complex, open-ended engineering problems.

The rapid deployment of AI agents in commercial settings has outpaced the development of evaluation methodologies that reflect production realities. Existing benchmarks measure agent capabilities through retrospectively curated tasks with well-specified requirements and deterministic metrics -- conditions that diverge fundamentally from production environments where requirements contain implicit constraints, inputs are heterogeneous multi-modal documents with information fragmented across sources, tasks demand undeclared domain expertise, outputs are long-horizon professional deliverables, and success is judged by domain experts whose standards evolve over time. We present AlphaEval, a production-grounded benchmark of 94 tasks sourced from seven companies deploying AI agents in their core business, spanning six O*NET (Occupational Information Network) domains. Unlike model-centric benchmarks, AlphaEval evaluates complete agent products -- Claude Code, Codex, etc. -- as commercial systems, capturing performance variations invisible to model-level evaluation. Our evaluation framework covers multiple paradigms (LLM-as-a-Judge, reference-driven metrics, formal verification, rubric-based assessment, automated UI testing, etc.), with individual domains composing multiple paradigms. Beyond the benchmark itself, we contribute a requirement-to-benchmark construction framework -- a systematic methodology that transforms authentic production requirements into executable evaluation tasks in minimal time. This framework standardizes the entire pipeline from requirement to evaluation, providing a reproducible, modular process that any organization can adopt to construct production-grounded benchmarks for their own domains.

We present SIR-Bench, a benchmark of 794 test cases for evaluating autonomous security incident response agents that distinguishes genuine forensic investigation from alert parroting. Derived from 129 anonymized incident patterns with expert-validated ground truth, SIR-Bench measures not only whether agents reach correct triage decisions, but whether they discover novel evidence through active investigation. To construct SIR-Bench, we develop Once Upon A Threat (OUAT), a framework that replays real incident patterns in controlled cloud environments, producing authentic telemetry with measurable investigation outcomes. Our evaluation methodology introduces three complementary metrics: triage accuracy (M1), novel finding discovery (M2), and tool usage appropriateness (M3), assessed through an adversarial LLM-as-Judge that inverts the burden of proof -- requiring concrete forensic evidence to credit investigations. Evaluating our SIR agent on the benchmark demonstrates 97.1% true positive (TP) detection, 73.4% false positive (FP) rejection, and 5.67 novel key findings per case, establishing a baseline against which future investigation agents can be measured.

GUI agents drive applications through their visual interfaces instead of programmatic APIs, interacting with arbitrary software via taps, swipes, and keystrokes, reaching a long tail of applications that CLI-based agents cannot. Yet progress in this area is bottlenecked less by modeling capacity than by the absence of a coherent full-stack infrastructure: online RL training suffers from environment instability and closed pipelines, evaluation protocols drift silently across works, and trained agents rarely reach real users on real devices. We present ClawGUI, an open-source framework addressing these three gaps within a single harness. ClawGUI-RL provides the first open-source GUI agent RL infrastructure with validated support for both parallel virtual environments and real physical devices, integrating GiGPO with a Process Reward Model for dense step-level supervision. ClawGUI-Eval enforces a fully standardized evaluation pipeline across 6 benchmarks and 11+ models, achieving 95.8% reproduction against official baselines. ClawGUI-Agent brings trained agents to Android, HarmonyOS, and iOS through 12+ chat platforms with hybrid CLI-GUI control and persistent personalized memory. Trained end to end within this pipeline, ClawGUI-2B achieves 17.1% Success Rate on MobileWorld GUI-Only, outperforming the same-scale MAI-UI-2B baseline by 6.0%.

Organizational knowledge used by AI agents typically lacks epistemic structure: retrieval systems surface semantically relevant content without distinguishing binding decisions from abandoned hypotheses, contested claims from settled ones, or known facts from unresolved questions. We argue that the ceiling on organizational AI is not retrieval fidelity but epistemic fidelity--the system's ability to represent commitment strength, contradiction status, and organizational ignorance as computable properties. We present OIDA, a framework that structures organizational knowledge as typed Knowledge Objects carrying epistemic class, importance scores with class-specific decay, and signed contradiction edges. The Knowledge Gravity Engine maintains scores deterministically with proved convergence guarantees (sufficient condition: max degree $< 7$; empirically robust to degree 43). OIDA introduces QUESTION-as-modeled-ignorance: a primitive with inverse decay that surfaces what an organization does not know with increasing urgency--a mechanism absent from all surveyed systems. We describe the Epistemic Quality Score (EQS), a five-component evaluation methodology with explicit circularity analysis. In a controlled comparison ($n{=}10$ response pairs), OIDA's RAG condition (3,868 tokens) achieves EQS 0.530 vs.\ 0.848 for a full-context baseline (108,687 tokens); the $28.1\times$ token budget difference is the primary confound. The QUESTION mechanism is statistically validated (Fisher $p{=}0.0325$, OR$=21.0$). The formal properties are established; the decisive ablation at equal token budget (E4) is pre-registered and not yet run.

Existing AI benchmarks lack the fidelity to assess economically meaningful progress on professional workflows. To evaluate frontier AI agents in a high-value, labor-intensive profession, we introduce BankerToolBench (BTB): an open-source benchmark of end-to-end analytical workflows routinely performed by junior investment bankers. To develop an ecologically valid benchmark grounded in representative work environments, we collaborated with 502 investment bankers from leading firms. BTB requires agents to execute senior banker requests by navigating data rooms, using industry tools (market data platform, SEC filings database), and generating multi-file deliverables--including Excel financial models, PowerPoint pitch decks, and PDF/Word reports. Completing a BTB task takes bankers up to 21 hours, underscoring the economic stakes of successfully delegating this work to AI. BTB enables automated evaluation of any LLM or agent, scoring deliverables against 100+ rubric criteria defined by veteran investment bankers to capture stakeholder utility. Testing 9 frontier models, we find that even the best-performing model (GPT-5.4) fails nearly half of the rubric criteria and bankers rate 0% of its outputs as client-ready. Our failure analysis reveals key obstacles (such as breakdowns in cross-artifact consistency) and improvement directions for agentic AI in high-stakes professional workflows.

Existing browser agent benchmarks face a fundamental trilemma: real-website benchmarks lack reproducibility due to content drift, controlled environments sacrifice realism by omitting real-web noise, and both require costly manual curation that limits scalability. We present WebForge, the first fully automated framework that resolves this trilemma through a four-agent pipeline -- Plan, Generate, Refine, and Validate -- that produces interactive, self-contained web environments end-to-end without human annotation. A seven-dimensional difficulty control framework structures task design along navigation depth, visual complexity, reasoning difficulty, and more, enabling systematic capability profiling beyond single aggregate scores. Using WebForge, we construct WebForge-Bench, a benchmark of 934 tasks spanning 7 domains and 3 difficulty levels. Multi-model experiments show that difficulty stratification effectively differentiates model capabilities, while cross-domain analysis exposes capability biases invisible to aggregate metrics. Together, these results confirm that multi-dimensional evaluation reveals distinct capability profiles that a single aggregate score cannot capture. Code and benchmark are publicly available at https://github.com/yuandaxia2001/WebForge.

The rapid deployment of LLM-based autonomous agents has introduced safety risks that extend far beyond traditional LLM concerns, prompting a proliferation of safety benchmarks since late 2023. However, these benchmarks have developed independently, with inconsistent threat models, incompatible metrics, and overlapping yet incomplete risk coverage. We present the first systematic analysis dedicated to agent safety benchmarks as evaluation instruments. We catalog 40 behavioral agent-safety benchmarks (2023-2026), plus 5 adjacent evaluator, defense, and dataset artifacts, propose a six-axis taxonomy of benchmark evaluation methodology, and apply it across the corpus to characterize how methodological choices shape safety conclusions. A coverage matrix reveals broad risk coverage but limited methodological convergence, while the taxonomy analysis shows a behavioral-benchmark core concentrated in sandboxed, constrained, and often safety-only evaluation. Across the landscape, we find that benchmark choice can yield contradictory safety conclusions, coverage counts often overstate evaluation depth, environment fidelity systematically shapes reported safety, the field disproportionately tests externally imposed rather than agent-internal risks, metric fragmentation limits comparison, and robustness remains effectively unbenchmarked. We ground these claims with a cross-benchmark consistency check, with 95% confidence intervals and Kendall's W concordance analysis, finding no evidence of ranking concordance across evaluation dimensions (W = 0.10, p = 0.94). We release structured metadata, full taxonomy codings, risk annotations, and all experimental artifacts, and propose minimum reporting standards for future benchmarks.

AI agents hold growing promise for accelerating scientific discovery; yet, a lack of frontier evaluations hinders adoption into real workflows. Expert-written benchmarks have proven effective at measuring AI reasoning, but most at this stage have become saturated and only measure performance on constrained outputs. To help address this gap, we introduce COMPOSITE-STEM, a benchmark of 70 expert-written tasks in physics, biology, chemistry, and mathematics, curated by doctoral-level researchers. Our benchmark combines exact-match grading and criterion-based rubrics with an LLM-as-a-jury grading protocol, allowing more flexible assessment of scientifically meaningful outputs. Using an adapted multimodal Terminus-2 agent harness within the Harbor agentic evaluation framework, we evaluate four frontier models. The top-performing model achieves 21%, demonstrating that COMPOSITE-STEM captures capabilities beyond current agent reach. All tasks are open-sourced with contributor permission to support reproducibility and to promote additional research towards AI's acceleration of scientific progress in these domains.

AI agents may be able to assist with emails and documents, but can they reliably complete everyday online workflows on real websites? Everyday online tasks offer a realistic yet unsolved testbed for evaluating the next generation of AI agents. To this end, we introduce ClawBench, an evaluation framework comprising 153 everyday online tasks that people need to accomplish regularly in their lives and work, spanning 144 platforms across 15 categories, from completing purchases and booking appointments to submitting job applications. These tasks require capabilities beyond existing benchmarks, such as obtaining relevant information from user-provided documents, navigating multi-step workflows across diverse platforms, and write-heavy operations like filling in many detailed forms correctly. Unlike existing benchmarks that evaluate agents in offline sandboxes with static pages, ClawBench operates on production websites, preserving the full complexity, dynamic nature, and interaction challenges of real-world web environments. An interception layer captures and blocks the final submission request, ensuring safe evaluation without real-world side effects. Our evaluations of 8 frontier models show that both proprietary and open-source models complete only a small portion of these tasks. For example, Claude Sonnet 4.6 achieves only 33.3%, which exposes gaps in current AI agents. Progress on ClawBench brings us closer to AI agents that can function as general-purpose assistants.

LLM-driven GUI agents are increasingly used in production systems to automate workflows and simulate users for evaluation and optimization. Yet most GUI-agent evaluations emphasize task success and provide limited evidence on whether agents interact in human-like ways. We present a trace-level evaluation framework that compares human and agent behavior across (i) task outcome and effort, (ii) query formulation, and (iii) navigation across interface states. We instantiate the framework in a controlled study in a production audio-streaming search application, where 39 participants and a state-of-the-art GUI agent perform ten multi-hop search tasks. The agent achieves task success comparable to participants and generates broadly aligned queries, but follows systematically different navigation strategies: participants exhibit content-centric, exploratory behavior, while the agent is more search-centric and low-branching. These results show that outcome and query alignment do not imply behavioral alignment, motivating trace-level diagnostics when deploying GUI agents as proxies for users in production search systems.

History textbooks often contain implicit biases, nationalist framing, and selective omissions that are difficult to audit at scale. We propose an agentic evaluation architecture comprising a multimodal screening agent, a heterogeneous jury of five evaluative agents, and a meta-agent for verdict synthesis and human escalation. A central contribution is a Source Attribution Protocol that distinguishes textbook narrative from quoted historical sources, preventing the misattribution that causes systematic false positives in single-model evaluators. In an empirical study on Romanian upper-secondary history textbooks, 83.3% of 270 screened excerpts were classified as pedagogically acceptable (mean severity 2.9/7), versus 5.4/7 under a zero-shot baseline, demonstrating that agentic deliberation mitigates over-penalization. In a blind human evaluation (18 evaluators, 54 comparisons), the Independent Deliberation configuration was preferred in 64.8% of cases over both a heuristic variant and the zero-shot baseline. At approximately $2 per textbook, these results position agentic evaluation architectures as economically viable decision-support tools for educational governance.

Towards an embodied generalist for real-world interaction, Multimodal Large Language Model (MLLM) agents still suffer from challenging latency, sparse feedback, and irreversible mistakes. Video games offer an ideal testbed with rich visual observations and closed-loop interaction, demanding fine-grained perception, long-horizon planning, and precise control. However, systematically evaluating these capabilities is currently hindered by heterogeneous action interfaces and heuristic verification. To this end, we introduce GameWorld, a benchmark designed for standardized and verifiable evaluation of MLLMs as generalist game agents in browser environments. Two game agent interfaces are studied: (i) computer-use agents that directly emit keyboard and mouse controls, and (ii) generalist multimodal agents that act in a semantic action space via deterministic Semantic Action Parsing. GameWorld contains 34 diverse games and 170 tasks, each paired with state-verifiable metrics for outcome-based evaluation. The results across 18 model-interface pairs suggest that even the best performing agent is far from achieving human capabilities on video games. Extensive experiments of repeated full-benchmark reruns demonstrate the robustness of the benchmark, while further studies on real-time interaction, context-memory sensitivity, and action validity expose more challenges ahead for game agents. Together, by offering a standardized, verifiable, and reproducible evaluation framework, GameWorld lays a robust foundation for advancing research on multimodal game agents and beyond. The project page is at https://gameworld-bench.github.io.

Web agents act through long interaction sequences, yet existing benchmarks evaluate only terminal success, discarding all process information and offering little guidance on improvement. In this work, we conduct a process-level analysis of web agents. We introduce WebStep, a benchmark of 1,800 task instances with controlled difficulty and automatic semantic state tracking. Each website exposes a deterministic semantic MDP alongside the GUI: the agent operates on the interface, while the environment records high-level states and transitions in the background, enabling fine-grained analysis without manual annotation. Based on the semantic trajectory, we first show that process metrics reveal differences invisible to outcome evaluation: three agents whose success rates cluster within 34-37% diverge in exploration reach versus execution accuracy. Then, decomposing by skill characterizes the nature of these differences, exposing opposite per-skill rankings hidden within the same website: e.g., on Q&A, Claude CUA outperforms OpenAI CUA by 30% on navigation actions yet underperforms it by 6.7% on inspection, pinpointing a concrete skill to improve even within a domain. Bifurcation analysis further localizes the decisive error that loses the task and shows that this error is agent-specific rather than shared. Finally, these differences widen as tasks grow harder: success rate is similar on easy tasks but separates sharply as exploration becomes more demanding. Our process-level analysis opens a new avenue in web agent evaluation, providing fine-grained and actionable insight into where and how each agent should be improved. Project page: https://jiwanchung.github.io/webstep

We present ATANT (Automated Test for Acceptance of Narrative Truth), an open evaluation framework for measuring continuity in AI systems: the ability to persist, update, disambiguate, and reconstruct meaningful context across time. While the AI industry has produced memory components (RAG pipelines, vector databases, long context windows, profile layers), no published framework formally defines or measures whether these components produce genuine continuity. We define continuity as a system property with 7 required properties, introduce a 10-checkpoint evaluation methodology that operates without an LLM in the evaluation loop, and present a narrative test corpus of 250 stories comprising 1,835 verification questions across 6 life domains. We evaluate a reference implementation across 5 test suite iterations, progressing from 58% (legacy architecture) to 100% in isolated mode (250 stories) and 100% in 50-story cumulative mode, with 96% at 250-story cumulative scale. The cumulative result is the primary measure: when 250 distinct life narratives coexist in the same database, the system must retrieve the correct fact for the correct context without cross-contamination. ATANT is system-agnostic, model-independent, and designed as a sequenced methodology for building and validating continuity systems. The framework specification, example stories, and evaluation protocol are available at https://github.com/Kenotic-Labs/ATANT. The full 250-story corpus will be released incrementally.

Large language models are increasingly deployed as autonomous agents for multi-step workflows in real-world software environments. However, existing agent benchmarks are limited by trajectory-opaque grading, underspecified safety and robustness evaluation, and narrow coverage of modalities and interaction paradigms. We introduce Claw-Eval, an end-to-end evaluation suite addressing these gaps with 300 human-verified tasks spanning 9 categories across three groups: general service orchestration, multimodal perception and interaction, and multi-turn professional dialogue. To enable trajectory-aware grading, each run is recorded through three independent evidence channels: execution traces, audit logs, and environment snapshots, yielding 2,159 fine-grained rubric items. The scoring protocol evaluates Completion, Safety, and Robustness, with Average Score, Pass@k, and Pass^k across three trials to distinguish genuine capability from lucky outcomes. Experiments on 14 frontier models show that: (1) Trajectory-opaque evaluation is systematically unreliable, missing 44% of safety violations and 13% of robustness failures detected by our framework. (2) Capability does not imply consistency, with Pass@3 remaining stable under error injection while Pass^3 dropping by up to 24 percentage points. (3) Agent capability is strongly multi-dimensional, with model rankings varying across task groups and metrics, indicating that our heterogeneous evaluation coverage is essential. Claw-Eval highlights directions for developing agents that are not only capable but reliably deployable.

Large Language Models (LLMs) have shown promise in simulating human behavior, yet existing agents often exhibit behavioral rigidity, a flaw frequently masked by the self-referential bias of current "LLM-as-a-judge" evaluations. By evaluating against empirical ground truth, we reveal a counter-intuitive phenomenon: increasing the intensity of prompt-driven reasoning does not enhance fidelity but rather exacerbates value polarization, collapsing population diversity. To address this, we propose the Context-Value-Action (CVA) architecture, grounded in the Stimulus-Organism-Response (S-O-R) model and Schwartz's Theory of Basic Human Values. Unlike methods relying on self-verification, CVA decouples action generation from cognitive reasoning via a novel Value Verifier trained on authentic human data to explicitly model dynamic value activation. Experiments on CVABench, which comprises over 1.1 million real-world interaction traces, demonstrate that CVA significantly outperforms baselines. Our approach effectively mitigates polarization while offering superior behavioral fidelity and interpretability.

Agentic AI systems capable of autonomous goal setting and proactive intervention introduce new challenges for regulating moral-emotional processes in learning environments. Existing frameworks typically treat emotion as reactive feedback or engagement optimization, overlooking the need for normative regulation across autonomous decision cycles.This paper proposes an ethical emotion regulation framework for agentic AI learning design inspired by Toegye Yi Hwang's moral-emotional philosophy. The Ethical Emotion Feedback System (EEFS) is reconstructed as a five-stage architecture aligned with agentic cycles, articulating stage-specific design principles and scenario classifications.An EEFS Evaluation Instrument is introduced to enable systematic assessment of moral-emotional alignment in agentic AI systems.

In today's artificial intelligence driven world, modern systems communicate with people from diverse backgrounds and skill levels. For human-machine interaction to be meaningful, systems must be aware of context and user expertise. This study proposes an agentic AI profiler that classifies natural language responses into four levels: Novice, Basic, Advanced, and Expert. The system uses a modular layered architecture built on LLaMA v3.1 (8B), with components for text preprocessing, scoring, aggregation, and classification. Evaluation was conducted in two phases: a static phase using pre-recorded transcripts from 82 participants, and a dynamic phase with 402 live interviews conducted by an agentic AI interviewer. In both phases, participant self-ratings were compared with profiler predictions. In the dynamic phase, expertise was assessed after each response rather than at the end of the interview. Across domains, 83% to 97% of profiler evaluations matched participant self-assessments. Remaining differences were due to self-rating bias, unclear responses, and occasional misinterpretation of nuanced expertise by the language model.

Computer-Use Agents (CUAs) leverage large language models to execute GUI operations on desktop environments, yet they generate actions without evaluating action quality, leading to irreversible errors that cascade through subsequent steps. We propose IntentScore, a plan-aware reward model that learns to score candidate actions from 398K offline GUI interaction steps spanning three operating systems. IntentScore trains with two complementary objectives: contrastive alignment for state-action relevance and margin ranking for action correctness. Architecturally, it embeds each candidate's planning intent in the action encoder, enabling discrimination between candidates with similar actions but different rationales. IntentScore achieves 97.5% pairwise discrimination accuracy on held-out evaluation. Deployed as a re-ranker for Agent S3 on OSWorld, an environment entirely unseen during training, IntentScore improves task success rate by 6.9 points, demonstrating that reward estimation learned from heterogeneous offline trajectories generalizes to unseen agents and task distributions.

Evaluating GUI agents presents a distinct challenge: trajectories are long, visually grounded, and open-ended, yet evaluation must be both accurate and interpretable. Existing approaches typically apply a single holistic judgment over the entire action-observation sequence-a strategy that proves unreliable on long-horizon tasks and yields binary verdicts offering no insight into where or why an agent fails. This opacity limits the utility of evaluation as a diagnostic tool for agent development. We introduce GUIDE (GUI Understanding and Interpretable Diagnostic Evaluation), a framework that decomposes trajectory assessment into three sequential stages mirroring the compositional structure of GUI tasks. Trajectory Segmentation partitions the full trace into semantically coherent subtask units. Subtask Diagnosis evaluates each unit in context, assigning a completion verdict and generating a structured error analysis with corrective recommendations. Overall Summary aggregates per-subtask diagnoses into a task-level judgment. By operating on bounded subtask segments rather than full trajectories, GUIDE mitigates the context overload that degrades existing evaluators as task complexity grows. We validate GUIDE on three benchmarks: an industrial e-commerce dataset of 932 trajectories, AGENTREWARDBENCH spanning five web agent tasks with 1302 trajectories, and AndroidBench for mobile device control. Across all settings, GUIDE substantially outperforms existing evaluators-achieving up to 5.35 percentage points higher accuracy than the strongest baseline-while producing structured diagnostic reports that directly inform agent improvement.