Daily Papers

Self-evolving agents improve future behavior from interaction experience, yet existing evaluations typically optimize under fixed execution conditions and do not test recovery after those conditions change. To address this gap, we introduce PACE-Bench (Physics Adaptation via Code Evolution), a simulator-grounded benchmark of 144 source-to-target adaptation pairs across six physics domains. Each pair links a source environment to a mutated target environment with the same goal and interface. A code-driven design that succeeds in the source fails in the target, where agents must iteratively adapt it into a working target design using diagnostic sandbox feedback within a limited attempt budget. We compare ten self-evolving methods from four paradigms. The benchmark remains far from saturated: Reflexion + Qwen3-14B succeeds on only 35.9\% of full-benchmark pairs, while GPT-5.5 solves 66.7\% of the Statics subset under the full budget. Together, these results show that simulator-grounded reflection is more reliable than unverified self-revision, while memory anchors agents to early designs and broad tree search explores without converging. Even revealing exact physical changes does not raise the performance ceiling, pointing to mechanism redesign rather than parameter inference as the central bottleneck. Data and code are available at https://github.com/thunlp/PACE-Bench.

We present RepoST, a scalable method to construct environments that provide execution feedback for repository-level code generation for both training and evaluation. Unlike existing works that aim to build entire repositories for execution, which is challenging for both human and LLMs, we provide execution feedback with sandbox testing, which isolates a given target function and its dependencies to a separate script for testing. Sandbox testing reduces the complexity of external dependencies and enables constructing environments at a large scale. We use our method to construct RepoST-Train, a large-scale train set with 7,415 functions from 832 repositories. Training with the execution feedback provided by RepoST-Train leads to a performance gain of 5.5% Pass@1 on HumanEval and 3.5% Pass@1 on RepoEval. We also build an evaluation dataset, RepoST-Eval, and benchmark 12 code generation models.

When a professional web developer's code fails a test, they do not simply re-read the stack trace. They open the application in a browser, click buttons, inspect computed styles, and run diagnostic commands to understand what went wrong. Existing feedback mechanisms for coding agents rely on screenshots, LLM-as-a-judge scoring, or natural-language corrections, but few interact with the live application the way a developer would. We introduce WatchPoint, a simulated-user system that mimics real developer behavior by generating and executing diagnostic scripts against the running application, producing structured observations that guide the coding model's retry. Unlike prior approaches that target single-file edits or evaluate using non-executable metrics, we operate on Web-Bench, a benchmark of 50 multi-file web projects comprising 1,000 sequentially dependent tasks, verified by deterministic end-to-end tests. WatchPoint recovers 57.6% of the tasks it diagnoses, and a controlled user study confirms the simulation's realism: human testers achieve a comparable recovery rate (54.5%), providing evidence that automated diagnostic scripts can substitute for interactive human testing on sequential web development tasks. We further identify a pattern of capability gaps that governs when simulated-user feedback is helpful and when it should be withheld.

LLM agents often retry after external validation rejects a candidate, but the interface between validation and the next model call remains underspecified. We introduce VeriHarness, a code-controlled agent loop in which models generate candidates while external validators control acceptance, budgets, and traces. We use it to compare raw diagnostics with feedback that identifies the failure location, observed value, and admissible alternatives. Across 50 paired TextWorld games under a four-call cap, feedback containing all three fields raises terminal success from 14/50 to 36/50 for Qwen2.5-Coder-14B (+44 percentage points) and from 8/50 to 29/50 for Llama-3.1-8B (+42 points). Ablations locate most of the gain in the admissible alternatives: feedback containing only the location and observed value remains near the raw diagnostic baseline. Presenting the complete repair information in prose instead of a keyed JSON record yields nearly the same success, providing no evidence that JSON syntax itself improves repair. The ordering persists across the tested call budgets and one sampled-decoding setting.

Most clinical benchmarks evaluate language models (LMs) on diagnosis using complete case descriptions. In clinical practice, however, patients present information in different ways, and clinicians must obtain relevant history and determine which examinations are needed before reaching a diagnosis. Diagnostic accuracy alone therefore cannot establish whether an agent gathered essential information or conducted an appropriate clinical assessment. Furthermore, existing benchmarks lack professional clinicians' verification. To address this gap, we introduce KlinikeBench, a benchmark of 333 clinician-authored tasks, each providing an isolated sandbox environment with a virtual patient, clinical tools, and task-specific success criteria. More than 35 clinicians contributed to case authoring and benchmark evaluation. In an empirical study, clinicians gave simulated dialogues higher mean quality ratings than reference conversations, which is adapted from real conversation. In each task, an LM has a fixed budget of turns to communicate with the patient, ask about relevant history, request examinations, follow action constraints, and record a final diagnosis. We score these steps separately as well as together. Across 31 models and seven model families, the best-performing models (e.g., GPT-6-astra and Claude Opus 5) succeed on less than 30% of tasks, even though their diagnosis accuracy reaches 90.7%. Some models benefit from talking with the patient; others diagnose well from a complete chart but perform much worse in conversation. Overall, KlinikeBench provides a testbed for evaluating the full clinical encounter and reveals a substantial gap between diagnostic accuracy and performance in interactive clinical assessment.

Test-time scaling for large language models (LLMs) can involve generating multiple candidate solutions for selection or revising an earlier attempt. To guide selection and revision, we need a verifier that can estimate candidate correctness and provide evidence of failure. In competitive programming, however, official judging systems provide authoritative verdicts under finite submission budgets, without revealing failure explanations or failing inputs. To provide informative feedback without consuming official submissions, we formulate a task in which a model judges and diagnoses a candidate program using only the problem statement and source code, with generated counterexamples serving as checkable evidence. To evaluate this task, we build USACO-JUDGE from the USACO 2026 contests, using accepted and unaccepted programs written by human experts and multiple LLMs. We then train ARBITER, a 35B mixture-of-experts (MoE) agentic verifier, with reinforcement learning in a generic execution sandbox. Training raises counterexample success on incorrect programs from 40.61% to 62.67% and mean binary verdict accuracy on USACO-JUDGE from 71.41% to 87.41%, approaching frontier verification performance. Beyond verification, ARBITER outperforms baselines in both code selection and refinement. In an IOI 2026 case study, ARBITER-guided refinement raises GPT-OSS-120B's expected score above the gold-medal threshold under the official submission limit.

Self-correction is particularly useful when a failure constrains the next repair. Coding agents benefit from this property because compilers, tests, and execution traces turn many failures into typed recovery signals, but broad language-agent tasks often expose only a coarse task failure. This creates a tension for generic recovery playbooks: they broaden the agent's context precisely when the system needs a narrower repair interface, mixing incompatible signals for invalid actions, missing procedures, and strict-format errors. Our insight is that development-set failures can recover part of the missing diagnostic substrate by deciding which recovery interventions are admissible before test-time correction. We propose DARC, a diagnosis-guided recovery harness that profiles task-family failure modes, prunes mismatched interventions from a shared recovery library, and freezes a verifier-selected success-cost policy for deployment. This causal order makes correction selective: the harness first determines what kind of failure can be repaired, then decides how much recovery evidence to spend. In ALFWorld, AppWorld, and XBRL Finance, the same protocol yields an action-validity harness, a procedural-recovery fallback, and a format-precision retrieval policy; in each evaluated setting it improves average task performance over base agents and broad playbooks while reducing environment steps or retrieval budget. Our experiments show that failures need not trigger uniformly more context: DARC turns self-correction from prompt expansion into recovery-interface design. DARC provides a practical route toward more reliable agents in domains where compiler-like feedback is absent: making failures actionable before making contexts larger.

Recent agent benchmarks increasingly ground evaluation in executable environments, from code repair to web navigation, app APIs, and function calling. Yet completing consequential work beyond code requires more than producing a plausible response or valid tool call: agents must gather missing information over multiple turns, follow domain policies, coordinate dependent tools, and realize the correct persistent state transition without collateral effects. In this paper, we introduce Thinkingbox, a sandbox for tool-agent-user interaction that provides isolated MCP-compatible tool sessions, complete execution traces, and outcome evaluation over terminal backend state. Built on this sandbox, Thinkingbox-bench contains 507 policy-conditioned workflows across numerous scenarios, including retail, hospitality, auto insurance, neobank internal IT, and consulting IT/HR support. Each attempt is evaluated by task-specific executable checks that accept valid trajectories while rejecting wrong, missing, or extra effects; designated tasks additionally check required properties of the final response. Across proprietary and open-weight models, the strongest achieves 65.36% pass@1, but only 25.25% pass^20. Moreover, many failed trials show clean termination and valid state-changing actions, showing that response or tool-call-level signals are not clear proxies for end-to-end task completion. Thinkingbox-bench reveals a large gap between occasionally finding a successful trajectory and reliably completing stateful business tasks. We release both Thinkingbox and Thinkingbox-Bench: https://github.com/microsoft/thinkingbox

In this paper, we present a framework for training large language models (LLMs) as diagnostic agents with reinforcement learning, enabling them to manage multi-turn diagnostic processes, adaptively select examinations, and commit to final diagnoses. Unlike instruction-tuned models trained on static case summaries, our method acquires diagnostic strategies through interactive exploration and outcome-based feedback. Our contributions are fourfold: (i) We present DiagGym, a diagnostics world model trained with electronic health records that emits examination outcomes conditioned on patient history and recommended examination, serving as a virtual clinical environment for realistic diagnosis training and evaluation; (ii) We train DiagAgent via end-to-end, multi-turn reinforcement learning to learn diagnostic policies that optimize both information yield and diagnostic accuracy; (iii) We introduce DiagBench, a diagnostic benchmark comprising 750 cases with physician-validated examination recommendations and 99 cases annotated with 973 physician-written rubrics on diagnosis process; (iv) we demonstrate superior performance across diverse diagnostic settings. DiagAgent significantly outperforms 10 state-of-the-art LLMs, including DeepSeek-v3 and GPT-4o, as well as two prompt-engineered agents. In single-turn settings, DiagAgent achieves 9.34% higher diagnostic accuracy and 44.03% improvement in examination recommendation hit ratio. In end-to-end settings, it delivers 15.12% increase in diagnostic accuracy and 23.09% boost in examination recommendation F1 score. In rubric-based evaluation, it surpasses the next-best model, Claude-sonnet-4, by 7.1% in weighted rubric score. These findings indicate that learning policies in interactive clinical environments confers dynamic and clinically meaningful diagnostic management abilities unattainable through passive training alone.

The emergence of large-scale multi-modal generative models has drastically advanced artificial intelligence, introducing unprecedented levels of performance and functionality. However, optimizing these models remains challenging due to historically isolated paths of model-centric and data-centric developments, leading to suboptimal outcomes and inefficient resource utilization. In response, we present a novel sandbox suite tailored for integrated data-model co-development. This sandbox provides a comprehensive experimental platform, enabling rapid iteration and insight-driven refinement of both data and models. Our proposed "Probe-Analyze-Refine" workflow, validated through applications on state-of-the-art LLaVA-like and DiT based models, yields significant performance boosts, such as topping the VBench leaderboard. We also uncover fruitful insights gleaned from exhaustive benchmarks, shedding light on the critical interplay between data quality, diversity, and model behavior. With the hope of fostering deeper understanding and future progress in multi-modal data and generative modeling, our codes, datasets, and models are maintained and accessible at https://github.com/modelscope/data-juicer/blob/main/docs/Sandbox.md.

As LLMs are increasingly used for pre-submission self-review, there is growing demand for feedback that not only identifies weaknesses but also guides authors toward concrete revisions. We study this as Actionable Peer-review Generation and decompose it into two subtasks: diagnostic claim generation and revision suggestion generation. We introduce ActReview, a rebuttal-guided post-training framework that connects paper-specific diagnoses to concrete, grounded revision plans. Our central insight is that author rebuttals reveal plausible actions for addressing reviewer concerns and can therefore provide latent supervision for revision-oriented feedback. From real review-rebuttal threads on OpenReview, we construct ActReview-40K by aligning reviewer weaknesses with author responses and grounding the resulting feedback in localized paper evidence. We post-train Qwen3-8B-Base with multi-task supervised fine-tuning followed by GRPO using candidate-aware, weakness-specific rubric rewards. We also introduce ActReview-Bench, a human-curated benchmark of 1,000 instances for evaluating diagnostic quality and revision usefulness. Experiments show that ActReview outperforms prior specialized review-generation models on actionability and grounding while remaining competitive with strong prompt-based LLMs. Human evaluation confirms improved revision usefulness while revealing a remaining gap in technical accuracy, and additional analyses support generalization to held-out papers and robustness across independent judges.

Software testing has progressed toward intelligent automation, yet current AI-based test generators still suffer from static, single-shot outputs that frequently produce invalid, redundant, or non-executable tests due to the lack of execution aware feedback. This paper introduces an agentic multi-model testing framework a closed-loop, self-correcting system in which a Test Generation Agent, an Execution and Analysis Agent, and a Review and Optimization Agent collaboratively generate, execute, analyze, and refine tests until convergence. By using sandboxed execution, detailed failure reporting, and iterative regeneration or patching of failing tests, the framework autonomously improves test quality and expands coverage. Integrated into a CI/CD-compatible pipeline, it leverages reinforcement signals from coverage metrics and execution outcomes to guide refinement. Empirical evaluations on microservice based applications show up to a 60% reduction in invalid tests, 30% coverage improvement, and significantly reduced human effort compared to single-model baselines demonstrating that multi-agent, feedback-driven loops can evolve software testing into an autonomous, continuously learning quality assurance ecosystem for self-healing, high-reliability codebases.

In this work, we introduce MedAgentSim, an open-source simulated clinical environment with doctor, patient, and measurement agents designed to evaluate and enhance LLM performance in dynamic diagnostic settings. Unlike prior approaches, our framework requires doctor agents to actively engage with patients through multi-turn conversations, requesting relevant medical examinations (e.g., temperature, blood pressure, ECG) and imaging results (e.g., MRI, X-ray) from a measurement agent to mimic the real-world diagnostic process. Additionally, we incorporate self improvement mechanisms that allow models to iteratively refine their diagnostic strategies. We enhance LLM performance in our simulated setting by integrating multi-agent discussions, chain-of-thought reasoning, and experience-based knowledge retrieval, facilitating progressive learning as doctor agents interact with more patients. We also introduce an evaluation benchmark for assessing the LLM's ability to engage in dynamic, context-aware diagnostic interactions. While MedAgentSim is fully automated, it also supports a user-controlled mode, enabling human interaction with either the doctor or patient agent. Comprehensive evaluations in various simulated diagnostic scenarios demonstrate the effectiveness of our approach. Our code, simulation tool, and benchmark are available at https://medagentsim.netlify.app/.

Future AI systems could conceal their capabilities ('sandbagging') during evaluations, potentially misleading developers and auditors. We stress-tested sandbagging detection techniques using an auditing game. First, a red team fine-tuned five models, some of which conditionally underperformed, as a proxy for sandbagging. Second, a blue team used black-box, model-internals, or training-based approaches to identify sandbagging models. We found that the blue team could not reliably discriminate sandbaggers from benign models. Black-box approaches were defeated by effective imitation of a weaker model. Linear probes, a model-internals approach, showed more promise but their naive application was vulnerable to behaviours instilled by the red team. We also explored capability elicitation as a strategy for detecting sandbagging. Although Prompt-based elicitation was not reliable, training-based elicitation consistently elicited full performance from the sandbagging models, using only a single correct demonstration of the evaluation task. However the performance of benign models was sometimes also raised, so relying on elicitation as a detection strategy was prone to false-positives. In the short-term, we recommend developers remove potential sandbagging using on-distribution training for elicitation. In the longer-term, further research is needed to ensure the efficacy of training-based elicitation, and develop robust methods for sandbagging detection. We open source our model organisms at https://github.com/AI-Safety-Institute/sandbagging_auditing_games and select transcripts and results at https://huggingface.co/datasets/sandbagging-games/evaluation_logs . A demo illustrating the game can be played at https://sandbagging-demo.far.ai/ .

Reality may establish that an outcome occurred without identifying which evolving procedure produced it or why. This distinction matters in production ML systems whose code, configuration, and artifacts change while external feedback accumulates. We examine it in a human-directed, agent-engineered quantitative trading system, using oracle to mean an external source of realized outcomes rather than a complete correctness specification. Across one year, the account gained and outperformed a broad market index, while annual alpha was not statistically distinguishable from zero under the main retrospective specification. Retrospectively selected subperiods include adverse relative performance and conditional candidate-level weakness under declared approximate references. Engineering records document changes during the episode, and complete recommendation-to-runtime binding is unavailable. The archive does not establish a common frozen instance or a unique cause. The case motivates an outcome--diagnosis gap: outcome evidence, evaluated-object identity, and causal explanation support distinct claims. We distinguish frozen instances, pre-specified adaptive procedures, and ad-hoc development; organize archive-relative claim identifiability and an evidence hierarchy; and propose a prospective production-binding protocol. An illustrative compatible-history example shows how factual binding can resolve a recommendation's referent without supplying its counterfactual effect. The protocol is proposed rather than prospectively validated. External feedback constrains outcome claims, while provenance and additional identification structure determine the resolution of diagnosis.

Automated research systems support literature synthesis, ideation, experiments, writing, and peer review, but their evaluation is dispersed across tasks, benchmarks, and studies that are difficult to compare directly. We review this literature from the perspective of evaluation design and evidence, covering six targets: literature synthesis, research ideation, executable workflows, scholarly writing and communication, automatic peer review, and end-to-end research. We compare task construction, evidence sources, evaluators, and scoring procedures to explain the capabilities assessed by different designs. Our synthesis highlights three recurring lessons: output checks, process checks, and human studies provide complementary information; evaluator calibration is specific to the property being assessed; and resource budgets and attempt selection are integral to interpreting performance comparisons. We identify diagnostic evaluation designs and documented gaps in supporting evidence, and translate these comparisons into reporting and audit recommendations for specific evaluation settings. The survey helps readers navigate existing evaluations, select appropriate benchmarks, and design subsequent studies.

Traditionally, AI research in medical diagnosis has largely centered on image analysis. While this has led to notable advancements, the absence of patient-reported symptoms continues to hinder diagnostic accuracy. To address this, we propose a Pre-Consultation Dialogue Framework (PCDF) that mimics real-world diagnostic procedures, where doctors iteratively query patients before reaching a conclusion. Specifically, we simulate diagnostic dialogues between two vision-language models (VLMs): a DocVLM, which generates follow-up questions based on the image and dialogue history, and a PatientVLM, which responds using a symptom profile derived from the ground-truth diagnosis. We additionally conducted a small-scale clinical validation of the synthetic symptoms generated by our framework, with licensed clinicians confirming their clinical relevance, symptom coverage, and overall realism. These findings indicate that the resulting DocVLM-PatientVLM interactions form coherent, multi-turn consultations paired with images and diagnoses, which we then use to fine-tune the DocVLM. This dialogue-based supervision leads to substantial gains over image-only training, highlighting the value of realistic symptom elicitation for diagnosis.

Benchmarks are necessary for healthcare evaluation, but are not sufficient for predicting deployment performance. Our position is that the evaluation--deployment gap arises not because of poorly designed benchmarks, but from implicit assumptions about how users interact with models that cannot be surfaced from benchmarks alone. To make this precise, we propose a classification of assumptions into two categories: task, which can be tested from conversation data alone, and outcome, which requires outcome data and behavioral studies for testing. Critically, outcome assumptions depend on human behavior, something that even well-designed benchmarks cannot directly observe. To demonstrate the operationality of this framework, we retrospectively analyze a healthcare RCT as a case study and find that the gap naturally separates into task and outcome gaps of roughly equal size. To address this, we make two contributions: first, we propose BenchmarkCards, an artifact that documents assumptions, and second, we propose staged evaluation, a procedure that systematically tests assumptions and evaluates performance.

Interactive dialogue games test a capability that static benchmarks largely leave implicit: a model must carry state across turns, interpret feedback, and choose valid actions under changing constraints. We study this setting in the LM Playschool Challenge with a 2B open-weight model, and find that many failures are not only broad knowledge failures but also local decision failures: repeated guesses, malformed actions, and violations of feedback that the model has just seen. These diagnostics motivate a training recipe organized around three steps: acquire broad game participation through supervised fine-tuning, repair mechanically verifiable failures within one targeted dialogue-game family using turn-local preference pairs, and preserve general capabilities beyond these dialogue games. In the official final evaluation, our submission improves public clemscore from 10.67 to 38.92 and closed in-domain score from 13.41 to 41.17, while approximately preserving aggregate static performance (44.14 vs. 44.24 for the baseline). Out-of-domain clemscore remains low at 7.88, with the largest gains concentrated in unseen variants of the targeted family. Our results suggest that broad SFT brings most of the model's capability improvement; turn-local supervision can be effective when failure detection is precise, with observed transfer concentrated primarily within-family.

This study investigates GPT-4's assessment of its performance in healthcare applications. A simple prompting technique was used to prompt the LLM with questions taken from the United States Medical Licensing Examination (USMLE) questionnaire and it was tasked to evaluate its confidence score before posing the question and after asking the question. The questionnaire was categorized into two groups-questions with feedback (WF) and questions with no feedback(NF) post-question. The model was asked to provide absolute and relative confidence scores before and after each question. The experimental findings were analyzed using statistical tools to study the variability of confidence in WF and NF groups. Additionally, a sequential analysis was conducted to observe the performance variation for the WF and NF groups. Results indicate that feedback influences relative confidence but doesn't consistently increase or decrease it. Understanding the performance of LLM is paramount in exploring its utility in sensitive areas like healthcare. This study contributes to the ongoing discourse on the reliability of AI, particularly of LLMs like GPT-4, within healthcare, offering insights into how feedback mechanisms might be optimized to enhance AI-assisted medical education and decision support.

Code agents can autonomously resolve GitHub issues, yet when they fail, current evaluation provides no visibility into where or why. Metrics such as Pass@1 collapse an entire execution into a single binary outcome, making it difficult to identify where and why the agent went wrong. To address this limitation, we introduce TRAJEVAL, a diagnostic framework that decomposes agent trajectories into three interpretable stages: search (file localization), read (function comprehension), and edit (modification targeting). For each stage, we compute precision and recall by comparing against reference patches. Analyzing 16,758 trajectories across three agent architectures and seven models, we find universal inefficiencies (all agents examine approximately 22x more functions than necessary) yet distinct failure modes: GPT-5 locates relevant code but targets edits incorrectly, while Qwen-32B fails at file discovery entirely. We validate that these diagnostics are predictive, achieving model-level Pass@1 prediction within 0.87-2.1% MAE, and actionable: real-time feedback based on trajectory signals improves two state-of-the-art models by 2.2-4.6 percentage points while reducing costs by 20-31%. These results demonstrate that our framework not only provides a more fine-grained analysis of agent behavior, but also translates diagnostic signals into tangible performance gains. More broadly, TRAJEVAL transforms agent evaluation beyond outcome-based benchmarking toward mechanism-driven diagnosis of agent success and failure.

Machine learning engineering (MLE) agents have made substantial progress, but learning through ML experimentation remains costly in time and computation. Synthetic environments reduce these costs while introducing variations in data and experimental settings that require task-specific diagnosis. Access to diagnostic tools alone does not ensure that agents learn when to use them or how to act on their findings. We introduce ToolMLBench, a suite of executable tools for data inspection, code verification, and experiment diagnosis, together with an SFT and RL pipeline for learning their use. Diagnostic calls acquire evidence whose value depends on subsequent decisions, so final outcomes provide limited guidance on which calls to reinforce. We address this challenge with SPICE, which measures how privileged context changes the likelihood of a sampled tool action and uses this difference as a turn-level reward alongside the final outcome. We train on 80 synthetic tasks and evaluate on 25 in-domain and 10 out-of-domain tasks. Providing tool interfaces and descriptions alone yields inconsistent gains across unadapted models. With the same diagnostic interface, our training pipeline raises in-domain success from 24.8% to 52.4% for Qwen3-8B and from 35.6% to 69.2% for Qwen3.5-35B-A3B. The latter also improves from 31% to 48% out-of-domain, supporting learned diagnostic tool use on held-out sources and targets.

Software development requires more than editing code: developers repeatedly run software, interact with its interfaces, visually inspect its behavior, and use these observations to decide what to change next and whether a change works. Existing coding agents and computer-use agents are largely studied in isolation, leaving this integrated development process underexplored. Diagnosing a runtime interaction failure requires agents to connect visual observations with the responsible code, then use the application again to verify the repair. We introduce CUA-SWE, a benchmark, environment, and evaluation pipeline for software engineering with computer use. Beyond studying how GUI feedback supports diagnosis and repair, we ask whether agents can complete software engineering tasks when required specification or operational information is available only through the running application's visual interface. CUA-SWE spans four software engineering domains and requires agents to modify code and configuration, execute commands, interact with running software, and inspect visual feedback within the same task. Each task includes deterministic, task-specific tests that verify whether the resulting software satisfies the requirements and preserves specified behavior. Our evaluation characterizes how frontier agents combine source-level execution with application screenshots and graphical interaction to produce verified software changes. We examine performance across domains and task information requirements, alongside the development behaviors associated with successful repairs. CUA-SWE provides a unified testbed for studying how agents use visual feedback and interaction to guide software engineering, with executable correctness criteria for the resulting software.

Recent advances in medical multi-modal models focus on specialized image analysis like dermatology, pathology, or radiology. However, they do not fully capture the complexity of real-world clinical diagnostics, which involve heterogeneous inputs and require ongoing contextual understanding during patient-physician interactions. To bridge this gap, we introduce PulseMind, a new family of multi-modal diagnostic models that integrates a systematically curated dataset, a comprehensive evaluation benchmark, and a tailored training framework. Specifically, we first construct a diagnostic dataset, MediScope, which comprises 98,000 real-world multi-turn consultations and 601,500 medical images, spanning over 10 major clinical departments and more than 200 sub-specialties. Then, to better reflect the requirements of real-world clinical diagnosis, we develop the PulseMind Benchmark, a multi-turn diagnostic consultation benchmark with a four-dimensional evaluation protocol comprising proactiveness, accuracy, usefulness, and language quality. Finally, we design a training framework tailored for multi-modal clinical diagnostics, centered around a core component named Comparison-based Reinforcement Policy Optimization (CRPO). Compared to absolute score rewards, CRPO uses relative preference signals from multi-dimensional com-parisons to provide stable and human-aligned training guidance. Extensive experiments demonstrate that PulseMind achieves competitive performance on both the diagnostic consultation benchmark and public medical benchmarks.

Autonomous research agents can generate experiments faster than researchers can validate them. Researchers have responded by scaling the proposer and ranking more samples with a learned judge or human reviewers. We argue that this *generate-and-rank* paradigm misses the problem of sparse feedback. Within a declared research problem, an agent follows the control loop of a greybox fuzzer: it proposes a candidate, executes it, observes feedback, and chooses what to try next. A fuzzer rarely finds a bug, but coverage makes partial progress observable on every execution. Fuzzers then use that signal to mutate inputs and allocate effort, rather than only to rank completed runs. Auto-research needs the same two capabilities. First, each experiment should expose a cheap, dense signal of epistemic progress before final scientific validation is available. Second, that signal should determine the next intervention so that the agent searches rather than repeatedly samples. Because the optimized progress signal is guidance rather than a verdict, final validation must still decide what counts as a discovery using evidence protected from adaptive reuse. We propose controlled tests of whether candidate signals predict validated progress, whether feedback-directed search yields more validated discoveries per unit cost than repeated sampling, and whether protected validation reduces false discoveries. Feedback architecture, not only generation, is a central bottleneck in auto-research.

Realistic practice and tailored feedback are key processes for training peer counselors with clinical skills. However, existing mechanisms of providing feedback largely rely on human supervision. Peer counselors often lack mechanisms to receive detailed feedback from experienced mentors, making it difficult for them to support the large number of people with mental health issues who use peer counseling. Our work aims to leverage large language models to provide contextualized and multi-level feedback to empower peer counselors, especially novices, at scale. To achieve this, we co-design with a group of senior psychotherapy supervisors to develop a multi-level feedback taxonomy, and then construct a publicly available dataset with comprehensive feedback annotations of 400 emotional support conversations. We further design a self-improvement method on top of large language models to enhance the automatic generation of feedback. Via qualitative and quantitative evaluation with domain experts, we demonstrate that our method minimizes the risk of potentially harmful and low-quality feedback generation which is desirable in such high-stakes scenarios.

An agent system's execution traces record how it fails, and procedures that improve such a system without changing model weights (trajectory selection, prompt and workflow optimization, runtime monitoring) read these traces for feedback. Yet raw traces are a poor medium for accumulating that feedback: long, instance-specific, and lacking a stable vocabulary for recurring failures. We argue that an agent system should instead maintain an explicit representation of how it fails, induced from its own behavior and reusable wherever failure feedback is needed. AdaMAST builds this representation by converting a target system's traces into a compact, evidence-grounded failure taxonomy: named failure codes organized along three fixed axes (system-level, role-specific, and domain-specific), with every name, definition, and evidence pattern induced from the traces; no code is hand-authored, no trace human-annotated. The taxonomy is not merely a post-hoc diagnostic but a shared feedback interface, improving agents in three ways. In agent-system search, taxonomy-coded diagnoses of failed candidates outperform free-form reflection on all five benchmarks we test. At runtime, taxonomy feedback raises SWE-agent's resolution on SWE-bench Verified Mini from 60% with free-text reflection to 70%, and improves Claude Code from 64.0% to 70.7% as a runtime skill. In trajectory selection, AdaMAST-Judge, a verifier built on the induced codes, improves best-of-5 accuracy on Terminal-Bench 2.0 by 8-15 points over Pass@1. The vocabulary itself is compact (an order-of-magnitude compression that preserves trace distinctions), human-faithful (matching expert failure annotations more closely than a hand-crafted reference vocabulary), and adaptive (taxonomies induced for different domains share few codes). Adaptive failure taxonomies close the loop between the traces agents produce and the procedures that improve them.

Large Language Models (LLMs) and code agents in software development are rapidly evolving from generating isolated code snippets to producing full-fledged software applications with graphical interfaces, interactive logic, and dynamic behaviors. However, current benchmarks fall short in evaluating such production-ready software, as they often rely on static checks or binary pass/fail scripts, failing to capture the interactive behaviors and runtime dynamics that define real-world usability - qualities that only emerge when an application is actively used. This is the blind spot of current evaluation: you don't know if an app works until you click through it, interact with it, and observe how it responds. To bridge this gap, we introduce RealDevWorld, a novel evaluation framework for automated end-to-end assessment of LLMs' ability to generate production-ready repositories from scratch. It features two key components: (1) RealDevBench, a diverse collection of 194 open-ended software engineering tasks across multiple domains, incorporating multimodal elements to reflect real-world complexity; and (2) AppEvalPilot, a new agent-as-a-judge evaluation system that simulates realistic, GUI-based user interactions to automatically and holistically assess software functional correctness, visual fidelity, and runtime behavior. The framework delivers fine-grained, task-specific diagnostic feedback, supporting nuanced evaluation beyond simple success/failure judgments. Empirical results show that RealDevWorld delivers effective, automatic, and human-aligned evaluations, achieving an accuracy of 0.92 and a correlation of 0.85 with expert human assessments, while significantly reducing the reliance on manual review. This enables scalable, human-aligned assessment of production-level software generated by LLMs. Our code is available on GitHub.

AI agents are becoming increasingly capable of generating scientific code, but generating code is not the same as improving the algorithms behind it. For numerical solvers, execution feedback can expose poor performance, but rarely reveals its underlying cause and how to address it. We introduce Auto-Diagnosis and Skill Discovery (ADSD), a framework that links numerical diagnosis to reusable solver self-improvement. ADSD follows a diagnosis-first paradigm that first explains why a solver performs poorly, then uses this diagnosis to guide the discovery of appropriate numerical methods. The resulting knowledge is packaged into reusable solver skills, turning solver improvement from trial-and-error editing into a structured process of diagnosis, discovery, and implementation. Across four challenging numerical domains--power flow equation, AC optimal power flow control, stiff ordinary differential equations, and heterogeneous diffusion PDEs--ADSD consistently improves solver accuracy, robustness, and efficiency. On GOC-500 power flow, for example, ADSD reduces mean solver error by nearly 71times, with improvements further transferring to unseen grid topologies and operating regimes.

Tool-using LLM agents are mostly evaluated assuming all tools work. When a tool times out, returns a week-stale value, or has its description poisoned in deployment, the developer needs a controlled way to reproduce the failure, test a fix, and confirm the fix worked before deployment. We present AgentCheck, an open-source web workbench that turns an MCP server into an intervention surface. AgentCheck runs an agent against its real tools and records every tool response, then re-runs the agent with the response perturbed by a fault (12 types) injector. Matching tool calls are replayed from cache, and later tool calls go live after the agent diverges. This yields a reproduce-intervene-confirm loop: the developer toggles a mitigation, re-runs against the identical fault, and sees if the failure goes away. Scoring has two parts: deterministic pass/fail rules, plus an LLM judge for interpretive labels, validated against human annotations. Across five agents, the best passes 105/120 scenarios and the weakest only 77. The failures are usually silent, confident use of incorrect tool outputs rather than crashes. On the weakest agent, a retry mitigation raises success on timeout error faults from as few as 30% of cases to 100%, whereas stale-data faults remain near 3-4 of 10 regardless of the mitigation. AgentCheck makes these failure modes reproducible, comparable, and verifiable before deployment.

Autoresearch agents are reshaping the research ecosystem, but they can also let flawed claims enter the literature at scale. Human advisors catch such issues in drafts through careful, traceable feedback, yet advisor-style assessment requires extensive manual effort and does not scale. To shift automated paper assessment from a judge to a diagnostician, we introduce PaperDoctor, an agent framework for pre-submission feedback with three key innovations. First, a holistic hierarchical framework evaluates writing, layout, references, code, theory, prior work, and experiments through three layers: L1 surface screening, L2 typed verifiers that route each claim to the appropriate evidence, and L3 reproducers that rerun experiments by priority. Second, each finding contains an observation, a pointer to specific evidence such as a sentence, equation, or code line, and a revision suggestion, making critiques auditable and actionable. Third, PaperDoctor selectively rebuilds and reruns experiments based on claim importance and compute budget, surfacing reproducibility gaps and quantitative limitations that are invisible from the manuscript alone. We evaluate PaperDoctor on 30 in-progress papers, yielding 70.6% agreement and all positive holistic scores, and on 40 manuscripts across machine learning, natural science, and social science, covering human- and AI-authored papers with code. Overall, PaperDoctor produces more auditable feedback than human and other agentic reviewers, pairs critiques with concrete suggestions by design, and complements dimensions often overlooked by human reviewers. We also develop an interactive interface that lets authors browse findings grounded in their paper. PaperDoctor reframes automated paper assessment as diagnosis rather than verdict, taking a concrete step toward AI advisors for more rigorous AI-assisted scientific discovery.

Clinical diagnosis is critical in medical practice, typically requiring a continuous and evolving process that includes primary diagnosis, differential diagnosis, and final diagnosis. However, most existing clinical diagnostic tasks are single-step processes, which does not align with the complex multi-step diagnostic procedures found in real-world clinical settings. In this paper, we propose a multi-step diagnostic task and annotate a clinical diagnostic dataset (MSDiagnosis). This dataset includes primary diagnosis, differential diagnosis, and final diagnosis questions. Additionally, we propose a novel and effective framework. This framework combines forward inference, backward inference, reflection, and refinement, enabling the LLM to self-evaluate and adjust its diagnostic results. To assess the effectiveness of our proposed method, we design and conduct extensive experiments. The experimental results demonstrate the effectiveness of the proposed method. We also provide a comprehensive experimental analysis and suggest future research directions for this task.

Autonomous improvement loops are hard to trust because the improvement process is usually external scaffolding bolted onto the agent: failures go unlogged, diagnoses cannot be replayed, and promote-or-discard decisions land in a side database rather than the agent's own history. We show that an event-sourced agent runtime removes that friction and turns controlled improvement into a first-class workflow. When the agent's state is a deterministic projection of an append-only event log, failures are recorded, a run replays exactly from its log, candidate patches scope to typed pipeline seams, gates are auditable, and every promotion or discard is itself an event. We demonstrate this with Regimes, a loop on the ActiveGraph runtime that diagnoses failed evaluations, proposes a repair at a pipeline point, and promotes it only after static checks, sandbox execution, in-sample evaluation, and held-out validation. The loop is target-agnostic: the same control flow runs against different tasks through a common interface. On LongMemEval-S the dominant failure is not retrieval but reconciliation: the evidence is already in the assembled context, yet the reader answers incorrectly. Across five seeded held-out splits, Regimes discovers reader-prompt repairs that improve final held-out accuracy by +0.05 to +0.10 in four splits and +0.01 in one over-promotion split; two splits are individually significant (seed 5 unadjusted for its sequential promotion structure), and the pooled count is descriptive only, since the splits share one 500-question pool. The durable contributions are ActiveGraph as an auditable substrate that makes controlled improvement loops tractable, the held-out-gated loop it supports, the failure-regime taxonomy routing each failure to a pipeline location (whose marginal value over an unrouted baseline is the primary open question), and the prompt-as-discovery-probe hypothesis.

Large language models (LLMs) show promise in medical diagnosis, but real-world deployment remains challenging due to high-stakes clinical decisions and imperfect reasoning reliability. As a result, careful inspection of model behavior is essential for assessing whether diagnostic reasoning is reliable and clinically grounded. However, debugging medical LLMs remains difficult. First, developers often lack sufficient medical domain expertise to interpret model errors in clinically meaningful terms. Second, models can fail across a large and diverse set of instances involving different input types, tasks, and reasoning steps, making it challenging for developers to prioritize which errors deserve focused inspection. Third, developers struggle to identify recurring error patterns across cases, as existing debugging practices are largely instance-centric and rely on manual inspection of isolated failures. To address these challenges, we present VeriLLMed, a visual analytics system that integrates external biomedical knowledge to audit and debug medical LLM diagnostic reasoning. VeriLLMed transforms model outputs into comparable reasoning paths, constructs knowledge graph-grounded reference paths, and identifies three recurring classes of diagnosis errors: relation errors, branch errors, and missing errors. Case studies and expert evaluation demonstrate that VeriLLMed helps developers identify clinically implausible reasoning and generate actionable insights that can inform the improvement of medical LLMs.

Medical question answering (QA) benchmarks often focus on multiple-choice or fact-based tasks, leaving open-ended answers to real patient questions underexplored. This gap is particularly critical in mental health, where patient questions often mix symptoms, treatment concerns, and emotional needs, requiring answers that balance clinical caution with contextual sensitivity. We present CounselBench, a large-scale benchmark developed with 100 mental health professionals to evaluate and stress-test large language models (LLMs) in realistic help-seeking scenarios. The first component, CounselBench-EVAL, contains 2,000 expert evaluations of answers from GPT-4, LLaMA 3, Gemini, and online human therapists on patient questions from the public forum CounselChat. Each answer is rated across six clinically grounded dimensions, with span-level annotations and written rationales. Expert evaluations show that while LLMs achieve high scores on several dimensions, they also exhibit recurring issues, including unconstructive feedback, overgeneralization, and limited personalization or relevance. Responses were frequently flagged for safety risks, most notably unauthorized medical advice. Follow-up experiments show that LLM judges systematically overrate model responses and overlook safety concerns identified by human experts. To probe failure modes more directly, we construct CounselBench-Adv, an adversarial dataset of 120 expert-authored mental health questions designed to trigger specific model issues. Expert evaluation of 1,080 responses from nine LLMs reveals consistent, model-specific failure patterns. Together, CounselBench establishes a clinically grounded framework for benchmarking LLMs in mental health QA.

Online medical consultation (OMC) restricts doctors to gathering patient information solely through inquiries, making the already complex sequential decision-making process of diagnosis even more challenging. Recently, the rapid advancement of large language models has demonstrated a significant potential to transform OMC. However, most studies have primarily focused on improving diagnostic accuracy under conditions of relatively sufficient information, while paying limited attention to the "inquiry" phase of the consultation process. This lack of focus has left the relationship between "inquiry" and "diagnosis" insufficiently explored. In this paper, we first extract real patient interaction strategies from authentic doctor-patient conversations and use these strategies to guide the training of a patient simulator that closely mirrors real-world behavior. By inputting medical records into our patient simulator to simulate patient responses, we conduct extensive experiments to explore the relationship between "inquiry" and "diagnosis" in the consultation process. Experimental results demonstrate that inquiry and diagnosis adhere to the Liebig's law: poor inquiry quality limits the effectiveness of diagnosis, regardless of diagnostic capability, and vice versa. Furthermore, the experiments reveal significant differences in the inquiry performance of various models. To investigate this phenomenon, we categorize the inquiry process into four types: (1) chief complaint inquiry; (2) specification of known symptoms; (3) inquiry about accompanying symptoms; and (4) gathering family or medical history. We analyze the distribution of inquiries across the four types for different models to explore the reasons behind their significant performance differences. We plan to open-source the weights and related code of our patient simulator at https://github.com/LIO-H-ZEN/PatientSimulator.

We explore the potential of Large Language Models (LLMs) to assist and potentially correct physicians in medical decision-making tasks. We evaluate several LLMs, including Meditron, Llama2, and Mistral, to analyze the ability of these models to interact effectively with physicians across different scenarios. We consider questions from PubMedQA and several tasks, ranging from binary (yes/no) responses to long answer generation, where the answer of the model is produced after an interaction with a physician. Our findings suggest that prompt design significantly influences the downstream accuracy of LLMs and that LLMs can provide valuable feedback to physicians, challenging incorrect diagnoses and contributing to more accurate decision-making. For example, when the physician is accurate 38% of the time, Mistral can produce the correct answer, improving accuracy up to 74% depending on the prompt being used, while Llama2 and Meditron models exhibit greater sensitivity to prompt choice. Our analysis also uncovers the challenges of ensuring that LLM-generated suggestions are pertinent and useful, emphasizing the need for further research in this area.

Medical vision-language model (VLM) evaluation is sensitive to workflow design, prompting strategy, and benchmark construction, yet most studies treat these factors in isolation. We introduce a reliability stress test for chest X-ray interpretation built on two balanced datasets (a private report-backed set and a curated MIMIC subset). Three medical VLMs (CheXagent, MedGemma-4B, and MedGemma-27B) are evaluated across three prompt styles and two workflows (single-VLM and multi-agent), producing 36 configurations. We show that exact-match accuracy alone can overstate the effectiveness of conservative models that default to "Normal" predictions. Diagnostic reliability also depends heavily on model family and scale: multi-agent reasoning helps some configurations but hurts others. Building on these observations, we propose a decision-time routing framework that selectively escalates to multi-agent inference only when beneficial, improving the cost-quality trade-off over fixed workflows. Our results highlight the need for evaluation protocols that jointly consider prompt sensitivity, failure-mode diversity, and workflow choice before clinical deployment.

Various test-time interventions for Computer Use Agents (CUAs), including critic models, have been developed to improve performance through pre-execution action evaluation in complex Graphical User Interface (GUI) environments. However, existing critics suffer from two key limitations: they (1) focus primarily on short-sighted decision loops (e.g., forgetting earlier actions) and (2) lack the visual grounding needed to detect flawed actions (e.g., clicking wrong UI elements). To address these, we introduce HiViG, a History-aware Visually Grounded test-time framework, built around a multimodal critic trained on real GUI trajectories to abstract past interactions into a compact record and to evaluate actions with visual grounding. At test time, HiViG integrates the critic into the policy decision loop to provide macro-action history, which summarizes the policy's completed achievements, and visually grounded critique, which verifies raw execution coordinates against the current screenshot to intercept errors before execution. Across web, mobile, and desktop benchmarks, HiViG consistently outperforms existing scalar and verbal critics, improving average success rates over the strongest baseline by 5.8% for Qwen3-VL-32B and 9.0% for Gemini-3-Flash, and demonstrates strong cross-platform generalization. Ablations show that macro-action history mitigates short-sighted planning and visually grounded critique reduces execution errors, with both components being critical for test-time scaling in long-horizon GUI tasks.

LLM-as-a-Judge frameworks are increasingly trusted to automate evaluation in place of human experts, yet their reliability in high-stakes medical contexts remains unproven. We stress-test this assumption for detecting incomplete patient-facing medical responses, evaluating three rubric granularities (General-Likert, Analytical-Rubric, Dynamic-Checklist) and three backbone models across two clinician-annotated datasets, including HealthBench, the largest publicly available benchmark for medical response evaluation. LLM Judges discriminate complete from incomplete responses at and slightly above near chance (AUC 0.49--0.66); at the threshold required to recall 90% of incomplete responses, clinicians must still review the vast majority of the dataset, offering no triage utility. Even when model and clinician verdicts agree, they rarely cite the same explanation; and when they diverge, false positives stem from over-flagging non-essential gaps while false negatives reflect outright detection failures. These results reveal that LLM Judges and clinicians apply fundamentally different completeness standards; a finding that undermines their use as autonomous evaluators or triage filters in clinical settings.

Language models can distinguish between testing and deployment phases -- a capability known as evaluation awareness. This has significant safety and policy implications, potentially undermining the reliability of evaluations that are central to AI governance frameworks and voluntary industry commitments. In this paper, we study evaluation awareness in Llama-3.3-70B-Instruct. We show that linear probes can separate real-world evaluation and deployment prompts, suggesting that current models internally represent this distinction. We also find that current safety evaluations are correctly classified by the probes, suggesting that they already appear artificial or inauthentic to models. Our findings underscore the importance of ensuring trustworthy evaluations and understanding deceptive capabilities. More broadly, our work showcases how model internals may be leveraged to support blackbox methods in safety audits, especially for future models more competent at evaluation awareness and deception.

Recent advances in persistent personal-agent frameworks are making human-centered agent networks realistic deployment targets: each user can be served by an AI agent that acts on the user's behalf, maintains state, and communicates with other agents through social and task relations. In these networks, everyday tool use becomes multi-party owned-agent collaboration over personal workspaces, where files, records, tools, and policies are not directly visible across owners. Existing agent benchmarks study tool use and collaboration, but they do not provide an end-to-end sandbox for verifiable cross-user agent collaboration with realistic user digital workspaces or test how harmful actions can travel through the human-centered agent network. We introduce WeClawArena, an auditable benchmark and runtime sandbox for multi-party owned-agent collaboration over personal workspaces. WeClawArena targets collaborative tool-use tasks in which personal workspaces serve as both operational tools and personal constraints. The benchmark contains 124 base tasks across six cross-user task domains and expands them into 620 scenario variants, with one benign control and four attack-vector variants per base task. The sandbox records peer messages, tool calls, resource operations, governed decisions, and final workspace states. WeClawArena reports utility and attack success rate separately and audits attack success from bounded runtime evidence, supporting diagnosis of task breakdown, privacy leakage, poisoned evidence, and invalid authority paths.

A benchmark is not merely a collection of test questions; it naturally induces an empirical measure over difficulty. In this work, we develop this measure-based view of benchmark evaluation. This perspective allows us to identify and formalize two central failure modes: measure mismatch, where the benchmark difficulty distribution differs from the intended difficulty distribution, and coverage gaps, where uncovered regions along the difficulty axis prevent benchmarks from detecting meaningful localized training progress. We propose two diagnostic metrics that capture these two failure modes and provide rigorous theoretical guarantees. Empirically, we analyze 25 popular benchmarks spanning 7 domains using carefully estimated item difficulties. Our results show that many benchmarks implicitly overweight certain difficulty regions, so their leaderboard conclusions should be interpreted with caution; others leave substantial portions of the difficulty axis uncovered, limiting their ability to monitor the model improvement process. Finally, we turn diagnosis into action by introducing BenchFill, a simple difficulty-targeted item rewriting algorithm that effectively reduces local coverage gaps.

During surgical training, real-time feedback from trainers to trainees is important for preventing errors and enhancing long-term skill acquisition. Accurately predicting the effectiveness of this feedback, specifically whether it leads to a change in trainee behavior, is crucial for developing methods for improving surgical training and education. However, relying on human annotations to assess feedback effectiveness is laborious and prone to biases, underscoring the need for an automated, scalable, and objective method. Creating such an automated system poses challenges, as it requires an understanding of both the verbal feedback delivered by the trainer and the visual context of the real-time surgical scene. To address this, we propose a method that integrates information from transcribed verbal feedback and corresponding surgical video to predict feedback effectiveness. Our findings show that both transcribed feedback and surgical video are individually predictive of trainee behavior changes, and their combination achieves an AUROC of 0.70+/-0.02, improving prediction accuracy by up to 6.6%. Additionally, we introduce self-supervised fine-tuning as a strategy for enhancing surgical video representation learning, which is scalable and further enhances prediction performance. Our results demonstrate the potential of multi-modal learning to advance the automated assessment of surgical feedback.

Complete orthogonal predictive coordinates do not by themselves bind a latent direction to a named intervention. We present a mathematical and synthetic audit motivated by vascular device-vessel suitcordance. Capacity-matched least-squares predictors were exactly equivalent under complete fixed output transforms, whereas an anchor-only observer recovered interpretations only within the span of known perturbation signatures. Six three-dimensional configurations across 64 seeds gave a maximum paired prediction discrepancy of 6.7e-15 but a median untransported edit error of 1.513. Coordinate transport removed that error. Noisy and weak anchors constrained calibration stability, and changing the representation basis required recalibration or verified transport. Across 256 additional fits in dimensions 3-24, prediction equivalence persisted within 4.0e-15. We then evaluated nine deliberate runnable fault classes across 64 seeds. All 576 faulty executions completed, but each violated at least one reconstruction, prediction, delivered-edit or scope contract; all 320 valid control records passed. Repeating a faulty implementation gave exact self-agreement despite error against the separately computed simulator expectation. For one omitted-direction defect, probe coverage followed its analytic law, and rank-aware abstention protected unsupported interpretations. Scalar-noise experiments exposed both missed weak faults and excessive rejection under narrow relative tolerances. These controls provide an executable separation of prediction, semantic support and scientific acceptance. They are synthetic numerical audits, not clinical validation, neural JEPA-Anything replication, agent learning or patient treatment-effect estimation.

Real-world clinical diagnosis is a complex process in which the doctor is required to obtain information from both interaction with the patient and conducting medical exams. Additionally, the doctor needs to adapt to different patient personas, as well as noisy and incomplete information that can happen at any time during the process. However, existing benchmarks for medical LLMs and methods for automatic diagnosis largely simplify this process by reducing it to single-turn question answering, noise-free conversations, or sequential exam making, etc., ignoring the interactive and uncertain nature of clinical diagnosis. In this paper, we aim to address this gap by formalizing clinical diagnosis as a Partially Observable Markov Decision Process (POMDP) with three action types: questioning the patient, ordering medical exams as tool calls, and issuing a diagnosis. We also introduce a systematic noise model comprising seven patient noise types and three exam noise types. Using our proposed environment, we train an effective diagnosis agent, MedExAgent, through a two-stage pipeline that first performs supervised finetuning on synthetic conversations structured after the Calgary-Cambridge model for clinical interviews, and then applies DAPO to optimize a composite reward capturing diagnostic accuracy, tool call quality, and exam cost including financial cost and patient discomfort. Through extensive experiments and ablation studies, we demonstrate that MedExAgent achieves diagnostic performance comparable to larger models while maintaining cost-efficient examination strategies.

As Large Multimodal Models (LMMs) scale up and reinforcement learning (RL) methods mature, LMMs have made notable progress in complex reasoning and decision making. Yet training still relies on static data and fixed recipes, making it difficult to diagnose capability blind spots or provide dynamic, targeted reinforcement. Motivated by findings that test driven error exposure and feedback based correction outperform repetitive practice, we propose Diagnostic-driven Progressive Evolution (DPE), a spiral loop where diagnosis steers data generation and reinforcement, and each iteration re-diagnoses the updated model to drive the next round of targeted improvement. DPE has two key components. First, multiple agents annotate and quality control massive unlabeled multimodal data, using tools such as web search and image editing to produce diverse, realistic samples. Second, DPE attributes failures to specific weaknesses, dynamically adjusts the data mixture, and guides agents to generate weakness focused data for targeted reinforcement. Experiments on Qwen3-VL-8B-Instruct and Qwen2.5-VL-7B-Instruct show stable, continual gains across eleven benchmarks, indicating DPE as a scalable paradigm for continual LMM training under open task distributions. Our code, models, and data are publicly available at https://github.com/hongruijia/DPE.

General-purpose health benchmarks increasingly anchor claims about LLM medical performance, but they are not always resolved by clinical specialty, making domain-specific performance hard to isolate. Mental health is of acute public-health concern as millions of people turn to LLMs for psychological support, and most existing evaluations are bespoke academic benchmarks that are difficult to integrate into developer workflows. We introduce HealthBench-Psych and HealthBench-Psych-Hard. We screened HealthBench's 5,000 physician-rubric conversations for mental-health relevance with a transparent LLM-applied rubric, then validated the subset through two rounds of blinded clinician review with concealed known-exclude controls, yielding 610 conversations (12.2% of the corpus). Evaluating 20 frontier and open models under a cross-vendor panel of three LLM judges, we find a statistically tied frontier cluster, measurable refusal behavior in two models, and near-identical rankings across judges (τge 0.92). We release the subset, pipeline, model responses, grades, and analysis code as a reusable resource.

Self-evolving skill systems promise to improve agents by turning execution feedback into persistent skill updates without changing the underlying model. Yet it remains unclear when further evolution helps, how successful and failed trajectories shape revision, and whether extra test-time computation can recover the same gains. To address these questions, we present a controlled evaluation framework across five benchmarks and three models. Our primary study contains 42 feedback runs across 14 supported model-benchmark settings. Within each setting, we hold the executor and optimizer configuration, revision procedure, validation rule, and round budget fixed, while varying only the feedback shown to the optimizer: successes and failures (Normal), failures only, or successes only. Evolution is sparse: only 55 of 388 candidates establish byte-distinct validation bests. Validation-based selection chooses an evolved skill in 11 of 14 settings, nine of which improve released-test performance. All 11 selections come from feedback conditions that include failed trajectories, although the relative ranking of Normal and Fail-only varies across settings. Validation and downstream evaluations on test, robustness, and transfer sometimes favor different feedback views. A broader SearchQA analysis covering eight models shows similarly sparse, feedback-dependent dynamics. In the GPT-5.5 test-time-scaling controls, oracle Parallel Sampling comes within 0.43 points of the evolved SearchQA skill but remains 30.96 points behind on SpreadsheetBench; Sequential Refinement recovers neither gain. Overall, persistent skill self-evolution is better understood as sparse, validation-filtered search with model- and benchmark-dependent returns, rather than steady improvement from additional rounds. The implementation is available at https://github.com/HKUST-KnowComp/rethinkskill.

Despite the rapid progress of Multimodal Large Language Models (MLLMs), their ability to perform reliable visual grounding in high-stakes clinical software environments remains underexplored. Existing GUI benchmarks largely focus on isolated, single-step grounding queries, overlooking the sequential, workflow-driven reasoning required in real-world medical interfaces, where tasks evolve across independent steps and dynamic interface states. We introduce MedSPOT, a workflow-aware sequential grounding benchmark for clinical GUI environments. Unlike prior benchmarks that treat grounding as a standalone prediction task, MedSPOT models procedural interaction as a sequence of structured spatial decisions. The benchmark comprises 216 task-driven videos with 597 annotated keyframes, in which each task consists of 2 to 3 interdependent grounding steps within realistic medical workflows. This design captures interface hierarchies, contextual dependencies, and fine-grained spatial precision under evolving conditions. To evaluate procedural robustness, we propose a strict sequential evaluation protocol that terminates task assessment upon the first incorrect grounding prediction, explicitly measuring error propagation in multi-step workflows. We further introduce a comprehensive failure taxonomy, including edge bias, small-target errors, no prediction, near miss, far miss, and toolbar confusion, to enable systematic diagnosis of model behavior in clinical GUI settings. By shifting evaluation from isolated grounding to workflow-aware sequential reasoning, MedSPOT establishes a realistic and safety-critical benchmark for assessing multimodal models in medical software environments. Code and data are available at: https://github.com/Tajamul21/MedSPOT.

The rapid increase of computed tomography (CT) scans and their time-consuming manual analysis have created an urgent need for robust automated analysis techniques in clinical settings. These aim to assist radiologists and help them managing their growing workload. Existing methods typically generate entire reports directly from 3D CT images, without explicitly focusing on observed abnormalities. This unguided approach often results in repetitive content or incomplete reports, failing to prioritize anomaly-specific descriptions. We propose a new anomaly-guided report generation model, which first predicts abnormalities and then generates targeted descriptions for each. Evaluation on a public dataset demonstrates significant improvements in report quality and clinical relevance. We extend our work by conducting an ablation study to demonstrate its effectiveness.

Coding agents are increasingly used to build programs from scratch as software engineers do rather than to resolve issues within a repository or to generate a repository from a given requirements document. However, scalable verifiable training environments that force agents to discover what is required and to decide how the system should be designed remain scarce. Moreover, existing agent workflows also cannot sustain long-horizon development in which the agent explores the reference program before implementation. To address these challenges, we introduce ProgramGym, a framework consisting of Program-Env and Program-Agent for scalable training environments of long-horizon program reproduction agents. Specifically, Program-Env constructs verifiable training environments from repositories through a feedback-driven loop that repairs runtime sandbox builds and the quality of black-box tests. Using this novel pipeline, we obtain 1,648 training environments spanning C, C++, Go, and Rust. Program-Agent combines a role-isolated workflow of analysis, implementation, and review with revision management for compilation checks and failure recovery. Experiments on ProgramBench show that both Program-Env and Program-Agent improve performance on program reproduction. Fine-tuning on trajectories with these constructed environments raises the performance of Qwen3.8-27B from 34.87% to 42.84%. Program-Agent further raises the base and SFT models to 55.71% and 56.25%, both exceeding some flagship models. We will release the ProgramGym to support further research on coding agents that build programs from scratch.

Generalist foundation models such as GPT-4 have displayed surprising capabilities in a wide variety of domains and tasks. Yet, there is a prevalent assumption that they cannot match specialist capabilities of fine-tuned models. For example, most explorations to date on medical competency benchmarks have leveraged domain-specific training, as exemplified by efforts on BioGPT and Med-PaLM. We build on a prior study of GPT-4's capabilities on medical challenge benchmarks in the absence of special training. Rather than using simple prompting to highlight the model's out-of-the-box capabilities, we perform a systematic exploration of prompt engineering. We find that prompting innovation can unlock deeper specialist capabilities and show that GPT-4 easily tops prior leading results for medical benchmarks. The prompting methods we explore are general purpose, and make no specific use of domain expertise, removing the need for expert-curated content. Our experimental design carefully controls for overfitting during the prompt engineering process. We introduce Medprompt, based on a composition of several prompting strategies. With Medprompt, GPT-4 achieves state-of-the-art results on all nine of the benchmark datasets in the MultiMedQA suite. The method outperforms leading specialist models such as Med-PaLM 2 by a significant margin with an order of magnitude fewer calls to the model. Steering GPT-4 with Medprompt achieves a 27% reduction in error rate on the MedQA dataset over the best methods to date achieved with specialist models and surpasses a score of 90% for the first time. Beyond medical problems, we show the power of Medprompt to generalize to other domains and provide evidence for the broad applicability of the approach via studies of the strategy on exams in electrical engineering, machine learning, philosophy, accounting, law, nursing, and clinical psychology.

Large language models have the potential to be valuable in the healthcare industry, but it's crucial to verify their safety and effectiveness through rigorous evaluation. For this purpose, we comprehensively evaluated both open-source LLMs and Google's new multimodal LLM called Gemini across Medical reasoning, hallucination detection, and Medical Visual Question Answering tasks. While Gemini showed competence, it lagged behind state-of-the-art models like MedPaLM 2 and GPT-4 in diagnostic accuracy. Additionally, Gemini achieved an accuracy of 61.45\% on the medical VQA dataset, significantly lower than GPT-4V's score of 88\%. Our analysis revealed that Gemini is highly susceptible to hallucinations, overconfidence, and knowledge gaps, which indicate risks if deployed uncritically. We also performed a detailed analysis by medical subject and test type, providing actionable feedback for developers and clinicians. To mitigate risks, we applied prompting strategies that improved performance. Additionally, we facilitated future research and development by releasing a Python module for medical LLM evaluation and establishing a dedicated leaderboard on Hugging Face for medical domain LLMs. Python module can be found at https://github.com/promptslab/RosettaEval

Recent developments in large language models (LLMs) have been impressive. However, these models sometimes show inconsistencies and problematic behavior, such as hallucinating facts, generating flawed code, or creating offensive and toxic content. Unlike these models, humans typically utilize external tools to cross-check and refine their initial content, like using a search engine for fact-checking, or a code interpreter for debugging. Inspired by this observation, we introduce a framework called CRITIC that allows LLMs, which are essentially "black boxes" to validate and progressively amend their own outputs in a manner similar to human interaction with tools. More specifically, starting with an initial output, CRITIC interacts with appropriate tools to evaluate certain aspects of the text, and then revises the output based on the feedback obtained during this validation process. Comprehensive evaluations involving free-form question answering, mathematical program synthesis, and toxicity reduction demonstrate that CRITIC consistently enhances the performance of LLMs. Meanwhile, our research highlights the crucial importance of external feedback in promoting the ongoing self-improvement of LLMs.

Clinical diagnosis at hospital admission must be made rapidly from limited, incomplete evidence. Existing diagnosis-prediction benchmarks are poorly suited to this setting: they restrict prediction to closed code sets, exclude free-text notes, and supervise with discharge diagnoses that incorporate the full inpatient course. We introduce EarlyDx, a large-scale benchmark for open-ended early diagnosis, built from 154,834 emergency department encounters in MIMIC-IV. Each encounter is restricted to records available at admission time t_0 and supervised by the diagnoses recorded during the ED encounter rather than at discharge. An LLM auditor further verifies every free-text label as supported, partially supported, or unsupported by that evidence; the primary evaluation scores only fully supported labels. Under a semantic LLM-as-judge protocol, no evaluated system --- frontier general, medical-specialized, or in-domain post-trained --- synthesizes admission-time evidence reliably. Zero-shot models score largely by extraction, recovering only 3-31% of diagnoses that must be inferred rather than read from the record; post-training raises inference-dependent recall to 56%, but a sizeable margin remains, and on time-critical conditions no system attains a clinician's balance of sensitivity and precision. We release the full construction and evaluation pipeline at here.

Healthcare agents must transform longitudinal clinical evidence into reliable actions through heterogeneous tools and stateful workflows, conditions under which model-centric systems may lose context, misuse tools, fail to verify consequential transitions, or cross authorization boundaries. Existing medical-agent benchmarks primarily emphasize terminal task outcomes, making it difficult to determine whether execution failures arise from the agent model, the runtime harness, or their interaction. We present the Medical Harness Protocol, an outcome-preserving framework that independently evaluates seven healthcare-specific process dimensions: Care Action Fulfillment, Clinical Resource Competence, Patient-Evidence Alignment, Clinical Workflow Continuity, Clinical Auditability, Closed-Loop Verification, and Clinical Conduct Boundaries. A normalized intervention trace further tracks whether runtime support is available, activated, delivered, adopted by the agent, and propagated into subsequent behavior. We instantiate the protocol as MedHarnessBench, comprising 342 executable tasks across FHIR/EHR operations, healthcare-administration interfaces, and multimodal clinical tools. Across more than 16,000 controlled task executions, MedHarness yields broad improvements in terminal task success, intermediate subtask completion, and healthcare-specific process quality. We further evaluate MedHarness against multiple general-purpose harnesses and use controlled single-fault perturbations to verify the sensitivity and discrimination of the seven-dimensional evaluation. These results demonstrate that healthcare-specific runtime support can substantially improve both task completion and execution quality, while revealing failures that terminal outcomes alone cannot capture. Code and resources are available at https://github.com/Tomry1114/Awesome-Agent-Harness-for-Healthcare.

Global healthcare providers are exploring use of large language models (LLMs) to provide medical advice to the public. LLMs now achieve nearly perfect scores on medical licensing exams, but this does not necessarily translate to accurate performance in real-world settings. We tested if LLMs can assist members of the public in identifying underlying conditions and choosing a course of action (disposition) in ten medical scenarios in a controlled study with 1,298 participants. Participants were randomly assigned to receive assistance from an LLM (GPT-4o, Llama 3, Command R+) or a source of their choice (control). Tested alone, LLMs complete the scenarios accurately, correctly identifying conditions in 94.9% of cases and disposition in 56.3% on average. However, participants using the same LLMs identified relevant conditions in less than 34.5% of cases and disposition in less than 44.2%, both no better than the control group. We identify user interactions as a challenge to the deployment of LLMs for medical advice. Standard benchmarks for medical knowledge and simulated patient interactions do not predict the failures we find with human participants. Moving forward, we recommend systematic human user testing to evaluate interactive capabilities prior to public deployments in healthcare.

Adding procedural skills to an LLM agent is typically evaluated by average improvement in task success. However, this metric hides an important cost: skills can also make agents worse. We measure both sides by comparing agents with and without skills across nearly 6,000 runs spanning two office automation benchmarks and three model harness stacks. This allows us to distinguish two outcomes. A regression is a task solved without skills but failed after skills are added. A residual failure is a task that fails both with and without skills. We find that regressions are substantial enough that the best performing skills outperform others primarily by regressing less, not by gaining more. We identify three causes of regression: (i) skill description osmosis, a skill changes an agent's behavior simply by being present in context, even when it is never invoked; (ii) grounding displacement, a skill's prescribed procedure overrides how the agent interprets its inputs; and (iii) verification displacement, where the procedure suppresses checks the agent would otherwise perform on its outputs. Analysing persistent failures reveals the same underlying pattern. Existing skills overemphasize procedural guidance the stage least often responsible for failure while under supporting grounding and verification, the dominant sources of remaining errors. After correcting evaluation artifacts and studying traces, we find many regressions and persistent failures recoverable through better grounding and verification. Procedural skills should be evaluated by decomposing their net effect into gains and regressions, not by aggregate improvement alone. We identify three regression modes skills should avoid, and find that reliability depends more on grounding and verification than on procedural skill choice.

Large language models (LLMs) increasingly act as autonomous agents, using tools to execute code, read and write files, and access networks, creating novel security risks. To mitigate these risks, agents are commonly deployed and evaluated in isolated "sandbox" environments, often implemented using Docker/OCI containers. We introduce SANDBOXESCAPEBENCH, an open benchmark that safely measures an LLM's capacity to break out of these sandboxes. The benchmark is implemented as an Inspect AI Capture the Flag (CTF) evaluation utilising a nested sandbox architecture with the outer layer containing the flag and no known vulnerabilities. Following a threat model of a motivated adversarial agent with shell access inside a container, SANDBOXESCAPEBENCH covers a spectrum of sandboxescape mechanisms spanning misconfiguration, privilege allocation mistakes, kernel flaws, and runtime/orchestration weaknesses. We find that, when vulnerabilities are added, LLMs are able to identify and exploit them, showing that use of evaluation like SANDBOXESCAPEBENCH is needed to ensure sandboxing continues to provide the encapsulation needed for highly-capable models.

Most medical multimodal benchmarks focus on static tasks such as image question answering, report generation, and plain-language rewriting. Patient education is more demanding: systems must identify relevant evidence across images, show patients where to look, explain findings in accessible language, and handle confusion or distress. Yet most patient education work remains text-only, even though combined image-and-text explanations may better support understanding. We introduce MedImageEdu, a benchmark for multi-turn, evidence-grounded radiology patient education. Each case provides a radiology report with report text and case images. A DoctorAgent interacts with a PatientAgent, conditioned on a hidden profile that captures factors such as education level, health literacy, and personality. When a patient question would benefit from visual support, the DoctorAgent can issue drawing instructions grounded in the report, case images, and the current question to a benchmark-provided drawing tool. The tool returns image(s), after which the DoctorAgent produces a final multimodal response consisting of the image(s) and a grounded plain-language explanation. MedImageEdu contains 150 cases from three sources and evaluates both the consultation process and the final multimodal response along five dimensions: Consultation, Safety and Scope, Language Quality, Drawing Quality, and Image-Text Response Quality. Across representative open- and closed-source vision-language model agents, we find three consistent gaps: fluent language often outpaces faithful visual grounding, safety is the weakest dimension across disease categories, and emotionally tense interactions are harder than low education or low health literacy. MedImageEdu provides a controlled testbed for assessing whether multimodal agents can teach from evidence rather than merely answer from text.

Keyword-matching benchmarks can credit small models for tool use they never perform. We document such a false positive in a matched-architecture pair of Spanish security language models and propose a ladder of strict, cheap diagnostics. A 661.6M parameter model (approx. 65% code/technical text; no dedicated SFT) and a 1,109M model (web-heavy multi-phase curriculum; 6B-token tool-SFT) share decoder, tokenizer, and special tokens, scoring almost identically on lenient tool-use metrics (B4: 0.660 vs. 0.650). Verbatim-reproduction checks on training examples separate them completely: the 600M emits valid tool calls with generalized arguments on 6/6 examples; the 1B does so on 0/6 across checkpoints. A first-token probe localizes the 1B's failure to a missing prior (prob. 10^{-4}--10^{-5} on <|tool_call|>), which was erased by its web-heavy training phase. A targeted SFT recipe (diverse corpus, 5x higher learning rate, 2,202 steps, ~3.3 GPU-hours) repairs the 1B using three orders of magnitude fewer tokens than the failed phase. On all 269 corpus rows, valid emission rises from 0.100 to 0.959 (600M: 0.926). On 238 unseen prompts, the repaired 1B passes 0.536 vs. the 600M's 0.428 (p = 0.004). Embedding-drift checks show the repair did not move the trigger token's tied embedding (97.7% of the bf16 table remains bit-identical), meaning changes live in the surrounding network. Both models over-trigger, rarely answering negative prompts without a call (0.09 for 600M, 0.17 for repaired 1B). Factorial analyses confirm all repair configurations install the format, though suppression benefits from a diverse corpus remain a hypothesis due to seed sensitivity. This cheap diagnostic ladder costs minutes of CPU time and should gate tool-use claims on small models.

The increasing deployment of AI agents in long-horizon tasks yields massive execution logs. Diagnosing failures within these records is crucial for reliability, as it transforms outcome-level signals into actionable interventions. The sheer scale of the data renders human review impractical, driving the need for automated root-cause attribution (RCA). However, automated RCA methods using LLMs suffer from low diagnostic accuracy, especially as execution traces grow larger. They struggle because relevant information is often sparse, distributed across distant actions, and disconnected from the visible failure, reducing root-cause attribution to a massive search problem. Existing RCA methods typically rely on one-shot LLM judgments to diagnose failures from execution traces. While effective for shorter trajectories, these judges tend to settle on a plausible diagnosis early, leaving critical evidence in longer traces unexamined. We introduce Continual Search, an iterative framework that nudges the judge, over successive turns, to keep searching for unresolved diagnostic evidence. We evaluate Continual Search across four existing RCA benchmarks. Recognizing the lack of massive execution traces in current benchmarks, we introduce MegaRCA-Mix to evaluate RCA at scale. MegaRCA-Mix provides a challenging testbed of 50 human-annotated failure trials spanning long-horizon, execution-heavy tasks. Across multiple benchmark suites and model families, Continual Search consistently improves attribution performance. On MegaRCA-Mix, for example, it improves GPT-5.5's F1 score by more than 40\%, from 0.349 to 0.498. More interestingly, within the same model family, lower-tier models can even surpass their higher-tier counterparts, demonstrating that effective search supersedes raw model scale.

Psychotherapists need repeated training and supervision by experts; however, scalability is problematic. Here we present MyMentorLLM, a multimodal voice- and text-based simulation environment for deliberate practice, used to generate 2,100 complete Cognitive Behavioural Therapy (CBT) training sessions. Each session links a DSM-5-TR-grounded patient (with major depressive, generalised anxiety or borderline personality disorder), a therapist-in-training and an expert supervisor. As an initial implementation, we adopted CBT because its structured procedures and competency-based supervision facilitate standardised simulation and evaluation. Sessions were analysed for emotional dynamics, therapeutic competence and diagnostic accuracy. Simulated patients expressed disorder-congruent emotional profiles, which trainee therapists mirrored as in real human counselling. The quality of supervision differed across LLMs: while most models overestimated trainees' competences, native speech-to-speech was closest to human scores. Supervisors' feedback led to better diagnoses in simulated psychotherapists in 5 out of 7 LLMs, and symptom identification accuracy increased with model size. This work shows that simulation of deliberate practice is possible for CBT training, although patient fidelity, calibration of supervisors, and harmful feedback should be evaluated together.

Many math- and science-oriented agent systems use hierarchical designs with specialized reviewer roles, assuming that a dedicated review stage should help turn wrong candidates into correct ones. We test this assumption on 4,181 verifier-grounded Omni-MATH problems using matched gpt-oss-120b actors. Collaboration adds little on the easiest tiers, but from tier 4 onward the gains open sharply; in this harder regime, broadcast-style peer discussion reaches higher final accuracy than a planner-executor-reviewer pipeline (PER). We ask whether this gap is explained by reviewer quality or by whether critique changes the next answer the protocol carries forward. It is not explained by reviewer precision alone: PER's reviewer is more precise than broadcast's (0.861 vs. 0.644), yet evaluator-verified useful critique is much less likely to change the next candidate and produces lower reviewer-guided repair. These results show that reviewer detection quality and critique uptake are empirically separable. Within matched PER interventions, forcing explicit acknowledgment lowers final accuracy, while embedding reviewer guidance directly in the solver's working context partially improves follow-through without closing the gap. Overall, reviewer-centric evaluation can overstate system quality: a protocol may spot errors well yet still fail to solve more problems if it does not act on those critiques.

Scientific Deep Research (DR) agents answer user queries by synthesizing research papers into multi-section reports. User feedback can improve their utility, but existing protocols only score the final report, making it hard to study and learn which intermediate actions DR agents should take to improve reports. We collect DRACULA, the first dataset with user feedback on intermediate actions for DR. Over five weeks, nineteen expert CS researchers ask queries to a DR system that proposes actions (e.g., "Add a section on datasets"). Our users select actions they prefer, then judge whether an output report applied their selections successfully, yielding 8,103 action preferences and 5,230 execution judgments. After confirming a DR agent can execute DRACULA's actions, we study the predictability of user-preferred actions via simulation-how well LLMs predict the actions users select-a step toward learning to generate useful actions. We discover: (1) LLM judges initially struggle to predict action selections, but improve most when using a user's full selection history, rather than self-reported or extrapolated user context signals; (2) Users' selections for the same query differ based on unstated goals, bottlenecking simulation and motivating affordances that let users steer reports; and (3) Our simulation results inform an online intervention that generates new actions based on the user's past interactions, which users pick most often in follow-up studies. Overall, while work extensively studies execution, DRACULA reveals a key challenge is deciding which actions to execute in the first place. We open-source DRACULA's study design, user feedback, and simulation tasks to spur future work on action feedback for long-horizon agents.

In medical education, physicians convert academic knowledge into clinical expertise through residency: years of training across thousands of encounters, with diverse sources of feedback and progressively greater autonomy. Much of clinical reasoning relies on the patient encounter, a dialogue in which a clinician elicits history, refines diagnostic hypotheses, and decides management under uncertainty. While large language models (LLMs) excel on static medical benchmarks, methods to optimize the full sequence of clinical decisions remain underdeveloped. We present ResidencyRL, a reinforcement learning (RL) method for training clinical artificial intelligence (AI) agents through simulated multi-turn clinical encounters (up to 60 dialogue turns and 8 tool calls per trajectory). ResidencyRL pairs the policy agent with LLM simulators capable of complex, adversarial behaviors, training against a structured reward aligned to diagnostic accuracy, management quality, communication, documentation, and safety. On held-out evaluations, the ResidencyRL agent improves diagnostic accuracy by 7.0% under adversarial conditions (88.0% vs. 81.0%) and reduces missed red flag rates by 31%, demonstrating rigorous mitigation of premature closure. Blinded expert clinicians validated these gains, preferring the trained agent in 87.6% of side-by-side comparisons. The procedural competencies transfer to unseen benchmarks: the agent outperforms the base model across all six clinical axes of the AMIE multi-visit benchmark, and shows consistent directional improvements on AgentClinic and CRAFT-MD. Our findings demonstrate that sequential clinical decision-making can be effectively learned through multi-turn RL in simulation, yielding robust, generalizable capabilities, paving the way towards clinical mastery. Prospective validation with real-world workflows remains necessary to establish clinical utility.

Users typically engage with LLMs interactively, yet most existing benchmarks evaluate them in a static, single-turn format, posing reliability concerns in interactive scenarios. We identify a key obstacle towards reliability: LLMs are trained to answer any question, even with incomplete context or insufficient knowledge. In this paper, we propose to change the static paradigm to an interactive one, develop systems that proactively ask questions to gather more information and respond reliably, and introduce an benchmark - MediQ - to evaluate question-asking ability in LLMs. MediQ simulates clinical interactions consisting of a Patient System and an adaptive Expert System; with potentially incomplete initial information, the Expert refrains from making diagnostic decisions when unconfident, and instead elicits missing details via follow-up questions. We provide a pipeline to convert single-turn medical benchmarks into an interactive format. Our results show that directly prompting state-of-the-art LLMs to ask questions degrades performance, indicating that adapting LLMs to proactive information-seeking settings is nontrivial. We experiment with abstention strategies to better estimate model confidence and decide when to ask questions, improving diagnostic accuracy by 22.3%; however, performance still lags compared to an (unrealistic in practice) upper bound with complete information upfront. Further analyses show improved interactive performance with filtering irrelevant contexts and reformatting conversations. Overall, we introduce a novel problem towards LLM reliability, an interactive MediQ benchmark and a novel question-asking system, and highlight directions to extend LLMs' information-seeking abilities in critical domains.

Background: Agent skills are increasingly deployed as modular, reusable capability units in AI agent systems. Medical research agent skills require safeguards beyond general-purpose evaluation, including scientific integrity, methodological validity, reproducibility, and boundary safety. This study developed and preliminarily evaluated a domain-specific audit framework for medical research agent skills, with a focus on reliability against expert review. Methods: We developed MedSkillAudit (skill-auditor@1.0), a layered framework assessing skill release readiness before deployment. We evaluated 75 skills across five medical research categories (15 per category). Two experts independently assigned a quality score (0-100), an ordinal release disposition (Production Ready / Limited Release / Beta Only / Reject), and a high-risk failure flag. System-expert agreement was quantified using ICC(2,1) and linearly weighted Cohen's kappa, benchmarked against the human inter-rater baseline. Results: The mean consensus quality score was 72.4 (SD = 13.0); 57.3% of skills fell below the Limited Release threshold. MedSkillAudit achieved ICC(2,1) = 0.449 (95% CI: 0.250-0.610), exceeding the human inter-rater ICC of 0.300. System-consensus score divergence (SD = 9.5) was smaller than inter-expert divergence (SD = 12.4), with no directional bias (Wilcoxon p = 0.613). Protocol Design showed the strongest category-level agreement (ICC = 0.551); Academic Writing showed a negative ICC (-0.567), reflecting a structural rubric-expert mismatch. Conclusions: Domain-specific pre-deployment audit may provide a practical foundation for governing medical research agent skills, complementing general-purpose quality checks with structured audit workflows tailored to scientific use cases.

We study auto research as a closed empirical loop driven by external measurement. Each submitted trial carries a hypothesis, an executable code edit, an evaluator-owned outcome, and feedback that shapes the next proposal. The output is not a generated paper or a single model checkpoint, but an auditable trajectory of proposals, code diffs, experiments, scores, and failure labels. We instantiate this loop with specialist agents that partition recipe surfaces and share measured lineage across trials. The central empirical finding is that lineage feedback lets agents turn evaluator outcomes, including crashes, budget overruns, size failures, and accuracy-gate misses, into later program-level recipe edits rather than one-shot suggestions. Across 1,197 headline-run trials plus 600 Parameter Golf control trials after one-time setup and launch, humans did not choose proposals, edit recipes, override scores, or repair failed trials during the search. In the three headline runs, the same submitted-trial loop reduces Parameter Golf validation bpb by 0.81%, raises NanoChat-D12 CORE by 38.7%, and reduces CIFAR-10 Airbench96 wallclock by 4.59%, with each task measured by its own external evaluator and legality checks. The trace includes a strict architecture-domain audit of 157 headline-run submissions and program rewrites such as a NanoChat attention-kernel path change. Within this scope the loop autonomously writes code, submits experiments, absorbs feedback, applies and combines known techniques inside each environment, and improves public starting recipes.

We introduce MPLSandbox, an out-of-the-box multi-programming language sandbox designed to provide unified and comprehensive feedback from compiler and analysis tools for Large Language Models (LLMs). It can automatically identify the programming language of the code, compiling and executing it within an isolated sub-sandbox to ensure safety and stability. In addition, MPLSandbox also integrates both traditional and LLM-based code analysis tools, providing a comprehensive analysis of generated code. MPLSandbox can be effortlessly integrated into the training and deployment of LLMs to improve the quality and correctness of their generated code. It also helps researchers streamline their workflows for various LLM-based code-related tasks, reducing the development cost. To validate the effectiveness of MPLSandbox, we integrate it into training and deployment approaches, and also employ it to optimize workflows for a wide range of real-world code-related tasks. Our goal is to enhance researcher productivity on LLM-based code-related tasks by simplifying and automating workflows through delegation to MPLSandbox.

Foundation models can serve as clinical agents through tool-use harnesses. However, conventional medical benchmarks assess reasoning over preselected evidence rather than the ability to seek it across clinical records and longitudinal imaging. We propose CASE: a series of role-specific Clinical Agents for Seeking Evidence, together with a tool-use harness and an agentic post-training framework for compact vision-language policy models. We further introduce a longitudinal multimodal benchmark built on UK Biobank, comprising 50,401 clinical questions derived from real-world ICD-10-coded diagnoses of 4,739 participants. Each question links to a patient-specific environment containing clinical context and multi-sequence MRI from baseline and follow-up visits, where agents autonomously select which visits, organs, modalities, slices, and specialist tools to inspect and compare. Supervised fine-tuning transfers evidence-seeking workflows from 14,734 frontier-model interaction trajectories, followed by agentic reinforcement learning on the learner's own environment interactions. Privileged on-policy self-distillation and rubric-based LLM feedback refine evidence-to-conclusion reasoning without prescribing tool sequences. Experiments show that CASE moves beyond question-answer imitation toward transferable investigation policies, strengthening evidence-grounded longitudinal reasoning. Under matched evaluation conditions, our Qwen3-VL-8B based agent achieves over 16% and 10% relative improvements in answer accuracy over GPT-5.4 and Claude Opus 4.8. Code will be available at https://github.com/VinyehShaw/CASE.

Eliciting detailed and actionable software requirements from end-users is a critical phase in the iterative development of a software product or application. To ensure the feedback collected is detailed and actionable, software teams can leverage the laddering interview technique. While effective for ensuring granular and actionable items from the software feedback, these interviews are subject to several limitations. They are traditionally a manual process associated with a time and financial burden, limiting scalability; interviewers must balance probing for depth while managing interviewee behavioral and cultural constraints. To address these limitations, we present LadderTeam, an open, reproducible framework that automates UX wireframe interviews using a dual-agent Large Language Model (LLM) architecture. An active interviewer agent executes one of three probing strategies (ACV, 5-Whys, and JTBD) to elicit actionable software requirements from usability feedback comments, while a concurrent background Judge agent evaluates probe-response pairs and triggers real-time guardrails to prevent topic drift. To rigorously evaluate LLM laddering without participant variance confounds, we introduce a controlled simulation methodology utilizing scripted ground-truth transcripts to isolate probe quality as the sole experimental variable. Across 216 interviews, LadderTeam achieved 99.1\% chain convergence and an 81.0\% ground-truth actionable response match (86.1\% reluctant personality, 75.9\% terse personality) with zero drift across all runs. All evaluation code, all transcripts, inputs, and a live demonstration platform will be open-sourced upon acceptance.

Traditional metrics for Medical Report Generation (MRG) predominantly rely on surface-level n-gram overlap, which fails to capture clinical factual accuracy and often overlooks catastrophic diagnostic errors. We address this fundamental limitation by proposing AtomiMed, a universal, modality-agnostic evaluation framework that decomposes complex medical narratives into a standardized, multi-level hierarchy of Atomic Clinical Facts, encompassing Disease-level entities and Attribute-level descriptors, including location, morphology, and severity. By implementing an Agentic Cross-Verification loop between ground-truth and predicted reports, AtomiMed simulates a multi-radiologist peer-review process to verify clinical consistency, thus enabling the decoupled assessment of diagnostic detection and descriptive accuracy. To facilitate standardized evaluation, we introduce MRGEvalKit, an open-source toolkit for automated hierarchical extraction, and curate OmniMRG-Bench, a comprehensive multi-modal benchmark covering X-ray, CT, MRI, and Ultrasound. Extensive experiments on multiple expert-annotated reader studies demonstrate that AtomiMed achieves significantly higher correlation with human radiologist judgment compared to traditional and model-based metrics. Our code are release at https://github.com/Venn2336/MRGEvalkit

Long-horizon agents are beginning to automate complete workflows that produce code, reports, and research artifacts. Medical imaging workflows are multi-stage and data-sensitive, while expert trajectories remain scarce and difficult to share. Structured benchmarks can localize failures through stage-level rubrics, but standard post-training discards these diagnostics before the next training round. We present Benchmark-as-Teacher (BaT), a recursive self-improvement system for agent post-training. BaT contains two linked components: the asynchronous Stage Bank data pipeline and BiCuRL (Bilevel Curriculum Reinforcement Learning), its self-improving post-training method. Stage Bank synthesizes content-isolated training states outside the policy-update loop. BiCuRL uses a fixed held-out evaluation to select the next stage curriculum, verifies rollouts with task rubrics, updates the policy with GRPO, and returns the candidate checkpoint to evaluation. On AutoMedBench-Lite, BaT-4B and BaT-9B more than double the Overall scores of their Qwen Instruct baselines. BaT-9B Agent reaches 79.6 Overall, exceeding Claude Opus 4.6 with Claude Code at 77.5.

This paper presents our team's participation in the MEDIQA-ClinicalNLP2024 shared task B. We present a novel approach to diagnosing clinical dermatology cases by integrating large multimodal models, specifically leveraging the capabilities of GPT-4V under a retriever and a re-ranker framework. Our investigation reveals that GPT-4V, when used as a retrieval agent, can accurately retrieve the correct skin condition 85% of the time using dermatological images and brief patient histories. Additionally, we empirically show that Naive Chain-of-Thought (CoT) works well for retrieval while Medical Guidelines Grounded CoT is required for accurate dermatological diagnosis. Further, we introduce a Multi-Agent Conversation (MAC) framework and show its superior performance and potential over the best CoT strategy. The experiments suggest that using naive CoT for retrieval and multi-agent conversation for critique-based diagnosis, GPT-4V can lead to an early and accurate diagnosis of dermatological conditions. The implications of this work extend to improving diagnostic workflows, supporting dermatological education, and enhancing patient care by providing a scalable, accessible, and accurate diagnostic tool.

Chest X-ray plays a central role in thoracic diagnosis, and its interpretation inherently requires multi-step, evidence-grounded reasoning. However, large vision-language models (LVLMs) often generate plausible responses that are not faithfully grounded in diagnostic evidence and provide limited visual evidence for verification, while also requiring costly retraining to support new diagnostic tasks, limiting their reliability and adaptability in clinical settings. To address these limitations, we present CXReasonAgent, a diagnostic agent that integrates a large language model (LLM) with clinically grounded diagnostic tools to perform evidence-grounded diagnostic reasoning using image-derived diagnostic and visual evidence. To evaluate these capabilities, we introduce CXReasonDial, a multi-turn dialogue benchmark with 1,946 dialogues across 12 diagnostic tasks, and show that CXReasonAgent produces faithfully grounded responses, enabling more reliable and verifiable diagnostic reasoning than LVLMs. These findings highlight the importance of integrating clinically grounded diagnostic tools, particularly in safety-critical clinical settings.

Artificial intelligence has significantly advanced healthcare, particularly through large language models (LLMs) that excel in medical question answering benchmarks. However, their real-world clinical application remains limited due to the complexities of doctor-patient interactions. To address this, we introduce AI Hospital, a multi-agent framework simulating dynamic medical interactions between Doctor as player and NPCs including Patient, Examiner, Chief Physician. This setup allows for realistic assessments of LLMs in clinical scenarios. We develop the Multi-View Medical Evaluation (MVME) benchmark, utilizing high-quality Chinese medical records and NPCs to evaluate LLMs' performance in symptom collection, examination recommendations, and diagnoses. Additionally, a dispute resolution collaborative mechanism is proposed to enhance diagnostic accuracy through iterative discussions. Despite improvements, current LLMs exhibit significant performance gaps in multi-turn interactions compared to one-step approaches. Our findings highlight the need for further research to bridge these gaps and improve LLMs' clinical diagnostic capabilities. Our data, code, and experimental results are all open-sourced at https://github.com/LibertFan/AI_Hospital.

Code editing is an essential step towards reliable program synthesis to automatically correct critical errors generated from code LLMs. Recent studies have demonstrated that closed-source LLMs (i.e., ChatGPT and GPT-4) are capable of generating corrective feedback to edit erroneous inputs. However, it remains challenging for open-source code LLMs to generate feedback for code editing, since these models tend to adhere to the superficial formats of feedback and provide feedback with misleading information. Hence, the focus of our work is to leverage open-source code LLMs to generate helpful feedback with correct guidance for code editing. To this end, we present Coffee, a collected dataset specifically designed for code fixing with feedback. Using this dataset, we construct CoffeePots, a framework for COde Fixing with FEEdback via Preference-Optimized Tuning and Selection. The proposed framework aims to automatically generate helpful feedback for code editing while minimizing the potential risk of superficial feedback. The combination of Coffee and CoffeePots marks a significant advancement, achieving state-of-the-art performance on HumanEvalFix benchmark. Codes and model checkpoints are publicly available at https://github.com/Lune-Blue/COFFEE.

Proactive agents must decide whether, when, and how far to assist without an explicit request. Existing evaluations often rely on automated scores or binary verdicts, offering limited human diagnosis of evidence use and intervention quality. To address these gaps, we introduce HALO, Human-Grounded GrAnular EvaLuation of PrOactive Agents. HALO-Arena covers 4 proactive need types, Dependency, Blockage, Risk, and Opportunity, across 10 everyday domains and 50 personal environments. Scenarios require agents to retrieve evidence from structured databases or unstructured documents before judging whether assistance is warranted and choosing its scope. On HALO-Arena, we construct HALO-8K, a dataset of 8,000 responses from 10 models across 400 scenarios, annotated by 20 experts for action level, acceptability verdicts, and human diagnosis. For rejected responses, human diagnosis covers timing, intensity, and other issues. Recorded reasoning tokens and retrieval rounds measure the effort behind these responses. The best proprietary and open-source models achieve 72.88% and 56.63% acceptability, respectively. Our central finding is that over-proactivity is the dominant observed failure mode, with Premature timing or Excessive intensity accounting for 61.20% of rejected responses. Further analyses yield complementary findings on the balance between intervention and restraint, evidence acquisition, and the calibration of timing and intensity.

Large Language Models (LLMs) are increasingly serving as autonomous agents, and their utilization of external tools via the Model Context Protocol (MCP) is considered a future trend. Current MCP evaluation sets suffer from issues such as reliance on external MCP services and a lack of difficulty awareness. To address these limitations, we propose MCPAgentBench, a benchmark based on real-world MCP definitions designed to evaluate the tool-use capabilities of agents. We construct a dataset containing authentic tasks and simulated MCP tools. The evaluation employs a dynamic sandbox environment that presents agents with candidate tool lists containing distractors, thereby testing their tool selection and discrimination abilities. Furthermore, we introduce comprehensive metrics to measure both task completion rates and execution efficiency. Experiments conducted on various latest mainstream Large Language Models reveal significant performance differences in handling complex, multi-step tool invocations. All code is open-source at Github.

Large language models (LLMs) are increasingly proposed as clinical agents, yet static, single-turn benchmarks cannot capture how a model dynamically delivers care across an encounter: gathering information, planning treatment, and adapting longitudinal management across successive patient states. Medical education has long addressed an analogous challenge through standardized patients (SPs): trained actors who consistently portray clinical cases, enabling realistic practice and objective, scripted assessment. Here we introduce MedSP1000, an SP-derived interactive benchmark for clinical-agent evaluation, including 1,638 SP cases with 24,602 trajectory-level peer-reviewed rubrics. MedSP1000 converts peer-reviewed SP teaching cases into executable scenarios with defined SP case scripts, clinical environment contexts, and human-validated structured rubric. In each simulation evaluation run, a clinical agent interacts in closed loop with a patient agent and an environment controller, and its behaviour is scored throughout the encounter against expert criteria specified in the original materials. Applying MedSP1000 to a range of general-purpose and medically specialized LLMs, we find that performance on static benchmarks does not reliably translate to such educational scenarios. The best-performing model, GPT-5.5, completes only 60.4% of expert-defined rubric items, whereas the strongest medically specialized model reaches 40.0%; increasing test-time compute produces no measurable gain. These results suggest that current LLMs, including agentic systems tuned for medicine, are not yet reliable enough to be safely integrated into actual clinical practice. More broadly, MedSP1000 shows how process-level, SP-style evaluation can reveal clinically relevant failure modes that single-turn benchmarks miss.

Large language models (LLMs) have become increasingly capable of following instructions and complex reasoning, making prompting a flexible interface for adapting models without parameter updates. Yet prompt design remains labor-intensive and highly sensitive to formatting, phrasing, and instruction order, motivating automated prompt optimization methods that reduce manual effort while preserving inference-time flexibility. However, existing methods often search over prompt candidates or use fixed critique-refine pipelines driven by individual examples or small batches, limiting their ability to capture systematic error patterns and make targeted edits grounded in failure history. We propose Reflective Prompt Tuning (RPT), a framework that uses LLM function calling to simulate the iterative workflow of human prompt engineers. An LLM optimizer calls a diagnostic function that evaluates the target model over an entire optimization set, summarizes recurring failure modes, and returns a structured diagnostic report. The optimizer uses this report, together with an accumulated memory of prior reports, to revise the prompt for the next iteration. RPT further supports confidence-aware optimization by using calibration signals in diagnostic feedback and final prompt selection. Across three reasoning tasks, RPT improves over initial prompts by up to 12.9 points, remains competitive with state of the art, and improves confidence calibration. Our analyses show that RPT is especially effective on multi-hop and mathematical reasoning, producing targeted prompt revisions that align with diagnosed failure patterns and lead to gains in task performance and calibration.

Objectives: While Large Language Models (LLMs) have been widely used to assist clinicians and support patients, no existing work has explored dialogue systems for standard diagnostic interviews and assessments. This study aims to bridge the gap in mental healthcare accessibility by developing an LLM-powered dialogue system that replicates clinician behavior. Materials and Methods: We introduce TRUST, a framework of cooperative LLM modules capable of conducting formal diagnostic interviews and assessments for Post-Traumatic Stress Disorder (PTSD). To guide the generation of appropriate clinical responses, we propose a Dialogue Acts schema specifically designed for clinical interviews. Additionally, we develop a patient simulation approach based on real-life interview transcripts to replace time-consuming and costly manual testing by clinicians. Results: A comprehensive set of evaluation metrics is designed to assess the dialogue system from both the agent and patient simulation perspectives. Expert evaluations by conversation and clinical specialists show that TRUST performs comparably to real-life clinical interviews. Discussion: Our system performs at the level of average clinicians, with room for future enhancements in communication styles and response appropriateness. Conclusions: Our TRUST framework shows its potential to facilitate mental healthcare availability.

Optimizing LLM-based agentic workflows is challenging for scaling AI capabilities. Current methods rely on coarse, end-to-end evaluation signals and lack fine-grained signals on where to refine, often resulting in inefficient or low-impact modifications. To address these limitations, we propose JudgeFlow, an Evaluation-Judge-Optimization-Update pipeline. We incorporate reusable, configurable logic blocks into agentic workflows to capture fundamental forms of logic. On top of this abstraction, we design a dedicated Judge module that inspects execution traces particularly failed runs and assigns rank-based responsibility scores to problematic blocks. These fine-grained diagnostic signals are then leveraged by an LLM-based optimizer, which focuses modifications on the most problematic block in the workflow. Our approach improves sample efficiency, enhances interpretability through block-level diagnostics, and provides a scalable foundation for automating increasingly complex agentic workflows. We evaluate JudgeFlow on mathematical reasoning and code generation benchmarks, where JudgeFlow achieves superior performance and efficiency compared to existing methods.

Proactive interventions by LLM critic models are often assumed to improve reliability, yet their effects at deployment time are poorly understood. We show that a binary LLM critic with strong offline accuracy (AUROC 0.94) can nevertheless cause severe performance degradation, inducing a 26 percentage point (pp) collapse on one model while affecting another by near zero pp. This variability demonstrates that LLM critic accuracy alone is insufficient to determine whether intervention is safe. We identify a disruption-recovery tradeoff: interventions may recover failing trajectories but also disrupt trajectories that would have succeeded. Based on this insight, we propose a pre-deployment test that uses a small pilot of 50 tasks to estimate whether intervention is likely to help or harm, without requiring full deployment. Across benchmarks, the test correctly anticipates outcomes: intervention degrades performance on high-success tasks (0 to -26 pp), while yielding a modest improvement on the high-failure ALFWorld benchmark (+2.8 pp, p=0.014). The primary value of our framework is therefore identifying when not to intervene, preventing severe regressions before deployment.

Large Multimodal Models (LMMs) have shown remarkable progress in the field of medical Visual Question Answering (Med-VQA), achieving high accuracy on existing benchmarks. However, their reliability under robust evaluation is questionable. This study reveals that state-of-the-art models, when subjected to simple probing evaluation, perform worse than random guessing on medical diagnosis questions. To address this critical evaluation problem, we introduce the Probing Evaluation for Medical Diagnosis (ProbMed) dataset to rigorously assess LMM performance in medical imaging through probing evaluation and procedural diagnosis. Particularly, probing evaluation features pairing original questions with negation questions with hallucinated attributes, while procedural diagnosis requires reasoning across various diagnostic dimensions for each image, including modality recognition, organ identification, clinical findings, abnormalities, and positional grounding. Our evaluation reveals that top-performing models like GPT-4V and Gemini Pro perform worse than random guessing on specialized diagnostic questions, indicating significant limitations in handling fine-grained medical inquiries. Besides, models like LLaVA-Med struggle even with more general questions, and results from CheXagent demonstrate the transferability of expertise across different modalities of the same organ, showing that specialized domain knowledge is still crucial for improving performance. This study underscores the urgent need for more robust evaluation to ensure the reliability of LMMs in critical fields like medical diagnosis, and current LMMs are still far from applicable to those fields.

Agent Skills package reusable instructions and assets for tool-using language-model agents. Progressive loading creates failure boundaries poorly represented by session-, model-, or tool-centric traces: a Skill can be discovered but not activated, activated without instructions, or appear successful without an independently verified outcome. We present Skill Runtime Intelligence, a passive runtime-intelligence system that reconstructs supported Skill-lifecycle stages across heterogeneous harnesses while preserving unsupported stages as unknown. Its Run Panorama separates immutable events, deterministic relations, inferred diagnoses, and controlled outcomes with four evidence grades; optional trace import and OTLP/HTTP export support existing observability deployments. Across six frozen repository profiles, three coding agents, and seven clean or fault-injected conditions, all 126 executions preserve source worktrees and each correlates to exactly one source session. Yet adapters expose three distinct semantics: no Skill runs; complete runs but no failure-like events; or failure-like events in every operational-failure and clean session. In a seven-template diagnostic study, semantic aliases and Panorama localize the same six non-clean boundaries but differ in exact/status behavior; both Raw views emit a failure status on all 18 clean cases, while Panorama emits none. A known-rule graph conforms to 126/126 frozen contracts, whereas a second model completes only 228/378 calls. These observations motivate executable adapter qualification and show that event presence is not boundary fidelity, composite exact scores mask distinct errors, and model explanations must not overwrite deterministic facts.

Tool-augmented LLM agents raise new security risks: tool executions can introduce runtime-only behaviors, including prompt injection and unintended exposure of external inputs (e.g., environment secrets or local files). While existing scanners often focus on static artifacts, analyzing runtime behavior is challenging because directly executing untrusted tools can itself be dangerous. We present MCP-SandboxScan, a lightweight framework motivated by the Model Context Protocol (MCP) that safely executes untrusted tools inside a WebAssembly/WASI sandbox and produces auditable reports of external-to-sink exposures. Our prototype (i) extracts LLM-relevant sinks from runtime outputs (prompt/messages and structured tool-return fields), (ii) instantiates external-input candidates from environment values, mounted file contents, and output-surfaced HTTP fetch intents, and (iii) links sources to sinks via snippet-based substring matching. Case studies on three representative tools show that MCP-SandboxScan can surface provenance evidence when external inputs appear in prompt/messages or tool-return payloads, and can expose filesystem capability violations as runtime evidence. We further compare against a lightweight static string-signature baseline and use a micro-benchmark to characterize false negatives under transformations and false positives from short-token collisions.

Making an open-weight language model generate text is not the same as making it finish coding work. This retrospective empirical systems experience report follows my engineering work from July to October 2026 across coding clients, model-specific serving recipes, vLLM, SGLang, llama.cpp, and MLX. Models behind a nominally compatible endpoint still differed in tool-call formatting, turn control, context behavior, cache reuse, and reliability. A served model, coherent text, executed tool calls, and code that passes independent behavioral checks proved to be distinct levels of progress. The evidence includes 4,165 coding-agent gateway turns from my own sessions, 623 debug jobs, and three separate delegation studies with hidden behavioral checks and a fixed Claude Opus 5.5 parent. Saved cold-prefix requests had a median time to first token of about 69.4 s, versus 2.8 s for high-reuse requests (observational). In a 45-trial study, direct coding passed 15/15 and delegation to a quantized Qwen3-Coder 30B-A3B worker passed 12/15 and 14/15. On a second 15-task suite under a fixed artifact contract, the completed replacement cohort had 45/45 for direct coding and 12/45 for a Nemotron Cascade 2 worker (11 worker substitutions). Sixteen sessions stopped by a subscription limit before any model call were replaced; counting them as failures gives 37/45 and 8/45. A Qwen3-Coder worker passed 0/45. In both studies with a direct arm, direct coding had lower API list-price-equivalent frontier cost per verified success. A post-hoc diagnostic found that 19 of 36 artifacts rejected only for renamed parameters passed the hidden checks unchanged; they are not counted as successes. The contribution is an evidence-qualified account of serving, agent protocol, session continuity, artifact validation, and verification in practice. The studies are not pooled and do not rank models or runtimes.

Conversational memory changes during use, so endpoint question answering alone cannot establish how a persistent state accumulates, ages, or incorporates revisions. We introduce LSREP, a Longitudinal State-Replay Evaluation Protocol combining ordered replay, explicit lifecycle schedules, repeated probes, evolving reference answers, and mechanism-fidelity checks. Its architectural case study is ICE v2, a local-first memory middleware with typed stores, retrieval fusion, and dynamic context budgets. The private, single-user instantiation contains 1,985 turns, 219 distinct probes, and 1,211 probe-checkpoint observations across 52 checkpoints. On three ordinary-density datasets, ICE v2 has a near-zero mean quality difference from vector-RAG while selecting 32% fewer fragments but using 6.6% more estimated prompt tokens. A fourth, dense dataset exposes catastrophic failures of the unbudgeted baseline. The fidelity audit limits attribution: procedural retrieval is defective, several mechanisms are unexercised, and graph utility is not established. In a complementary matched public diagnostic, ICE v2 loses decisively to pure vector-RAG on LongMemEval: 50.8% versus 72.8% in the evidence-only oracle and 43.0% versus 69.5% in full-S. Paired differences are -22.0 points (95% CI [-26.6, -17.4]) and -26.5 ([-31.3, -21.8]). Conservative abstention accompanies severe multi-session and temporal failures. ICE uses less context in this diagnostic, establishing a quality-cost trade-off rather than superior efficiency. Together, replay, fidelity auditing, and public endpoint testing expose distinct failure modes that neither architectural descriptions nor aggregate scores identify alone.

Execution feedback can guide coding agents toward correct repository repairs, but only when the tests capture the behavior requested by the issue. Agent-generated tests can encode incomplete or incorrect behavioral targets; when the same trajectory writes both the patch and the test, their errors can agree and create false confidence. We introduce ExecCritic, combining a test--verify--revise scaffold with a role-specific reinforcement learning recipe for training agents within it. The scaffold separates test construction from source-code repair: a Test agent independently generates repository-native tests, a fail-closed harness qualifies and freezes them, and a Repair agent revises source code from their execution feedback without changing the tests. Both roles use Qwen-3.5-35B-A3B as the backbone and are trained separately. In Learn to Test, the Test agent learns to produce behaviorally valid tests that distinguish correct from incorrect patches. In Test to Improve, the Repair agent learns both direct task resolution and feedback-guided revision. On SWE-bench Verified, test quality determines whether feedback helps: holding the base Repair agent fixed, tests from the base Test agent reduce resolved rate from a no-test baseline of 61.2% to 57.3%, whereas tests from GPT-5.6-sol raise it to 65.3%. Role-specific post-training raises the Qwen Test agent's Base-to-Gold success from 22.2% to 62.2%; composing the two post-trained Qwen agents reaches 72.6%, an 11.4-point gain over the original no-test baseline without stronger-model or Oracle feedback at evaluation time. Code is publicly available at https://github.com/MSR-Orchard/execcritic.

Multi-agent systems (MAS) are critical for automating complex tasks, yet their practical deployment is severely hampered by the challenge of failure attribution. Current diagnostic tools, which rely on statistical correlations, are fundamentally inadequate; on challenging benchmarks like Who\&When, state-of-the-art methods achieve less than 15\% accuracy in locating the root-cause step of a failure. To address this critical gap, we introduce the first failure attribution framework for MAS grounded in multi-granularity causal inference. Our approach makes two key technical contributions: (1) a performance causal inversion principle, which correctly models performance dependencies by reversing the data flow in execution logs, combined with Shapley values to accurately assign agent-level blame; (2) a novel causal discovery algorithm, CDC-MAS, that robustly identifies critical failure steps by tackling the non-stationary nature of MAS interaction data. The framework's attribution results directly fuel an automated optimization loop, generating targeted suggestions whose efficacy is validated via counterfactual simulations. Evaluations on the Who\&When and TRAIL benchmarks demonstrate a significant leap in performance. Our method achieves up to 36.2\% step-level accuracy. Crucially, the generated optimizations boost overall task success rates by an average of 22.4\%. This work provides a principled and effective solution for debugging complex agent interactions, paving the way for more reliable and interpretable multi-agent systems.

Vision-language foundation models (VLMs) show promise for diverse imaging tasks but often underperform on medical benchmarks. Prior efforts to improve performance include model finetuning, which requires large domain-specific datasets and significant compute, or manual prompt engineering, which is hard to generalize and often inaccessible to medical institutions seeking to deploy these tools. These challenges motivate interest in approaches that draw on a model's embedded knowledge while abstracting away dependence on human-designed prompts to enable scalable, weight-agnostic performance improvements. To explore this, we adapt the Declarative Self-improving Python (DSPy) framework for structured automated prompt optimization in medical vision-language systems through a comprehensive, formal evaluation. We implement prompting pipelines for five medical imaging tasks across radiology, gastroenterology, and dermatology, evaluating 10 open-source VLMs with four prompt optimization techniques. Optimized pipelines achieved a median relative improvement of 53% over zero-shot prompting baselines, with the largest gains ranging from 300% to 3,400% on tasks where zero-shot performance is low. These results highlight the substantial potential of applying automated prompt optimization to medical AI systems, demonstrating significant gains for vision-based applications requiring accurate clinical image interpretation. By reducing dependence on prompt design to elicit intended outputs, these techniques allow clinicians to focus on patient care and clinical decision-making. Furthermore, our experiments offer scalability and preserve data privacy, demonstrating performance improvement on open-source VLMs. We publicly release our evaluation pipelines to support reproducible research on specialized medical tasks, available at https://github.com/DaneshjouLab/prompt-triage-lab.

LLM agents are increasingly explored for automating root cause analysis (RCA) in cloud-native systems, creating a need to evaluate both diagnostic correctness and the quality of the supporting investigation. Existing static benchmarks offer repeatable inputs but limited system-facing interaction, while live testbeds expose realistic tools but hinder controlled comparison because incident evidence varies across runs; both paradigms focus primarily on final answers. To address these limitations, we present Cloud-OpsBench, an evaluation infrastructure for interactive and evidence-grounded cloud RCA. It comprises 754 runtime-verified cases across 57 fault types on two microservice workloads spanning application services and Kubernetes platform layers. Each fault is captured as a state snapshot and replayed through standard diagnostic interfaces, with outcome labels and diagnostic evidence graphs that enable matched comparisons and process-level analysis. Across ten LLM agents, the strongest Joint RCA Accuracy (JRA) reaches 0.76 on OnlineBoutique and 0.68 on TrainTicket, while the corresponding Evidence Closure Rates (ECR) are only 0.38 and 0.15. This outcome--process gap shows that final-answer correctness alone substantially overestimates agents' ability to perform evidence-grounded diagnosis.

Automated issue solving aims to resolve real-world issues in software repositories. The most popular benchmarks for automated issue solving are SWE-bench and its human-filtered subset SWE-bench Verified. These benchmarks leverage testing to validate generated patches. However, because testing is rarely exhaustive, a patch may pass the tests but nevertheless fail to match the developers' expectations. Unfortunately, it is currently unclear to what extent evaluations performed with SWE-bench suffer from such plausible but incorrect patches. This paper presents an in-depth empirical study of the correctness of plausible patches generated by three state-of-the-art issue-solving tools evaluated on SWE-bench Verified. We extensively test and inspect generated patches, and compare them against human-written ground truth patches. The core of our methodology is a novel technique PatchDiff for differential patch testing, which automatically exposes behavioral discrepancies between two patches. Our findings reveal critical weaknesses in SWE-bench's patch validation mechanism, which causes 7.8% of all patches to count as correct while failing the developer-written test suite. Moreover, our novel automated technique reveals that even more (29.6%) plausible patches induce different behavior than the ground truth patches. These behavioral differences are often due to similar, but divergent implementations (46.8%) and due to generated patches that adapt more behavior than the ground truth patches (27.3%). Our manual inspection shows that 28.6% of behaviorally divergent patches are certainly incorrect. Combined, the different weaknesses lead to an inflation of reported resolution rates by 6.2 absolute percent points. Our findings are a call to arms for more robust and reliable evaluation of issue-solving tools. We envision our automated differential patch testing technique to be useful for this purpose.

The meaningful use of electronic health records (EHR) continues to progress in the digital era with clinical decision support systems augmented by artificial intelligence. A priority in improving provider experience is to overcome information overload and reduce the cognitive burden so fewer medical errors and cognitive biases are introduced during patient care. One major type of medical error is diagnostic error due to systematic or predictable errors in judgment that rely on heuristics. The potential for clinical natural language processing (cNLP) to model diagnostic reasoning in humans with forward reasoning from data to diagnosis and potentially reduce the cognitive burden and medical error has not been investigated. Existing tasks to advance the science in cNLP have largely focused on information extraction and named entity recognition through classification tasks. We introduce a novel suite of tasks coined as Diagnostic Reasoning Benchmarks, DR.BENCH, as a new benchmark for developing and evaluating cNLP models with clinical diagnostic reasoning ability. The suite includes six tasks from ten publicly available datasets addressing clinical text understanding, medical knowledge reasoning, and diagnosis generation. DR.BENCH is the first clinical suite of tasks designed to be a natural language generation framework to evaluate pre-trained language models. Experiments with state-of-the-art pre-trained generative language models using large general domain models and models that were continually trained on a medical corpus demonstrate opportunities for improvement when evaluated in DR. BENCH. We share DR. BENCH as a publicly available GitLab repository with a systematic approach to load and evaluate models for the cNLP community.

Quantification of real-time informal feedback delivered by an experienced surgeon to a trainee during surgery is important for skill improvements in surgical training. Such feedback in the live operating room is inherently multimodal, consisting of verbal conversations (e.g., questions and answers) as well as non-verbal elements (e.g., through visual cues like pointing to anatomic elements). In this work, we leverage a clinically-validated five-category classification of surgical feedback: "Anatomic", "Technical", "Procedural", "Praise" and "Visual Aid". We then develop a multi-label machine learning model to classify these five categories of surgical feedback from inputs of text, audio, and video modalities. The ultimate goal of our work is to help automate the annotation of real-time contextual surgical feedback at scale. Our automated classification of surgical feedback achieves AUCs ranging from 71.5 to 77.6 with the fusion improving performance by 3.1%. We also show that high-quality manual transcriptions of feedback audio from experts improve AUCs to between 76.5 and 96.2, which demonstrates a clear path toward future improvements. Empirically, we find that the Staged training strategy, with first pre-training each modality separately and then training them jointly, is more effective than training different modalities altogether. We also present intuitive findings on the importance of modalities for different feedback categories. This work offers an important first look at the feasibility of automated classification of real-world live surgical feedback based on text, audio, and video modalities.

Context: Test-driven development (TDD) is a widely employed software development practice that involves developing test cases based on requirements prior to writing the code. Although various methods for automated test case generation have been proposed, they are not specifically tailored for TDD, where requirements instead of code serve as input. Objective: In this paper, we introduce a text-to-testcase generation approach based on a large language model (GPT-3.5) that is fine-tuned on our curated dataset with an effective prompt design. Method: Our approach involves enhancing the capabilities of basic GPT-3.5 for text-to-testcase generation task that is fine-tuned on our curated dataset with an effective prompting design. We evaluated the effectiveness of our approach using a span of five large-scale open-source software projects. Results: Our approach generated 7k test cases for open source projects, achieving 78.5% syntactic correctness, 67.09% requirement alignment, and 61.7% code coverage, which substantially outperforms all other LLMs (basic GPT-3.5, Bloom, and CodeT5). In addition, our ablation study demonstrates the substantial performance improvement of the fine-tuning and prompting components of the GPT-3.5 model. Conclusions: These findings lead us to conclude that fine-tuning and prompting should be considered in the future when building a language model for the text-to-testcase generation task

Unstructured clinical notes contain essential patient information but are challenging for physicians to search and interpret efficiently. Although large language models (LLMs) have shown promise in question answering (QA), most existing systems lack transparency, usability, and alignment with clinical workflows. This work introduces an interactive QA system that enables physicians to query clinical notes via text or voice and receive extractive answers highlighted directly in the note for traceability. The system was built using OpenAI models with zero-shot prompting and evaluated across multiple metrics, including exact string match, word overlap, SentenceTransformer similarity, and BERTScore. Results show that while exact match scores ranged from 47 to 62 percent, semantic similarity scores exceeded 87 percent, indicating strong contextual alignment even when wording varied. To assess usability, the system was also evaluated using simulated clinical personas. Seven diverse physician and nurse personas interacted with the system across scenario-based tasks and provided structured feedback. The evaluations highlighted strengths in intuitive design and answer accessibility, alongside opportunities for enhancing explanation clarity.

Evaluations should do more than measure a models current performance. They should tell us what to fix for the next model iteration and provide a way to generate targeted post training data. Most evaluation pipelines identify weak examples, topics, or categories, but they leave the underlying capability failure implicit: they say where a model fails, not why. We introduce CRAFT, a method that converts any rubric based evaluation dataset into a model specific diagnosis of weak capabilities. CRAFT treats each grading criterion as a capability probe: it extracts a capability description from every prompt rubric pair, clusters these descriptions into a hierarchical capability tree, scores the target model at every node, and selects low performing nodes dynamically across tree levels, at the granularity where each failure is clearest. The selected weak capabilities then direct the generation of targeted supervised finetuning data. Holding the data generation, finetuning, and evaluation setup fixed, we compare CRAFT against prompt level EvalTree clustering and untargeted random generation on four open source models, two professional domains (finance and legal), and 13 held out benchmarks disjoint from the diagnostic data. CRAFT achieves the strongest finance domain average for all four models under repeated temperature decoding; on legal domain, it is strongest for three of four models and remains within the decoding variance bands of the best baseline on the fourth. Diagnosing weaknesses at the level of rubric criteria, rather than prompts or categories, thus yields both a sharper picture of what a model cannot do and measurably better models after finetuning on that diagnosis.

Long-horizon coding agents need timely corrections, yet feedback can be ineffective or even harmful when it misjudges ongoing work or fails to address the underlying problem. Existing critics focus on evaluating trajectories and generating feedback, but rarely track what happens after feedback is delivered. We present Opera, a verbal critic framework that treats each correction as a persistent note, followed until the diagnosed problem is resolved. Opera decides when to review through periodic and event-driven triggers, diagnoses issues with typed operators, audits feedback against visible evidence before delivery, and tracks the agent's subsequent actions to distinguish mere compliance from actual resolution. As a test-time critic, Opera improves the resolve rate of non-critic agents by up to 12.4, 15.0, and 8.9 percentage points on Terminal-Bench 2.1, a SWE-Bench Pro subset, and DeepSWE v1.1, respectively, across four policy models, and achieves the highest mean resolve rate among competitive critic baselines on all three benchmarks, and also improves policy models when the policy critiques itself. Beyond inference, Opera-guided rollouts provide approximately on-policy training data: fine-tuning Qwen3.5-9B on them improves its resolve rate on held-out SWE-Bench Pro repositories by 10.2 percentage points without a critic at inference time, matching fine-tuning on rollouts from a stronger model, while preserving its performance when switching harness, i.e., from Openhands to Terminus-2, which the latter substantially degrades. Our code is available at: https://github.com/dongyuanjushi/Opera.

Community Notes, the crowd-sourced misinformation governance system on X (formerly Twitter), allows users to flag misleading posts, attach contextual notes, and rate the notes' helpfulness. However, our empirical analysis of 30.8K health-related notes reveals substantial latency, with a median delay of 17.6 hours before notes receive a helpfulness status. To improve responsiveness during real-world misinformation surges, we propose CrowdNotes+, a unified LLM-based framework that augments Community Notes for faster and more reliable health misinformation governance. CrowdNotes+ integrates two modes: (1) evidence-grounded note augmentation and (2) utility-guided note automation, supported by a hierarchical three-stage evaluation of relevance, correctness, and helpfulness. We instantiate the framework with HealthNotes, a benchmark of 1.2K health notes annotated for helpfulness, and a fine-tuned helpfulness judge. Our analysis first uncovers a key loophole in current crowd-sourced governance: voters frequently conflate stylistic fluency with factual accuracy. Addressing this via our hierarchical evaluation, experiments across 15 representative LLMs demonstrate that CrowdNotes+ significantly outperforms human contributors in note correctness, helpfulness, and evidence utility.

Large Language Models (LLMs) based autonomous agents demonstrate multifaceted capabilities to contribute substantially to economic production. However, existing benchmarks remain focused on single agentic capability, failing to capture long-horizon real-world scenarios. Moreover, the reliance on human-in-the-loop feedback for realistic tasks creates a scalability bottleneck, hindering automated rollout collection and evaluation. To bridge this gap, we introduce AgencyBench, a comprehensive benchmark derived from daily AI usage, evaluating 6 core agentic capabilities across 32 real-world scenarios, comprising 138 tasks with specific queries, deliverables, and rubrics. These scenarios require an average of 90 tool calls, 1 million tokens, and hours of execution time to resolve. To enable automated evaluation, we employ a user simulation agent to provide iterative feedback, and a Docker sandbox to conduct visual and functional rubric-based assessment. Experiments reveal that closed-source models significantly outperform open-source models (48.4% vs 32.1%). Further analysis reveals significant disparities across models in resource efficiency, feedback-driven self-correction, and specific tool-use preferences. Finally, we investigate the impact of agentic scaffolds, observing that proprietary models demonstrate superior performance within their native ecosystems (e.g., Claude-4.5-Opus via Claude-Agent-SDK), while open-source models exhibit distinct performance peaks, suggesting potential optimization for specific execution frameworks. AgencyBench serves as a critical testbed for next-generation agents, highlighting the necessity of co-optimizing model architecture with agentic frameworks. We believe this work sheds light on the future direction of autonomous agents, and we release the full benchmark and evaluation toolkit at https://github.com/GAIR-NLP/AgencyBench.

Evaluations (Evals) are a deployment bottleneck for real-world AI applications: public benchmarks rarely match a team's users, context, or policies, and human review is often tedious to scale. Motivated by our work with AI applications in the public sector, this project addresses recurring evaluation challenges encountered when applications must satisfy local policy and governance requirements. We present Kaleidoscope, an integrated workflow for contextual functional evaluation that links persona-based test generation, contextualized rubrics, and human review for reliability-gated automated scoring. Generated test cases are scored against application-specific rubrics; human annotations provide reviewable labels; and LLM judges automate scoring only when their agreement with those labels meets a configured threshold. Kaleidoscope is therefore a practical, inspectable, iterative workflow for product teams. We report early evidence from a three-week pilot across four organizational use cases and custom-rubric judge experiments on 108 annotated Q\&A pairs spanning four domains and 14 evaluation dimensions. The results highlight useful features for end-to-end reliable, automated scoring.

There has been a rapidly growing interest in Automatic Symptom Detection (ASD) and Automatic Diagnosis (AD) systems in the machine learning research literature, aiming to assist doctors in telemedicine services. These systems are designed to interact with patients, collect evidence about their symptoms and relevant antecedents, and possibly make predictions about the underlying diseases. Doctors would review the interactions, including the evidence and the predictions, collect if necessary additional information from patients, before deciding on next steps. Despite recent progress in this area, an important piece of doctors' interactions with patients is missing in the design of these systems, namely the differential diagnosis. Its absence is largely due to the lack of datasets that include such information for models to train on. In this work, we present a large-scale synthetic dataset of roughly 1.3 million patients that includes a differential diagnosis, along with the ground truth pathology, symptoms and antecedents for each patient. Unlike existing datasets which only contain binary symptoms and antecedents, this dataset also contains categorical and multi-choice symptoms and antecedents useful for efficient data collection. Moreover, some symptoms are organized in a hierarchy, making it possible to design systems able to interact with patients in a logical way. As a proof-of-concept, we extend two existing AD and ASD systems to incorporate the differential diagnosis, and provide empirical evidence that using differentials as training signals is essential for the efficiency of such systems or for helping doctors better understand the reasoning of those systems.

This paper proposes one of the first clinical applications of multimodal large language models (LLMs) as an assistant for radiologists to check errors in their reports. We created an evaluation dataset from two real-world radiology datasets (MIMIC-CXR and IU-Xray), with 1,000 subsampled reports each. A subset of original reports was modified to contain synthetic errors by introducing various type of mistakes. The evaluation contained two difficulty levels: SIMPLE for binary error-checking and COMPLEX for identifying error types. LLaVA (Large Language and Visual Assistant) variant models, including our instruction-tuned model, were used for the evaluation. Additionally, a domain expert evaluation was conducted on a small test set. At the SIMPLE level, the LLaVA v1.5 model outperformed other publicly available models. Instruction tuning significantly enhanced performance by 47.4% and 25.4% on MIMIC-CXR and IU-Xray data, respectively. The model also surpassed the domain experts accuracy in the MIMIC-CXR dataset by 1.67%. Notably, among the subsets (N=21) of the test set where a clinician did not achieve the correct conclusion, the LLaVA ensemble mode correctly identified 71.4% of these cases. This study marks a promising step toward utilizing multi-modal LLMs to enhance diagnostic accuracy in radiology. The ensemble model demonstrated comparable performance to clinicians, even capturing errors overlooked by humans. Nevertheless, future work is needed to improve the model ability to identify the types of inconsistency.

Software is one of the most powerful tools that we humans have at our disposal; it allows a skilled programmer to interact with the world in complex and profound ways. At the same time, thanks to improvements in large language models (LLMs), there has also been a rapid development in AI agents that interact with and affect change in their surrounding environments. In this paper, we introduce OpenDevin, a platform for the development of powerful and flexible AI agents that interact with the world in similar ways to those of a human developer: by writing code, interacting with a command line, and browsing the web. We describe how the platform allows for the implementation of new agents, safe interaction with sandboxed environments for code execution, coordination between multiple agents, and incorporation of evaluation benchmarks. Based on our currently incorporated benchmarks, we perform an evaluation of agents over 15 challenging tasks, including software engineering (e.g., SWE-Bench) and web browsing (e.g., WebArena), among others. Released under the permissive MIT license, OpenDevin is a community project spanning academia and industry with more than 1.3K contributions from over 160 contributors and will improve going forward.

Critical peer review of scientific manuscripts presents a significant challenge for Large Language Models (LLMs), partly due to data limitations and the complexity of expert reasoning. This report introduces Persistent Workflow Prompting (PWP), a potentially broadly applicable prompt engineering methodology designed to bridge this gap using standard LLM chat interfaces (zero-code, no APIs). We present a proof-of-concept PWP prompt for the critical analysis of experimental chemistry manuscripts, featuring a hierarchical, modular architecture (structured via Markdown) that defines detailed analysis workflows. We develop this PWP prompt through iterative application of meta-prompting techniques and meta-reasoning aimed at systematically codifying expert review workflows, including tacit knowledge. Submitted once at the start of a session, this PWP prompt equips the LLM with persistent workflows triggered by subsequent queries, guiding modern reasoning LLMs through systematic, multimodal evaluations. Demonstrations show the PWP-guided LLM identifying major methodological flaws in a test case while mitigating LLM input bias and performing complex tasks, including distinguishing claims from evidence, integrating text/photo/figure analysis to infer parameters, executing quantitative feasibility checks, comparing estimates against claims, and assessing a priori plausibility. To ensure transparency and facilitate replication, we provide full prompts, detailed demonstration analyses, and logs of interactive chats as supplementary resources. Beyond the specific application, this work offers insights into the meta-development process itself, highlighting the potential of PWP, informed by detailed workflow formalization, to enable sophisticated analysis using readily available LLMs for complex scientific tasks.

Prompt-based probing has been widely used in evaluating the abilities of pretrained language models (PLMs). Unfortunately, recent studies have discovered such an evaluation may be inaccurate, inconsistent and unreliable. Furthermore, the lack of understanding its inner workings, combined with its wide applicability, has the potential to lead to unforeseen risks for evaluating and applying PLMs in real-world applications. To discover, understand and quantify the risks, this paper investigates the prompt-based probing from a causal view, highlights three critical biases which could induce biased results and conclusions, and proposes to conduct debiasing via causal intervention. This paper provides valuable insights for the design of unbiased datasets, better probing frameworks and more reliable evaluations of pretrained language models. Furthermore, our conclusions also echo that we need to rethink the criteria for identifying better pretrained language models. We openly released the source code and data at https://github.com/c-box/causalEval.

Banking assistants must use account-specific information to answer requests and, in many cases, take actions through tools. Evaluating only the final response misses important errors. An assistant may ask for information it already has, rely on stale context, select the wrong account, or write an invalid value after stating the correct one. We introduce IndicBankBench, a 799-case benchmark for Indian retail banking spanning five operational domains, a capability/refusal domain, and twenty primary axes. Cases are evaluated at four stages: safety, action and tool use, response adequacy, and advisory quality. Tool use and most safety checks are deterministic. A narrow resolver handles only ambiguous confirmation-before-write cases, while a separate LLM judge evaluates semantic response adequacy. We run every case three times and report strict pass^3, which requires success on all trials. Across the eleven evaluated models, strict reliability ranges from 43.7% to 58.2%, whereas at-least-once success ranges from 60% to 74%. This gap shows that at-least-once success can overstate dependable banking behavior. The case-level diagnostics also distinguish systems that ask unnecessary questions from those that act but fail to reconcile customer context or fully resolve the request. We release the cases, mock environment, and evaluation harness.

By simply composing prompts, developers can prototype novel generative applications with Large Language Models (LLMs). To refine prototypes into products, however, developers must iteratively revise prompts by evaluating outputs to diagnose weaknesses. Formative interviews (N=8) revealed that developers invest significant effort in manually evaluating outputs as they assess context-specific and subjective criteria. We present EvalLM, an interactive system for iteratively refining prompts by evaluating multiple outputs on user-defined criteria. By describing criteria in natural language, users can employ the system's LLM-based evaluator to get an overview of where prompts excel or fail, and improve these based on the evaluator's feedback. A comparative study (N=12) showed that EvalLM, when compared to manual evaluation, helped participants compose more diverse criteria, examine twice as many outputs, and reach satisfactory prompts with 59% fewer revisions. Beyond prompts, our work can be extended to augment model evaluation and alignment in specific application contexts.

LLM-based simulated clients are increasingly used to train novice counselors, evaluate LLM therapists, and generate synthetic data. However, current simulators produce overly cooperative clients that disclose too readily, accept therapeutic reframes without resistance, and resolve core issues within a single session. We trace these issues to profiles that lack causal depth and behavioral mechanisms that treat all content as equally accessible. We present PatientAct, a framework for client simulation grounded in established clinical theories. Our profiles integrate the 5Ps clinical case formulation, providing causal depth without tying the design to any single therapeutic modality. During simulation, profiles include a dynamic memory layer in which items carry trust thresholds (e.g., symptoms are available early, whereas formative memories require a sustained therapeutic alliance). At each turn, the client's emotional reaction and behavior are modeled before generating a response. If the therapist approaches gated content, PatientAct expresses resistance in terms of quantity, content, and style rather than defaulting to cooperation or a single resistance pattern. We evaluate our framework on 40 clinical situations and demonstrate that it generates diverse profiles with high clinical plausibility. Moreover, PatientAct significantly outperforms the baselines, yielding substantial gains in resistance quality and behavioral realism. Our code and data will be publicly available via github.com/Sahandfer/PatientHub.

Artificial intelligence holds great promise for expanding access to expert medical knowledge and reasoning. However, most evaluations of language models rely on static vignettes and multiple-choice questions that fail to reflect the complexity and nuance of evidence-based medicine in real-world settings. In clinical practice, physicians iteratively formulate and revise diagnostic hypotheses, adapting each subsequent question and test to what they've just learned, and weigh the evolving evidence before committing to a final diagnosis. To emulate this iterative process, we introduce the Sequential Diagnosis Benchmark, which transforms 304 diagnostically challenging New England Journal of Medicine clinicopathological conference (NEJM-CPC) cases into stepwise diagnostic encounters. A physician or AI begins with a short case abstract and must iteratively request additional details from a gatekeeper model that reveals findings only when explicitly queried. Performance is assessed not just by diagnostic accuracy but also by the cost of physician visits and tests performed. We also present the MAI Diagnostic Orchestrator (MAI-DxO), a model-agnostic orchestrator that simulates a panel of physicians, proposes likely differential diagnoses and strategically selects high-value, cost-effective tests. When paired with OpenAI's o3 model, MAI-DxO achieves 80% diagnostic accuracy--four times higher than the 20% average of generalist physicians. MAI-DxO also reduces diagnostic costs by 20% compared to physicians, and 70% compared to off-the-shelf o3. When configured for maximum accuracy, MAI-DxO achieves 85.5% accuracy. These performance gains with MAI-DxO generalize across models from the OpenAI, Gemini, Claude, Grok, DeepSeek, and Llama families. We highlight how AI systems, when guided to think iteratively and act judiciously, can advance diagnostic precision and cost-effectiveness in clinical care.

The growing capabilities of large language models (LLMs) have led to their use as substitutes for human feedback for training and assessing other LLMs. These methods often rely on `constitutions', written guidelines which a critic model uses to provide feedback and improve generations. We investigate how the choice of constitution affects feedback quality by using four different constitutions to improve patient-centered communication in medical interviews. In pairwise comparisons conducted by 215 human raters, we found that detailed constitutions led to better results regarding emotive qualities. However, none of the constitutions outperformed the baseline in learning more practically-oriented skills related to information gathering and provision. Our findings indicate that while detailed constitutions should be prioritised, there are possible limitations to the effectiveness of AI feedback as a reward signal in certain areas.

Feedback is crucial for every design process, such as user interface (UI) design, and automating design critiques can significantly improve the efficiency of the design workflow. Although existing multimodal large language models (LLMs) excel in many tasks, they often struggle with generating high-quality design critiques -- a complex task that requires producing detailed design comments that are visually grounded in a given design's image. Building on recent advancements in iterative refinement of text output and visual prompting methods, we propose an iterative visual prompting approach for UI critique that takes an input UI screenshot and design guidelines and generates a list of design comments, along with corresponding bounding boxes that map each comment to a specific region in the screenshot. The entire process is driven completely by LLMs, which iteratively refine both the text output and bounding boxes using few-shot samples tailored for each step. We evaluated our approach using Gemini-1.5-pro and GPT-4o, and found that human experts generally preferred the design critiques generated by our pipeline over those by the baseline, with the pipeline reducing the gap from human performance by 50% for one rating metric. To assess the generalizability of our approach to other multimodal tasks, we applied our pipeline to open-vocabulary object and attribute detection, and experiments showed that our method also outperformed the baseline.

Agentic tasks require a large language model to interact with the world, navigating information and gathering evidence across multiple steps with restricted resources. Due to this complexity, agentic task failures arise from various sources, and pinpointing these failure causes is essential to diagnose and improve agentic systems. Existing benchmarks, however, tend to focus on a single leaderboard score, leaving the underlying failure modes opaque. To fill this gap, we introduce AgentHop, a diagnostic benchmark of 1,011 multiple-choice questions paired with a controlled seven-tool sandbox under fixed token, turn, and tool-call constraints. AgentHop reveals model vulnerabilities by dissecting a single accuracy score along four axes of agent operation: retrieval, synthesis, tool-call, and resource management. Across 19 models, we find that behavior clusters by model family, with tool-call signatures revealing distinct family fingerprints: GPT models commit early, Anthropic and GLM checkpoints verify before committing, DeepSeek and Kimi over-search, and Gemini-3 Pro stays balanced. Decomposed axes further expose within-family structure: Claude Opus 4.6 and Sonnet 4.6 land within one accuracy point yet diverge on retrieval-versus-synthesis emphasis, with Opus retrieving more and Sonnet synthesizing better. We release the full benchmark set and the harness to support diagnostic agent benchmarking.

Outcome-supervised search agents learn when and how to retrieve evidence, but terminal rewards neither localize intermediate errors nor redirect an ongoing trajectory before those errors compound. Treating corrective feedback as a learned in-trajectory intervention couples the two roles: the agent must decide when to request and use feedback, while the critic must infer useful corrections from outcome-confounded rollouts whose failure patterns shift as the agent improves. We introduce CAFE (Coupled Agent--Feedback Evolution), a framework in which a shared-parameter model alternates between search-agent and critic roles. CAFE initializes feedback-conditioned recovery from trajectories built around the base agent's own failures, then couples online and offline optimization. During online RL, a comparative feedback estimate uses a prompt-level call--skip success gap to shape request returns, while feedback-aware advantage shaping reweights token advantages before and after feedback. Offline, rollout-derived preference optimization learns feedback from matched successful and unsuccessful trajectories. On seven agentic search benchmarks, CAFE outperforms the evaluated RL-based search agents on average, retains its gains across all six out-of-domain benchmarks, and reduces answer-level hallucinations. One-sided ablations show that improving only the agent or only the critic eventually plateaus, whereas alternating the two updates continues to improve performance. These findings suggest that a self-improving search agent needs feedback that co-evolves with the policy it guides.

Tumor documentation in Germany is largely done manually, requiring reading patient records and entering data into structured databases. Large language models (LLMs) could potentially enhance this process by improving efficiency and reliability. This evaluation tests eleven different open source LLMs with sizes ranging from 1-70 billion model parameters on three basic tasks of the tumor documentation process: identifying tumor diagnoses, assigning ICD-10 codes, and extracting the date of first diagnosis. For evaluating the LLMs on these tasks, a dataset of annotated text snippets based on anonymized doctors' notes from urology was prepared. Different prompting strategies were used to investigate the effect of the number of examples in few-shot prompting and to explore the capabilities of the LLMs in general. The models Llama 3.1 8B, Mistral 7B, and Mistral NeMo 12 B performed comparably well in the tasks. Models with less extensive training data or having fewer than 7 billion parameters showed notably lower performance, while larger models did not display performance gains. Examples from a different medical domain than urology could also improve the outcome in few-shot prompting, which demonstrates the ability of LLMs to handle tasks needed for tumor documentation. Open source LLMs show a strong potential for automating tumor documentation. Models from 7-12 billion parameters could offer an optimal balance between performance and resource efficiency. With tailored fine-tuning and well-designed prompting, these models might become important tools for clinical documentation in the future. The code for the evaluation is available from https://github.com/stefan-m-lenz/UroLlmEval. We also release the dataset as a new valuable resource that addresses the shortage of authentic and easily accessible benchmarks in German-language medical NLP.

Agent harnesses enable large language models to act through tools, observe the results, and iteratively work toward a task goal. Verification checks and revises agents' work to improve output quality. Yet there is limited guidance on how harnesses should organize this effort. We study verification as an axis of test-time scaling by examining when to intervene, who checks the work, and how checking leads to revision. Our framework spans four layers of the agent lifecycle: initial instructions, mid-turn reminders, completion gates, and independent review. We compare 17 designs across seven frontier coding models on Terminal-Bench (TB) and SWE-bench Verified (SWE), and introduce VisualBench for expert evaluation of visual and interactive artifacts. Initial instructions and execution-time reminders yield modest, model-dependent average gains on the coding benchmarks, without substantially increasing average output token use. Completion gates can yield larger gains but may substantially increase cost. Independent review by a critic offers a more favorable accuracy-cost tradeoff, improving average accuracy by 5.9-9.5 percentage points on TB and 3.9-6.6 on SWE at 1.13-1.25 and 1.00-1.04 times baseline cost, respectively. On VisualBench, independent review achieves net preferences of 13.2-23.8 points over the baseline across its four designs, exceeding the evaluated L0-L2 designs. Additional review rounds or independent critics further improve average TB accuracy by up to 3.5 percentage points. Together, these results provide empirical guidance for verification design in agent harnesses and support independent review as a practical axis of test-time scaling.