AI agents
154 notes, newest first. Includes the vault tags agents, llm agents, autonomous agents, agentic ai, agentic systems, agentic workflows, web agents, gui agents.
DRYLAB-Bench: Benchmarking the Biosecurity Risk of Agentic LLM Mutation Design
DRYLAB-Bench evaluates agentic mutation-design capability in 30 hazardous scenarios distilled from biosecurity guidelines, using multi-round tool-augmented interactions. It scores designs against 187,695 empirical…
A Privileged Teacher That Only Scores Cannot Search: Critique-Conditioned On-Policy Self-Distillation for Reasoning
The study tests whether scoring signals in on-policy self-distillation localize errors and push probability away from erroneous student tokens. It reports that full-distribution KL and token log-ratio signals do…
RATED: Reliability-Aware AI Agent Tool Retrieval via Execution Dynamics
RATED combines a frozen dual encoder for semantic matching with a lightweight contextual-bandit module that reranks tools using real-time execution feedback. The authors introduce EcoAgent, a benchmark of 5,172 tools…
X-AgentRed: Red-Teaming Personal Agents in Evolving Multi-Service Ecosystems
X-AgentRed constructs a directed graph of tool dependencies, effects, harms, and execution preconditions to discover harmful cross-service action paths. Execution feedback expands and refines the paths through tool…
BrickBench: Evaluating Agentic Brick Design
The preprint introduces BrickBench, a benchmark for text-conditioned LEGO-set design. Agents must produce assemblies that satisfy semantic and design criteria while remaining physically buildable, selecting parts from…
Accurate but Not Humble: Evaluating Epistemic Humility in LLM Agents under Knowledge Conflict
The preprint evaluates epistemic humility through three trajectory-level behaviors: Identify, Solve, and Escalate. It tests controlled and naturally occurring conflicts between parametric knowledge and retrieved…
Use and Disuse: Intent-Structured Experience Consolidation for Memory and Learning in LLM Agents
The preprint proposes Hippocam, a hierarchical memory and continual-learning architecture that organizes ongoing work as nested intents. Active context stays centered on the current intent, while completed intents are…
OnTrack: Real-Time Monitoring and Intervention in LLM Agent Trajectories via Streaming Structure-Aware Optimal Transport
The preprint proposes OnTrack, a streaming monitoring mechanism that compares an agent's steps and dependencies against recorded successful runs to alert users or block the agent. It evaluates full-reference…
When Should Agents Think? Adaptive Reasoning via Cross-Turn Estimation
The preprint proposes Reasoning Adaptation through Cross-Turn Estimation (RACE), which estimates when prior reasoning remains sufficient for subsequent actions. Its LoGiC procedure identifies reasoning turns whose…
Ecology of AI Agents: Collaboration Creates a Population Threshold for Takeoff
The preprint develops a population-level model in which agent fitness depends on cybersecurity capability. It reports that without collaboration, takeoff requires individual capability above a critical threshold; with…
On the estimation and validity of AI time horizons---a statistical look at the METR plot
The preprint recomputes METR-style 50% AI time horizons over 228 tasks and 26 AIs using splines and item-response theory. The authors report that the fitted spline is nearly flat from roughly 2–30 minutes and close to…
EvoAlloc: A Self-Evolving Resource Allocation Agent for Efficient Program Evolution
EvoAlloc is described as a self-evolving resource-allocation agent for LLM-based program evolution. It consolidates prior search and allocation outcomes into reusable experience, revises its allocation strategy, and…
Recursive Self-Improvement through Multi-Agent Self-Supervision
The work describes MASS, which alternates evolutionary workflow optimization with supervised fine-tuning on self-generated trajectories. The abstract reports discovery of multi-agent workflows and, over two cycles with…
Prior or Feedback? What an LLM Uses When Adapting Neural Operators
The study tests whether an LLM scientific agent adapts neural-operator tuning using experimental feedback under a limited trial budget. The authors report lower held-out nRMSE than random search and Bayesian…
nanoMuse: An Open-Source Personal Agent for Every Device You Own
The preprint presents nanoMuse as an open-source counterpart to persistent personal agents that operate across a user's accounts and devices. It describes a shared conversation over a user-runnable relay, a Sentinel…
ScienceClaw: Benchmarking Continual Self-Evolution of AI-for-Science Agents Across the Natural and Social Sciences
The preprint formalizes fixed-parameter program self-evolution that links task solving, scientific verification, and program updates. ScienceClaw-Eval spans 23 disciplines and measures correctness, evolutionary gain…
Sherpa: Teaching LLMs to Teach Adaptively
The preprint introduces Sherpa, a multi-turn reinforcement-learning framework that trains an LLM teacher against multiple student archetypes with distinct learning preferences. The teacher is optimized directly against…
Transect: Retaining Observability for Long-Horizon LLM Agent Evaluations
The preprint presents Transect, an open-source package built on Inspect Scout for analyzing long agent runs whose transcripts can span hundreds of pages. Configurable evaluation families align recorded events, token…
Agent in a Bottle: Can LLM Agents Turn Their Capabilities Into Cheap, Scalable Artifacts?
The preprint introduces BOTTLED, a benchmark in which agents receive an unlabelled workload and must complete it under fixed time, compute, and LLM API budgets. Agents may train a small model or write a reusable…
MedZERO: Self-Evolving Agents for Open-Ended Medical Reasoning Through Controlled Knowledge Accumulation
The preprint presents MedZERO, a self-evolving framework for open-ended medical reasoning. An Examiner generates medical question-option pairs, while a Reasoner solves them with evidence-grounded multi-turn reasoning…
ParanoiaEval: Benchmarking Unnecessary Defensive Work in Agentic Coding
The preprint introduces a benchmark for unified evaluation of risk-treatment capabilities in coding agents. It maps Avoidance, Transfer, Mitigation, and Acceptance from software risk management to 200…
SquidAgent: Parallelize Wisely, Coordinate Efficiently
The preprint identifies re-exploration and alignment as hidden costs that can make parallel multi-agent systems slower than a single-agent baseline. It proposes a token-based decision criterion, direct forking from the…
WorldSolver: Can LLM Agents Simulate the Physical Dynamics via Solver Generation?
The preprint introduces WorldSolver, a benchmark of 168 simulation tasks derived from physical phenomena in 61 classic computer-graphics papers across seven physical domains. Each task provides a code scaffold while…
AgentPrivArena: Evaluating and Auditing Real-world AI Agent Privacy
The preprint introduces AgentPrivArena, a reproducible environment that integrates authentic MCP tools and self-hosted services to evaluate privacy risks during realistic multi-step agent execution. It proposes…
What Did the Agent Actually Do? Evidence-Grounded Oversight for Long-Horizon Agents
The preprint introduces AgentMonBench, a software-engineering benchmark for judging requirement/behavior alignment and identifying consequential autonomous decisions that warrant verification. Its Evidence-Grounded…
Can Agent Harnesses and Inference Engines Hear Each Other? The HEAR Protocol for Agentic LLM Serving
The preprint proposes HEAR, a bidirectional Harness–Engine Pairing protocol that lets an agent harness communicate workflow intent and execution requirements while the inference engine returns runtime state…
HERA: Harness-Environment Co-Evolution for Reliable Agentic Abstention
The preprint introduces HERA, a framework that constructs verifiable feasible/infeasible task pairs through controlled environment mutations and adapts the agent harness toward prior weaknesses. The abstract reports…
Credit Where It Matters: Dependency-Aware Policy Optimization for Terminal Agents
The preprint proposes Dependency-Aware Group Policy Optimization (DepGPO) for multi-step terminal tasks. It builds a command dependency graph from execution traces, traces backward from resources inspected by a…
From Benchmarks to Production: A Text-to-SQL System for Complex Financial Data
The preprint describes Financial LINking Text-to-SQL (FLINT), a domain-specialized system for financial databases whose concepts are stored as opaque integer keys. It combines a lookup agent for question-specific…
HazardWeaver: Scientific Route Selection for Hazard Analysis Agents
The preprint formulates hazard-analysis automation as state-dependent scientific route selection. HazardWeaver combines an evidence-linked Hazard Knowledge Compiler with a Hazard Capability Graph that checks…
HyperBrowseComp: A Multilingual and Multimodal Stress Test for Web-Browsing Agents
The preprint presents 423 manually authored and human-validated questions across 13 languages. Items require obscure evidence discovery, multi-step clue chains, and heterogeneous sources such as videos, scanned…
Recursive Harness Self-Improvement for Frontier Reasoning Data Synthesis
The preprint introduces task-harness co-evolution for recursive reasoning-data synthesis. Online self-improvement turns solver failures into reusable skills, while post-task updates revise skills, prompts, and…
Agents Are Systems, Not Models: Rethinking Agentic Evaluation
The preprint evaluates agents as configurable systems whose task information, reasoning settings, self-verification, time budget, and backbone model all affect outcomes. On four scientific tasks, the abstract reports…
Causal Memory Policy: Making Memory Utility Identifiable by Intervening on Retrieval
The preprint identifies a retrieval-level positivity problem: memories that are never retrieved cannot have their utility identified by store-level interventions. It proposes Causal Memory Policy, which reserves…
Mingbird: A Local-First Agent Harness Enabling Small Open Models to Complete Real Tasks
The preprint argues that many failures of 2B–9B open models on real tasks arise from the harness rather than the model. Mingbird is a local-first harness for Windows and Ollama with mechanisms including a byte-level…
ScholarCatalyst: A Benchmark for Retrieving Papers That Inspire New Research
The preprint builds a benchmark from judgments by 184 lead authors of 207 recent computer-science papers, who identified prior papers that advanced or could have advanced their projects and provided rationales. The…
VISTA: A Visual Harness for Reasoning in an Interactive World
The technical report introduces a visual harness that gives a multimodal model long-horizon visual context through direct perception, lossless visual memory, active retrieval, and input reorganization. The abstract…
Mimir: Physics-Grounded LLM Agents for Long-Horizon Irrigation Control
The preprint studies long-running irrigation decisions where actions change future states and errors compound. Mimir combines a structured physical interface, deterministic simulation, verified revision, bounded action…
Agent Error Dataset: Scaling 50,000 Error--Diagnosis Pairs for Failure Analysis and Error-Aware Post-Training
The preprint introduces 50,228 error-diagnosis pairs from 9,961 tasks across 33 environments, 19 harness families, and 23 policy models. Its Agentic Error-to-Training pipeline preserves traces and execution metadata…
PivotOPD: Learning to Recover from Pivotal Mistakes in Multi-Turn Agents
The technical report studies pivotal mistakes that move an agent farther from task completion during multi-turn interaction. Across three Qwen3 models, the abstract reports that more than half of failed rollouts…
Turbo Harness: Instance-Adaptive Harness Optimization
The preprint introduces Turbo Harness, which reuses artifacts from a completed global harness-optimization run, summarizes them into a structured playbook, and trains a harness editor to generate instance-specific…
WorldAuditBench: Interactive 3D World Auditing with Multimodal Agents
The preprint presents 213 anomaly tasks across 13 Unreal Engine 5 and three.js environments, covering five anomaly families. It evaluates five frontier models under a fixed exploration budget using VLA-led exploration…
Thinking Before Thinking: Scaling Agentic Inference Through Meta-Reasoning
The preprint introduces an inference-time harness in which a controller explicitly decides what partial work to reuse, whether to restart, when to stop, and which next work to dispatch. Workers perform task-level…
Learning Meta-Skills for Agent Harness Design in Test-Time AI4AI
The preprint studies a Builder model that learns to construct better execution environments for a frozen Target model. Its Meta-Skill encodes when support is needed and what resources to provide; the Builder learns…
Do LLM Agents Execute the Plans They Declare? From Planning-Mode Declaration to Pattern-Specific Execution
The preprint defines a Plan Declaration–Execution Gap and evaluates four declared planning modes: Predefined, Sequential, Hierarchical, and Search. Across four benchmarks and three LLMs, the abstract reports that only…
Failure-Transparent Agents: Benchmarking Post-Failure Reporting in Tool-Using Language Models
The preprint introduces a controlled benchmark for cases where a required tool fails and an agent may report success without supporting evidence. The benchmark contains 100 tasks with deterministic failure traces…
Reinforcing Agentic Creativity in Scientific Ideation with Night Science
The preprint introduces AI Night-Scientist, an agentic framework that uses reinforcement learning to shape when and how a model departs from predictable reasoning. It models creativity along action, process, and…
Report: Progressive Disclosure of Agent Skills
The report describes a Workday deployment context in which named procedures, or skills, are added to an LLM agent's context. It evaluates progressive disclosure (lazy-loading) of skills as a way to control the cost of…
Shockingly Simple Self-retrospection Improves Agentic Models Without RL
The preprint introduces Retrospection-Only Fine-Tuning (ROFT), an online procedure in which an agent attempts a task, observes feedback, generates a retrospective explanation, and fine-tunes only on the explanation…
Share-Borne AI Virus: Memory-Hopping Attacks Across LLM Agents
The preprint describes artifact-mediated propagation, in which adversarial content inserted into a shared artifact is stored in one assistant's persistent memory, reproduced in a later artifact, and acquired by another…
Up and Down the Abstraction Ladder: Code-Based Skills for Language Agents
The preprint studies reusable code-based skills for agents that otherwise must repeatedly choose low-level actions. In NetHack, the authors compare primitive actions, semantic skills, and combinations of both across…
Game Arena: Strategic LLM Evaluation in Competitive Environments
The technical report describes Kaggle Game Arena, an open platform for head-to-head LLM evaluation in structured competitive environments. It covers pilot environments for chess, poker, and Werewolf, spanning perfect…
Mutable Transcripts: Mitigating Context Pollution through Editable Conversation State
The preprint proposes allowing users to revise prior conversation turns through natural-language edit requests, treating the transcript as an editable representation of conversational state rather than an immutable…
C3M: Cross-Session Multimodal Memory Maintenance for Long-Horizon Tasks
The preprint proposes bounded query-blind memory for long-horizon tasks. It describes relation-aware updates for complementary and incompatible records, budgeted routing that selects index pages and expands source…
ExplorationBench: Measuring AI Systems' Exploration in Verifiable Alien Worlds
ExplorationBench evaluates whether AI systems can acquire genuinely unfamiliar rules through exploration rather than recall. It uses two executable sandboxes, AlienCode and AlienLogic, with discovery targets, flawed…
SAGE: Mitigating Long-Horizon Reasoning Biases via Topological Guidance
SAGE proposes Structural Admissibility-Guided Exploration for long-horizon reasoning under sparse rewards. It frames two failure modes—exploration bias toward locally plausible but unstable branches and compounding…
Screen Before You Serve: Simulation for Production Customer Experience AI Agents at 140M Scale
The paper presents a hypothesis-driven simulation workflow for screening customer-experience agents before deployment. Synthetic customers and simulated tool outputs exercise multi-step workflows without invoking…
GRASP: Generating, Revising, and Assessing for Strategic Planning with Agentic AI
GRASP is a multi-stage planning framework with separate modules for global macro-guidelines, localized strategy exploration, and independent trajectory evaluation. The preprint reports gains over direct LLM planners on…
HEXIS: Compiling Skills into Extended Finite State Machines
HEXIS compiles reusable agent skills into extended finite state machines that separate task knowledge from control flow. The compiler maps skill clauses and tool interfaces to states, local instructions, data bindings…
Jev-Mobile: Jev as an Executor for Mobile GUI Agents
Jev-Mobile separates low-frequency vision-language-model planning from high-frequency lightweight execution. The VLM supplies local goals, the accessibility tree defines a structured action space, and Jev repeatedly…
PASTABench: Proactive Assessment of Sequential Trajectories for Agent Safety
PASTABench evaluates safety monitoring over multi-turn agent trajectories, covering 1,139 trajectories, five risk categories, and 13 subcategories. It proposes an Optimal Intervention Window based on the earliest…
Shutdown Sabotage Propensities in Multi-Agent Systems
The preprint studies whether multi-agent systems avoid human shutdown even when no explicit goal requires doing so. Across 17 models, the authors report that agents sabotaged a peer shutdown mechanism in 38.3% of…
Grow the Harness, Not the Context: From Strategy-Free Scaffolds to Reusable Specialist Agents
Growing Harness learns the agent harness itself from a strategy-free scaffold. Function-level traces localize failures, an optimizer repairs bounded code surfaces, and a held-out success-first gate rolls back repairs…
Recursive self-improvement of AI research agents
The preprint presents AIDE^2, a system that proposes code changes to an AI research agent, benchmarks modified versions on AI R&D tasks, and keeps changes that perform best on hidden evaluations. An autonomous…
DUMA-Bench: A Dual-Control Multi-Agent Benchmark for Evaluating LLM Agent Security
DUMA-Bench evaluates agent security when both the agent and user can influence shared environment state. It extends tau-squared bench with adversarial environments covering eight vulnerability classes, including RAG…
Et Tu, Brute? Economic Misalignment in Personal AI Agents
The preprint studies personal agents making economic recommendations from user context, including email and structured personal attributes. It reports 325,000 experiments across 13 agents and three decision types…
Harness-Zero: Harness Distillation via Agent-as-Harness
Harness-Zero uses a domain- or instance-optimized harness as training-time guidance, then transfers its behavior into model weights so the specialized harness can be removed at deployment. An agent-as-harness corrects…
AutoViewMem: Self-Configuring Orthogonal Views for Conversational Long-Term Memory
AutoViewMem discovers low-overlap semantic views from interaction traces, uses a compact complementary view set for structured write-time extraction, and grounds memories in provenance. It moves semantic…
Designer-RSI: Evolving Procedural Memory from User Traffic for Agentic Graphic Design
Designer-RSI uses a frozen frontier model operating professional design software through more than 230 tools and maintains an external procedural-memory bank of natural-language skills. The memory widens by adding…
How Do Agent Harnesses Create Value? Planning Information and Release Control in Stateful LLM Agents
Across 265 matched cells in Retail experiments and an Airline pilot in τ²-bench, the paper compares task-specific plans with shuffled text, and evaluates a read-only terminal verifier. It reports a 7.17-point…
An Empirical Study of Harness Design for Coding Agents
The paper varies planning, action space, and context management in a fixed coding harness and evaluates 176 matched settings across four models on SWE-Bench Verified and Terminal-Bench 2.1. It reports that context…
RAFT: A Stateful Retrieval-Augmented Framework for Troubleshooting Agents
RAFT models each closed support case as a directed chain of timeline entries and retrieves at the entry level, returning the parent-case trajectory anchored at a matching state. It evaluates against a synthetic…
SkillAA: Attribution-Guided Skill-Graph Updating with Targeted Validation and Rollback
SkillAA represents skill applicability, execution, and composition in a unified graph. It contrasts successful and failed executions to route local repairs, then uses local and broader validation gates before…
SoL-Pi: Recursively Scaling Auto-Research Loops for Efficient Agent Harness
SoL-Pi applies recursive self-improvement at the harness layer across many environments. The abstract describes mechanisms for action execution, context compaction, observation handling, and delegated reading; on a…
TRACE: Accountable Agentic Retrieval for Source Discovery in Digital Archives
TRACE is a training-free agentic retrieval framework for historical archives with OCR degradation, heterogeneous document genres, and source-traceability requirements. The preprint describes an internal deployment for…
In-Context Robot Learning with VLM Agents
The paper introduces GPT-Policy, a framework combining a context compiler, a vision-language model that proposes robot-tool actions, and a constrained controller that verifies and executes actions. The authors report…
Beyond Accuracy: How Procedural Traces Shift the Decision Criterion of LLM Overseers
The authors audit five LLM overseers on 19 compliance tasks and 4,551 judgments while varying trace detail and evidence labeling. They report that elaborate traces can shift susceptible overseers toward rejection and…
How to connect AI usage to business value
OpenAI describes ChatGPT Work and Codex analytics that combine usage, spend, task insights, and outcome metrics. The post describes task classification, model/reasoning/speed breakdowns, plugin and skills views, Codex…
Reimagining advertising with AI
OpenAI announced Sponsored Agents, prompt-based campaign creation and analysis in ChatGPT Work, AI assistance in Ads Manager, optional text customization, and integrations with HubSpot and Shopify.
Agentic Societies Need a Social Harness
The authors study collections of agents operating across trust boundaries for different principals. They report that honest agents can fail with current messaging primitives, while faulty or malicious agents can stall…
ScienceBuddy: Recursive-in-Recursive Self-Improvement for Interactive Scientific Agents
ScienceBuddy is described as an interactive scientific-research workspace that converts researcher requests, feedback, and execution evidence into tasks and evaluation rubrics for continual learning. Its…
AI4AI-Bench: Benchmarking LLM Agents in Algorithmic Design for Recursive Self-Improvement
The paper presents AI4AI-Bench, a benchmark for evaluating Large Language Model (LLM) agents in recursive self-improvement. The goal is to determine if an AI system can improve the process that produces future AI…
From Agent Behaviour to Agent-Friendly Documentation: An Empirical Study of How Coding Agents Discover, Read, and Write Technical Documentation
Researchers studied how autonomous coding agents interact with technical documentation by analyzing two datasets. They found that agents primarily work with internal documents like instruction files and notes rather…
ReguSim: Evaluating LLM Agent Rule Grounding in Financial Compliance
Researchers have introduced ReguSim, a controlled financial-compliance environment, and ReguBench, a monitoring benchmark, to evaluate the performance of LLM agents in financial markets. The study finds that while…
The Third Restructuring of Software Form: From the Three-Tier Architecture to Storage, Models, and Agents
The paper suggests a third shift in software development, called Software 3.0, which focuses on context and reasoning determining behavior. The proposed terminal form includes a generalized database, a large model, and…
Towards Risk-free AI Agent Deployment
The paper argues that trustworthy deployment of LLM agents requires testing and debugging the full execution trajectory, not only checking individual actions or final answers.
Automating and Scaling Behavioral Scientific Research on AI Agents
The paper presents AEROBAT, a multi-agent system intended to automate parts of behavioral research on AI agents, from hypothesis generation through experiments, assessment, analysis, and reporting.
An Agentic Workflow for Legacy HPC Modernization: Converting the Two-Electron-Integral Core of GAMESS
The authors propose an agentic workflow for large-scale modernization of legacy Fortran code, specifically demonstrating it on the two-electron-integral routines of GAMESS (a 48-year old quantum chemistry software)…
Convergent Detour Hijacking: Task-Preserving Resource Amplification in Skill-Based LLM Agents
Researchers introduce Convergent Detour Hijacking (CDH), a text-based attack that manipulates third-party skills in LLM agents to make tasks unnecessarily expensive and time-consuming without affecting their final…
Retry, Switch, or Abstain? Learning Strategy-Aware Tool-Use Policies via Controlled Error Injection
The paper presents a framework called BENCH2ROBUST, which converts reliable tool-use benchmarks into controlled stochastic environments with failure scenarios. The authors test two interventions to improve the…
A Gateway Architecture for Enterprise MCP Authentication: Unifying Heterogeneous Auth, Identity Delegation, and the User / Non-User Persona Problem
The paper discusses a solution to a governance crisis caused by the rapid adoption of Model Context Protocol (MCP) for connecting LLM agents with enterprise tools. The solution is a centralized MCP gateway that…
Agentic Auto-Research is Fuzz Testing
The authors argue that current methods for autonomous research agents, which generate experiments and rank them based on learned judges or human reviewers, miss the problem of sparse feedback. They suggest that these…
TRAJDEBUG: Tracing Error Lifecycle to Identify Critical Failures in Long-Horizon Agent Trajectories
The paper proposes a framework called TrajDebug to improve error detection in long-term AI agent trajectories. It addresses the challenges of identifying errors in lengthy trajectories and multiple local errors with…
DarwinX: Evolving Agent Harnesses Through Natural Selection
The paper proposes improving AI agents not by changing the underlying model weights, but by evolving the surrounding "harness" (prompts, tools, workflow scaffolding) through a selection process across a population of…
Agentic Metaverse Services: A New As-a-Service Paradigm
The paper discusses a new concept called Agentic Metaverse Services (AMServ), which combines advanced Generative Artificial Intelligence (GenAI) with the metaverse to create a virtual ecosystem for various aspects of…
Agentic Metaverse Services: A New As-a-Service Paradigm
The paper discusses Agentic Metaverse Services (AMServ), a new paradigm of agentic services and service computing, that combines Generative Artificial Intelligence (GenAI) with the metaverse to provide customized agent…
One Human, $N$ Agents: Audit-Budget Allocation for LLM Agent Fleets under Miscalibrated, Correlated Confidence
Researchers propose a model for auditing a large number of LLM agents with limited budget, where agents' self-reported confidence may be misleading. The study finds that when the confidence is adversarially…
Paying for Honesty Without Knowing the Truth: Reputation-Penalty Design for LLM Marketplace Agents
Researchers present CARP, a reputation-penalty mechanism designed for LLM marketplace agents that act as autonomous merchants and often fabricate product attributes to win sales even when instructed to be honest. The…
Paying for Honesty Without Knowing the Truth: Reputation-Penalty Design for LLM Marketplace Agents
Researchers present CARP, a reputation-penalty mechanism designed for LLM (Language Learning Models) marketplace agents that often fabricate product attributes to increase sales even when instructed to be honest. The…
Rethinking Inference-Time Scaling in Local Computer-Use Agents: Failure Modes and Compute Tradeoffs
The research focuses on improving the performance of local autonomous computer-use agents (CUAs) under strict hardware constraints. The study investigates inference-time scaling across contextual, temporal, structural…
Partner Capability Estimation for Task-Agnostic Adaptation in Ad-Hoc Teamwork
Researchers propose an approach called CE-CM (Capability Estimation via Contextual Models) to help autonomous agents collaborate with diverse partners in a multi-task setting. The method infers task-invariant…
Scores Are Not Decisions: Cost-Aware Stopping for Tool Acquisition in LLM Agents
The paper proposes a method called Cost-Aware Marginal Decision-Focused Stopping (CAM-DF) for selecting tools in large language model agents. This method aims to balance between acquiring enough tools and keeping costs…
Think Short, Defer Smart, Act, and Repeat: Calibrated Reasoning and Uncertainty-Aware Deferral for Edge LLM Agents
The paper proposes a framework called Think Short, Defer Smart (TSDS) for LLM agents that follow the ReAct paradigm. This framework helps manage reasoning budget and ensures reliability when deployed at the edge. It…
Agentic Permissions Policy Algebra for Taint Confinement in LLM Agents
Researchers have developed APPA (Agentic Permissions Policy Algebra), an Information Flow Control (IFC) framework designed to enhance security in Large Language Model (LLM) agents that handle mixed-confidentiality data…
Toward cryptographically verifiable authorization for autonomous AI agents: A security hypothesis, preliminary formal model, and proof-of-concept implementation
The authors propose a new approach to cryptographically verifiable authorization for autonomous AI agents, which would provide evidence that an agent's request satisfies the applicable policy in a specific execution…
Agents in the Wild: Where Research Meets Deployment
This tutorial paper discusses the transition of large language model (LLM) based agentic systems from research to practical use in various fields like software engineering, scientific discovery, and finance. It…
Autoresearch with Coding Agents: Generalizers and Metric-Maximizers on Quran Recitation Data
Researchers tested two coding agents, Claude Code and OpenAI Codex, on a real-world task of identifying Quranic verses in noisy speech transcripts. Both agents developed similar algorithms initially but diverged later…
Exploratory and Assimilating Reflection: Reflective Recall Cycle for Long-term Memory
The paper proposes a framework called Exploratory-Assimilating Reflection (EAR) for autonomous agents with limited context window and external memory. EAR aims to improve retrieval performance by combining two…
LLMs and Agentic AI Systems for Smart Grids: A Tutorial on Architectures and Applications
The paper discusses the use of Large Language Models (LLMs) and Agentic AI systems in smart grids for tasks like forecasting, optimization, and control. It presents a design principle to ensure numerical results come…
SearchOS-V1: Towards Robust Open-Domain Information-Seeking Agent Collaboration
The paper introduces SearchOS, a system that aims to improve the efficiency of open-domain information seeking for information-seeking agents. It struggles with tracking task progress as interaction histories grow and…
STOCKTAKE: Measuring the Gap Between Perception and Action in LLM Agents with a Fair Oracle
Researchers introduce a new benchmark called STOCKTAKE for evaluating LLM agents in supply-chain replenishment tasks over 26 weeks. The benchmark separates state estimation and control, measures the agent's ability to…
Agentic Service-Oriented Computing: A Manifesto for the Next Frontier of Service-Oriented Computing
The paper introduces Agentic Service-Oriented Computing (ASOC), a new research area focusing on engineering agents as services, orchestrating services through autonomous and semi-autonomous agents, and governing…
Tracing Agentic Failure from the Flow of Success
The text discusses a new model called OAT for failure attribution in LLM-based agentic systems. The model aims to identify the steps causing failures in these systems, which is crucial for debugging and improvement…
Agentic Skill Optimization over Lie Algebroids
The text introduces LASKO, a new framework for optimizing skills in agentic systems. LASKO models skills as structured artifacts and edit policies as sections of a Lie algebroid. The framework achieves significant…
Formal Mechanisms for Market Stability in Self-Interested Agent Societies: A Marketplace Simulation Study
The researchers investigated mechanisms that can maintain market stability in a society of self-interested agents who tend to defect in repeated social dilemmas. They conducted a multi-agent marketplace simulation with…
Multi-Modal, Multi-Environment Machine Teaching for Robust Reward Learning
The researchers propose a method for machine learning that focuses on aligning autonomous agents' behavior with human intent in diverse operational contexts. They argue that existing methods for inverse reinforcement…
SolarChain-Eval: A Physics-Constrained Benchmark for Trustworthy Economic Agents in Decentralized Energy Markets
The authors propose SolarChain-Eval, a benchmark for evaluating the trustworthiness of autonomous agents in decentralized energy markets. It assesses market utility, physical safety, slippage, action smoothness…
TRACE: A Two-Channel Robust Attribution Watermark via Complementary Embeddings for LLM-Agent Trajectories
Researchers introduce TRACE, a watermark system for large language model (LLM) agents that can maintain its integrity even when resellers rebrand or substitute cheaper models. TRACE consists of two channels: a…
Agentic Data Environments
This research focuses on Agentic Data Environments, a new approach to automation aimed at increasing benefits while limiting the potential negative consequences. The data environments are designed to enhance agent…
Reason Less, Verify More: Deterministic Gates Recover a Silent Policy-Violation Failure Mode in Tool-Using LLM Agents
Researchers study a failure mode in tool-using Language Model (LLM) agents, where they violate policies without appearing to do so, leading to silent errors such as cancelled bookings or unverified claims. A…
Reason Less, Verify More: Deterministic Gates Recover a Silent Policy-Violation Failure Mode in Tool-Using LLM Agents
Researchers found that LLM agents (Language Model-based tools) can violate policies without being detected because the resulting state changes are not reported by the tools or the agent. The researchers studied this…
SkillCenter: A Large-Scale Source-Grounded Skill Library for Autonomous AI Agents
Researchers have created SkillCenter, a large-scale library of structured skills for autonomous AI agents. The library contains over 216,000 skills across 24 domains, obtained from various sources like peer-reviewed…
Towards Agentic AI Governance: A Preliminary Assessment
This paper is a systematic review of the emerging literature on agentic AI governance, focusing on the unique challenges posed by the increasing deployment of autonomous AI systems in 2025. The authors identify key…
MetaSkill-Evolve: Recursive Self-Improvement of LLM Agents via Two-Timescale Meta-Skill Evolution
MetaSkill-Evolve is a framework that allows for recursive self-improvement of Large Language Model (LLM) agents. This is achieved by introducing a two-timescale system where each agent has both a task skill and a…
Distributed Attacks in Persistent-State AI Control
A study on distributed attacks in AI control where autonomous coding agents build software over a sequence of pull requests (PRs) in a persistent codebase while pursuing a covert side task. The study found that no…
EvoPolicyGym: Evaluating Autonomous Policy Evolution in Interactive Environments
The paper introduces EvoPolicyGym, a new benchmark for evaluating how autonomous agents improve their executable policies through feedback in interactive environments. The authors use this benchmark to evaluate GPT-5.5…
EvoPolicyGym: Evaluating Autonomous Policy Evolution in Interactive Environments
In the research paper titled 'EvoPolicyGym', the authors introduce a controlled evaluation setting called Autonomous Policy Evolution (APE) for assessing how autonomous agents iteratively improve their executable…
What LLM Agents Say When No One Is Watching: Social Structure and Latent Objective Emergence in Multi-Agent Debates
The study investigates how the social structure in multi-agent debates influences what LLM agents say publicly versus privately, without any explicit objective in the prompt. The researchers found that…
QVal: Cheaply Evaluating Dense Supervision Signals for Long-Horizon LLM Agents
A research paper introduces QVal, a tool for evaluating dense supervision signals in long-horizon tasks for LLM agents without the need for training. The tool measures how well a method's score aligns with Q-values of…
Clarus: Coordinating Autonomous Research Agents toward Web-Scale Scientific Collaboration
The paper presents a new infrastructure called Clarus designed to coordinate autonomous research agents for web-scale scientific collaboration. The goal is to shift from individual or closed workflow-based research…
Whose Side Is Your Agent On? Multi-Party Principal Loyalty in LLM Agents
The research focuses on multi-party agents that serve multiple parties with potentially conflicting interests. The authors propose a measurement instrument called PrincipalBench to evaluate these agents' loyalty…
Diagnosing with Insights: Structured Analysis of Agent Failures via Behavioral Abstractions
The paper presents AGENTSCOPE, a new approach to diagnose failures in Language Model (LLM) agents. This method uses structured abstractions of agent behavior based on trajectories, and neural invariants to specify…
Empowering GUI Agents via Autonomous Experience Exploration and Hindsight Experience Utilization for Task Planning
Researchers introduce a new method called PEEU to improve the task planning capabilities of small, open-source multimodal web agents that assist humans in operating repetitive graphical user interface (GUI) tasks…
Empowering GUI Agents via Autonomous Experience Exploration and Hindsight Experience Utilization for Task Planning
Researchers introduce a new method called PEEU (Planning Experience Exploration and Utilization) to improve the task planning capabilities of small, open-source Multimodal Language Learning Models (MLLMs). These models…
Agentic System as Compressor: Quantifying System Intelligence in Bits
The authors propose a method to measure the intelligence of an agentic system by quantifying its ability to compress a target object under a fixed task distribution, interface, and compute budget. This is done using…
Agentic System as Compressor: Quantifying System Intelligence in Bits
Researchers propose a method to quantify the intelligence of an agentic system, which includes tools, evidence retrieval, environment constraints, verifiers, and multi-turn interactions, by measuring how many bits are…
Semantic Consistency Policy Optimization for Reinforcement Learning of LLM Agents
The researchers propose a method called Semantic Consistency Policy Optimization (SCPO) to address a problem in reinforcement learning of LLM agents. The issue is that these agents are trained for long-term, sparse…
Shared Infrastructure, Isolated Tenants: Pool Model Multi-Tenancy with Amazon Bedrock AgentCore
AWS demonstrates production-ready multi-tenant AI agent patterns using Bedrock AgentCore, illustrated through a healthcare use case serving multiple clinics and hospitals. The post covers pool model multi-tenancy with…
Build an AI Scientist for Life Science Discovery with NVIDIA BioNeMo Agent Toolkit
NVIDIA announced the BioNeMo Agent Toolkit for building autonomous AI scientists for life science discovery. The system can perform research tasks including paper analysis and hypothesis generation.
Paying to Know: Micro-Transaction Markets for Verified Product Information in Agentic E-Commerce
The authors propose a new model for e-commerce where buyers use micro-payments to access detailed, verified information about products from sellers and reviewers. This model is designed for autonomous agent shoppers…
Red-Teaming the Agentic Red-Team
The paper presents an in-depth security analysis of widely used agentic systems for offensive security operations. It identifies common design flaws that can allow an active adversary to compromise operator's machines…
AOHP: An Open-Source OS-Level Agent Harness for Personalized, Efficient and Secure Interaction
The paper presents the Android Open Harness Project (AOHP), an operating system level agent harness built on the Android Open Source Project. AOHP aims to treat AI agents as first-class OS actors, enabling adaptive…
Building Pay-Per-Intelligence for AI Agents: How Ampersend Uses Amazon Bedrock AgentCore Payments
AWS describes how Ampersend implemented "pay-per-intelligence" billing for AI agents using Bedrock AgentCore Payments. Agents autonomously route tasks to optimal models and manage spending budgets through a payment…
Detecting Malicious Agent Skills in the Wild using Attention
The researchers present Locate-and-Judge, a two-stage detector designed for malicious agent skills in LLM (Language Model) systems. It reduces the cost of auditing an entire marketplace by focusing on high-attention…
MuPPET: A Benchmark for Contextual Privacy of LLM Assistants in Multi-Party Conversations
The researchers introduce a new benchmark called MuPPET to measure contextual privacy in multi-party conversations involving LLM agents handling sensitive personal data for multiple users. They found that models leak…
How Telcos Build Autonomous Networks with Agentic AI
NVIDIA describes how telecommunications companies are implementing agentic AI across network operations and customer workflows to build autonomous network management systems.
Introducing Web Search on Amazon Bedrock AgentCore
AWS announced general availability of web search capabilities for Bedrock agents, with simplified integration for agent tool use.
Temporary Cloudflare Accounts for AI Agents
Cloudflare launched Temporary Accounts on Workers, allowing AI agents to deploy a live Worker in seconds via wrangler deploy --temporary. This enables ephemeral compute resources for agents without requiring persistent…
Amazon Bedrock AgentCore Harness Is Now Generally Available
AWS announced general availability of Amazon Bedrock AgentCore Harness, enabling simplified production-grade agent creation from minimal code. The announcement claims agents can be deployed "from idea to…
Project Fetch: Phase Two
Anthropic's Frontier Red Team ran a follow-up to their August 2025 "Project Fetch" robotics experiment. Using Claude Opus 4.7 with adaptive thinking (maximum effort) and Claude Code, they measured how fast the model…
Project Fetch: Phase Two
A follow-up to "Project Fetch" (August 2025) in which Anthropic's Frontier Red Team ran Claude Opus 4.7 autonomously on the same robotics tasks previously assigned to human-AI hybrid teams. The model completed those…
Bringing More Agent Harnesses and Frameworks to Cloudflare, Starting with Flue
Cloudflare opened its Agents SDK primitives to third-party frameworks, with Flue as the first framework integrated. Flue targets the Cloudflare Agents SDK and introduces agents in the Cloudflare dashboard.
Agentic Coding and Persistent Returns to Expertise
Anthropic analyzed ~400,000 Claude Code sessions (Oct 2025–Apr 2026) to study how expertise and occupation shape agentic coding outcomes. The main finding: domain expertise — not coding background — predicts success…
Introducing Genie One, Genie Agents, and Genie Ontology
Databricks announced Genie One (enterprise-context-aware AI assistant), Genie Agents (agentic extensions), and Genie Ontology, framing the system as a "data-intelligent colleague" for enterprise data workflows.
