Harness Engineering

Folio: Recoverable Multi-Resolution Context Folding for Long-Horizon Language Agents

MULTILOOP: Evolving Multi-Agent Harnesses for Complex Information-Seeking Tasks

DISCERN: Preserve at Write, Resolve at Read for Persistent LLM-Agent Memory

TreeAgent: Structure-Aware Memory Agent for Processing Long Hierarchical Data

Veritas: A Verifier Generation Harness for Agent Tasks

MIRIX-2.0: Multi-Agent Memory System

Harness Matters: Measuring and Dissecting Security Resilience in Web Agents

Can LLM Agents Organize Their Own Memory?

Agent Architecture Cost-Success Trade-offs: A Controlled Evaluation across Enterprise-Style SOPs

Compaction, decay and reinforcement in a two-tier conversational memory: a paired evaluation against retrieval-augmented generation

BOXBOX: Recursive Harness Self-Improvement for Long-Running Performance Optimization

Pdb or Not to Be: Comparing Debugging Interfaces With Pass$_{X}$@$k$

Persistent Retrieval for Conversational Agents: A Study of Retrieval Efficiency

CORE: Coarse-to-Fine Search with Context Management for Deep Research

Beneath the Surface: What We Overlook When We Talk About Agent Evolution

Evolver: Turning Agent Exploration into Evolving Capabilities

The Memory Trap: Language Agents Overgeneralize from Benign Interactions

Signed but Not Safe: Requirement Integrity in Multi-Agent LLM Handoffs

The Last Mile to the Real World: Agent-Native Primitives as Reusable Building Blocks for Solving Real-World Tasks

EnvOpt: Automated Optimization for Agent Environments

How Should Agents See Code? Rethinking Code Representations for Agentic Software Engineering

SPILLWAY: Separating Target Admission from Sanitization to Defend Tool-Using Agents against Prompt Injection

When Do Summaries Become Instructions? Unsupported Policy in LLM Agents' Self-Written Handoff Summaries

HarnessFlow: MCTS-Guided Harness Optimization for LLM Agents

The Harness Is the Model: Agent Scaffolding Changes a Fixed LLM's Compliance with Banking Conduct Rules, Unevenly Across Models

Route to Remember: Adaptive Memory Selection for LLM Agents

Recursa: Toward Recursively Self-Improving Foundation Model Systems

WASP-Mem: Let Agent Organize Procedural Memory. Human Spend Effort on Diagnosis

Do Agents Still Follow the Rules? Benchmarking Rule Compliance under Agent Context Compaction

ParetoSaddler: Pareto-Guided Optimization of Agent Harnesses for Quality and Cost

Delivered, Not Used: Evaluating Agent Memory at Its Point of Use

SPAR: Experience-Driven Strategy Evolution in Multi-Session B2B Sales

MAPLE: A Multimodal Mobile Agent with Playbook-Learned Experience

Skill++: Towards Lifelong Agents via Routing-Aware Skill Refinement

Exo: An Architecture for Self-Improving Systems at Runtime

MEMENTO: Learning Context Management for Long Context Processing Agents

RoboArchitect: Automating the Design of Robotic Agents with Persistent Memory

Context as Infrastructure: Knowledge Compilation for Retrieval in Agentic Systems with Harnesses

Accept Changes, Reject Attacks: Robust Memory Management for LLM Agents

Agentic World Modeling for Agent Test-time Scaling without Environment Replica

VeriCUA: Verification as Agentic Exploration for Computer-Using Agents

"Shall We Approve?" Consequence-Aware Approval for Autonomous Coding Agents

Don't Lose the Thread: Empowering Long-Horizon LLM Agents with Cognitive Resource Self-Allocation

KARMA: Training-Free Knowledge-Atomic Harness Evolution for Industrial Advertising Agents

Sandman: Building and Operating Sandboxes for Agentic Reinforcement Learning on Shared Kubernetes

Letting AI Formulate the Research Question: Autonomous Research Agents

Compile Away Uncertainty: Self-Evolving Agent Skills via Programmatic Compilation

MetaVul: Agentic Harness Discovery for Vulnerability Detection

What Does "Now" Mean to Your Computer-Use Agent? Stale Screenshots, Page Loads, and Which Check to Run Before a Click

Routed Continual Learning

Think Before You Act: Decision Gate for LLM Agents

Masked Erosion: How Acceptance Gates Lose Redundant Safeguards in Self-Improving Agents

Skeleton Coding: Program Structure as a Specification for Agents

Scaling Lean Theorem Proving for Real-World Code Verification

Scan Before You Install: Detecting and Mitigating Local and Composite Vulnerabilities in Skills

GEPA-Curator: Curated Learning for Self Improving Agents

Your Policy Is Secretly a Harness: The Decision Geometry of Agent Interfaces

When Skills Tell Agents Not to Ask: A Large-Scale Study of Oversight Directives in Agent Skills

Does Your Generalist Skill Optimizer Really Incentivize Agentic Tasks? Rethinking Skill Optimization via Bipartite Attribution

Evolving Agent Teams

When the Critic Falls Behind: Co-Evolving Rubrics and Harnesses for Continual Agent Self-Improvement

The View from Nowhere: Population-Centered Agent Evolution

The Inbox as a Coordination Primitive for LLM Agent Societies

Free to Look, Free to Reason: Simple LLM Harnesses Outdiagnose Hand-Crafted Root Cause Analysis Agents

The Hidden Costs of Forgetting: A Study of Trajectory Deletion in Software Engineering Agents

Generation and Verification: Human-Agent Collaboration for Safety-Critical Automotive Software Engineering

How Do Coding-Agent Harnesses Self-Evolve? An Empirical Study of Retrospective Adaptation

Rethinking Verification in Agent Harnesses

EMMA: Structured Episodic Memory for Multi-Agent Workflow Synthesis

CTXAgent - Self-Controlled Context Management of LLM Agents

Matched Capabilities, Different Behaviors: A Controlled Study of MCP and CLI Interfaces for Language Agents

AgentBaton: Efficient Model Collaboration in Agentic Tasks

Memory Tax: Dissecting the Net Value of Agent Memory

Holding a Steady Course: Knowledge Evolution from Expertise and Lessons for Continual Agent Learning

FlashScript: Compiling Recurring Agent Workflows into Reusable Neuro-Symbolic Programs

HarnessMM: Recursive Self-Improvement for Harness Token Efficiency

Tool-Calling LLM Agents Under Realistic MCP Faults: Scale Does Not Purchase Fault Tolerance

When Agents Fail: Model or Harness--Who's to Blame?

Beyond Gain: Measuring Skill Internalization for Agent Skill Retirement

RSIHub: A Composable Framework for Agent-Evolution Methods

Benchmarks Are Surprisingly Compaction-Friendly

Adaptive Skill Granularity for Coding Agents: Runtime Routing and Macro-to-Micro Recovery

When Is a Memory Safe Enough to Act On? Action-Conditioned Verification for Long-Lived Agents

Predictable performance is suboptimal, and what it means for model routing

AfterDeleteBench: Auditing Post-Deletion Information Persistence in LLM Agents

Memory-Aware Reward Modeling: Learning Human Preferences for LLM Memory Use

Metacognitive Scaffolds Create Productive Friction for AI-Assisted Knowledge Work

Is Complex Agentic Memory Always Needed? Naive RAG Is Stronger Than We Thought

Mining the Design Space of Multi-Agent Workflows

Can LLMs Improve Their Own Harness? A Controlled Study on ARC-AGI-3

Why Bother Me Again? Benchmarking Personal Agent Memory for Interaction Preferences

SplatLab: Autonomous Scientific Discovery for 3DGS Research

SWE-VERIFIER: Selective Verification for Cost-Efficient Coding Agents

Charts as Epistemic Actions for Experience-Driven Agentic Data Analysis

MemoryTax: The Bitter Lesson of Agent Memory

Harness Scaling for Generalizable Agentic Reinforcement Learning

AgentPolicyForge: From Executable Environments to Policy Benchmarks and Adaptive Guards

PerMIT: Securing Persistent Memory for LLM Agents via Machine-Checked Integrity Theorems and Scalable Community Trust

Back Up Before You Regret: Measuring Backup Awareness and Automating Backup in Coding Agents

Forging More Tasks into Better Harnesses: Agentic Harness Evolution via Successive Scale Racing

Evolving Agent Interfaces

Object-Oriented Harness: Evolving Agent Harnesses through Object-Oriented Design

Does Multi-Agent Collaboration Pay When Agents Serve Users? Testing Agent Scaling in Dual-Control Settings

More Memory, Worse Agents: Error Reproduction and Anti-Persistence in LLM Agents

TEX: Test-Time Scaling Testing Agents via Execution-based Cross-Validation

Understanding the Benefits of Scaffolds through Computational Separations

Same Simulation, Different Evidence: Where Retrieved Tool Output Costs LLM Agents Accuracy

When the harness decides success: state that cannot be re-observed

Are Self-Evolving Agents Continual Learners?

StopLoss: Teaching Agents to Predict Failure and Stop Early

Epi: Towards a Persistent, Self-Improving Agent for Autonomous Research

SPIDES: Information-Flow Control for Browser Agents

From Approval to Abuse: Privilege Escalation Attacks on Harness-Protected Coding Agents

Self-Evolving Coding Agents via Harness-Policy Optimization

TokenEconomizer: Experience-Driven Black-Box Cost Control for Coding Agents

Learning Self-Orchestration in Decentralized Agent Swarms

Learning to Steer LLM Agents and Multi-Agent Systems

LEASH: Uncertainty-Aware Weak-to-Strong Escalation and Oversight for Computer-Use Agents

Harness-Delta Attribution: Dissecting Gains From Harness Evolution

Librarian Agents: Training LLMs to Curate Long-Context Workspaces that Anticipate Future Queries

Beyond Downstream QA: Rethinking How We Evaluate Agent Memory

Rethinking the Role of Tool Graph Memory in Multi-Turn Tool Use

Trajectory-Level Speculation with Semantic Verification for Tool-Using Agents

JevAgentHarness: Benchmarking, Deploying and Training Decision Models That Share Control with LLMs

Memorilla: Latent Semantic Memory for LLMs

Bias Control and Injection for Multi-Agent Aggregation

TRACER: Trajectory-Aware Tool Retrieval for LLM Agents

MemComposer: Learning to Assemble Memories for LLM Agents via Reinforcement Learning

Probe-and-Resume: When and How to Evaluate Agent Execution

Same Context, Different Agents: A Behavioral Study of Semantic Layers, Orchestration and Compaction in Data Agents

Prune the Summary, Not the History: Source-Judged Compaction for Long-Horizon Agents

Evolving Together, Failing Apart: The Skill-Policy Co-evolution Trap in Self-Improving Agents

Scaling Discovery Requires Forgetting

Doc2Harness: Benchmarking LLMs on Long-Horizon Harness Engineering of Coding Agents

TreeFlex: Agents Navigating Their Own Search Trees

L3M: A Verifiable Harness and Runtime for Self-Extending and Self-Improving Agents

Why Do Agent Teams Fail Across Boundaries? A Failure Taxonomy for Agent-to-Agent Systems with Private States

Whose turn is it? Context-guided model routing for stateful agents

Plan or Retrieve First? A Scale Study of Tool Discovery Architectures

The Agentic Scientist: Orchestrating Coding Agents for Automated Scientific (Co-)Discovery

Scaling Inference-Time Compute with Teams of Coding Agents

DuoDiscover: Scaling Autonomous Discovery Beyond the Evaluation Bottleneck

Scaling Multi-Agent Intelligence with Substrate-Centric Coordination

Google Earth AI Geospatial Reasoning Agent: Capabilities and a Systematic Evaluation

Harmony Harness: Multi-Objective Self-Evolution of Agent Harnesses

Harness^2: Recursive Agent Harnessing for an Open World

Humanize MLE Agents: Multi-Agent Session Alternation Beyond Single-Agent Limits

Proactive Agentic Orchestration

SWE-Critix: Post-Training Generalist Verifiers for Assessing Coding-Agent Trajectories

LongHarness: Advancing Agentic Reasoning over 100M-Token Corpora via Stateful Evidence Management

Evo$^2$: Evolving the Evolver through Weight-Harness Co-Evolution

EvoBelief: Falsifiable World Models for Self-Improving LLM Agents

Recurrent Harness for Agentic Continual Learning

Memento 3: Model-Based Recursive Self-Improvement through Reflective Rulebooks

DeltaReplay: Task-Relative Memory Reuse for Mobile GUI Agents

TRACE: Diagnosing Verifier Brittleness in Agentic Evaluation

Harness Evolution Hits a Ceiling: When Weight Training Should Begin

One Skill Too Many: How Co-Installed Skills Conflict in Coding Agents

Memory Type Varies: Empowering LLM Agents for Long-Term Memory with Diverse Strategies

Who Verifies the Verifier? Co-Evolving Inspectable Graders with Self-Improving Agents

DuplexAgent-RSI: Recursive Harness Improvement for Full-Duplex Voice Agent Collaboration

When Interfaces Speak: Data-Aware Generative UI Harness for Active Interaction

CoTrace: Data Recipes for Training Terminal Agents with Harness-Model Co-Evolution

SkillForge: Co-Evolving Skills and Agents via Dynamic Skill Lifecycles

Secure-CUA: Controlling Untrusted Influence in Computer-Use Agents

From Investigation Failures to Reliable SOC Agents: Understanding and Improving LLM-Based Alert Triage

Finding Blind Spots in AppWorld and WorkArena Task Verifiers

Humanize: Judgement Engineering for Agentic Coding

VeriFine: Scaling Verification for Self-Improvement in Embodied Reasoning

Does an Agent's History Tell You When Compaction Will Hurt? A Modest, Bounded Effect on the TRACE Paired-Replay Corpus

SquidAgent: Parallelize Wisely, Coordinate Efficiently

EMHO: EMbodied Agent Harness Optimization via Experience Traces

When Tools Lie: Reliability of Mathematical Agents Under Corrupted Tool Feedback

DAEDALUS: Bootstrapping Agent Memory from Self-Generated Tasks

ReFold: Training-Free Reversible Inter-Turn Context Folding for Long-Horizon Agents

Harness Engineering for Software Engineering via Modular Executable Dev-Primitives

Do I Need the Cloud? Uncertainty-Aware Step-Level Handoff for Small Language Model Agents

Persistent Memory in Multi-Agent LLM Inference: What It Costs, What It Buys, and When You Can Tell

Stateless Language Agents: Scaling Long-Horizon Automated Research

Decoupled Multi-Agent Orchestration

Selective Critique for Cost-Aware LLM Agents in Long-Horizon Decision Making

Understanding and Mitigating Inference-Time Overreliance Using Agentic Memory

The Right Memory in the Wrong Context: Verifying Retrieval Admissibility in Long-Term Agent Memory

Catching Developers in the Flow: Low-Latency Agentic Program Repair at Google Scale

Back to the Future: Rethinking EDA Infrastructure for Agentic Systems in Chip Design Verification

AMBER: Training Long-Horizon Web Agents through Append-Only Memory

HERA: Harness-Environment Co-Evolution for Reliable Agentic Abstention

AECP: Artifact-Exclusive Communication Protocol for Multi-Agent Code Generation

When to Remember, When to Abstain: Category-Conditioned Retention for Reliable Agent Memory

Capability-Driven Self-Evolution of Agent Memory

VERA: Scaling Verifiable Environments for Agentic co-Evolution

Mining Agent Skills from Production Traces

TeleTune: Evolving Agent Skills From Offline Telemetry

Harness-Search: Guiding Long-Horizon Search through Multi-Agent Coordination

MESH-Harness: Self-Improving Agent Harnesses via Bandit-Guided Compositional Evolution

Causal Improvement Graph for Agentic Harness Optimization

Complex Agents, Shallow Tests: Demystifying and Enhancing Test Adequacy of Agent Harness in the Wild

ForkPilot: Self-Evolving Policy for Retrospective Search in Long-Horizon Agents

GitSwarm: Decentralized Compounding Inference

MemTrace: State-Consistent Memory for Long-Horizon Coding Agents

Principles that Guide, Actions that Inform: Agent Evolution via Knowledge Abstraction

What Does a Harness Buy? Tokens, Mostly

Do Tool Calls Execute as Intended? Measuring and Repairing Intent-Execution Correspondence in LLM Agents

SHarP: Saliency-based Pruning of Agent Harnesses

CUAWright: A Minimal Unified Interface for Digital Agents

IdeaScientist: Orchestrating Agents for Grounded Scientific Ideation

Sentry: Learning to Recover from LLM Agent Failures at Test Time

Continual Graph Memory for Mathematical Research Agents

When History Fails to Become Experience: Action Calibration in Language Agents

VERSE: Verified Self-Evolving Optimizer for Agent Harnesses

Learning What to Investigate Next: Meta-Reasoning for Long-Horizon Research Agents

When Terminal-Agent Training Stalls: Demystifying Data Generation and Verification Challenge

Coco: An Agentic Copilot for the Hardware--Software Co-Design Lifecycle

VISTA: A Visual Harness for Reasoning in an Interactive World

Agents Are Systems, Not Models: Rethinking Agentic Evaluation

DeFA: Dependency-Guided Failure Attribution for LLM Agents

VeriHarness: Scaling Agentic Verification for Long-Horizon Tasks

GUI-HARVEST: Self-Improving GUI Agents through Evidence-Driven Harness Evolution

Finding the Right Fit: Model-Harness Interactions across Agent Tasks

ActiveSaddler: Automated Curriculum Learning for Agent Harness Optimization

Sapien: A Stateful Policy Engine for Autonomous AI Agents

Self-Evolving Coding Rules for AI Coding Agents

Cogentic: Multi-Agent Orchestration for Automated Proof Discovery

How Much of a Harness Does a Strong Agent Need for Autonomous ML Engineering?

Learning from Research: Toward Lifelong Agent Harness Evolution

Mid-Harness: Scaling Actions Between Model and Harness for Terminal Agents

RankEvolve: A Reliable Multi-Agent Auto-Research Harness for Evolving Ranking Models

DAGent: Evaluate-then-Grow Planning for Deep Research Agents

Schema: Discovering Unknown Environments via Agentic Program Induction

AIM: Agentic Idea Management for Automated Research

MILO: Automated Harness Discovery via Orchestrated Multi-Agent Evolution

Thinking Before Thinking: Scaling Agentic Inference Through Meta-Reasoning

Learning Meta-Skills for Agent Harness Design in Test-Time AI4AI

SelfSearch: Reward-Free Search for Self-Improving Agents

AgentBug-Smith: Automatically Reproducing Real-World Harness Bugs in Agentic Systems

Mixture of Self-Improving Branches For Agent Harness Optimization

FOCUS: Training-Free Decision-Preserving Context Compression for LLM Agents

Do Agent Benchmarks Do What They Say? An Executable-Contract Audit of Tool-Using Agent Environments

Follow the Entities: A Corpus Map for Agentic Search

VACE: Validation-Gated Alternating Co-Evolution of Agent Models and Harnesses

Traverse: Learning When to Remember, Reset, and Redirect for Long-Horizon Web Search

Harness Evolution as Learning: Approximation, Generalization, and Optimization Limits of Self-Improving Personal Agents

GitHarness: Git Init Your Harness Working Memory for Perpetual User Requirements

Adapting Context Compression for Long-Horizon Agents with Counterfactual Continuations

Engineering Simplicity: Simple Mechanism Interfaces Steer LLM Agents

DeepRewind: Predicting and Repairing Premature Commitments in Deep Research Agents

SCOUT: Synergizing Reasoning and Tool-Use for Computer-Use Safety

LongCat-DeepResearch Technical Report

SAGE: A Statistical Acceptance Gate for Self-Evolving Agents

Harness Learning Enables Generalizable Test-Time Adaptation

FlowState: Execution State as Memory for Long-Horizon LLM Agents

Just-In-Time Agent Memory with Runtime Agentic Research

Raven: The Harness of Harnesses for Composable Agentic Intelligence

LiteEvo: Automated, Cost-Efficient Harness Evolution for Generalization to Unseen Tasks

Beyond Token Savings: A Systematic Study of Context Compression in LLM Agents

More Programs or More Rolls? Separating Coverage from Specialization in LLM Harnesses

Agents as Software: A Programming Languages Agenda for Agent Reliability

Compress What You See, Not What You Say: Anchored Context Distillation for Latent-Observation Software Engineering Agents

Mutable Transcripts: Mitigating Context Pollution through Editable Conversation State

Up and Down the Abstraction Ladder: Code-Based Skills for Language Agents

MoMHa: Multi-Objective Optimization of LLM Harnesses over Accuracy, Safety, and Tokens

From Tapping to Hopping: Augmenting Mobile GUI Agents with App-Native Deeplinks

SkillEvoReg: Regularizing Agent Skill Evolution Against Overfitting

Learning What to Skip: Counterfactual Credit Assignment for Efficient Multi-Agent LLM Workflows

Probing Stability-Plasticity Tradeoffs in Agent Memory through Cognitive Experimental Paradigms

Stealth Apart, Harm Together: Skill Cascading Attacks on Skill-Based Agent Systems

When Can Agents Forget Their Reasoning? ICLR for Long-Horizon Agent Context Compression

IterSynth: Rethinking Deep Search Agents via Role-Decoupled Iterative Synthesis

A Wrong Turn Does Not Ruin the Journey: Deviation-Guided Skill Self-Evolution for LLM Agents

Where Does Exactly-Once Live? Model, Harness, and Tool-Contract Effects on Duplicate Side Effects in LLM Agents

Control the Harness, Control the Cost: Routing and Governing AI Coding Agents in the Enterprise

Just-in-Time Memory: Learning to Curate Task-Adaptive Memory for LLM Agents

ChipMEM: Verification-Grounded Memory for EDA Agents

Harness as a Language: A Minimalist Agent Framework With Maximal Expressivity

Agensh: Scaling Organizational Intelligence to 1,024 Agents

CliffCompaction: Cost-Efficient Compaction for Long-Horizon Coding Agents

Grow the Harness, Not the Context: From Strategy-Free Scaffolds to Reusable Specialist Agents

When Does Execution Provenance Help Agent Memory Retrieval?

How Strongly Should Task State Influence an LLM Agent?

Making Agents More Consistent: Skills Should Form Habits for Repeat Tasks

Harness-Zero: Harness Distillation via Agent-as-Harness

RRSI: Regularized Recursive Self-Improvement of Agent Harnesses

Qwen-Audio-Agent Technical Report

Silent Failures in Agent-Tool Interaction: An Audit of ToolUniverse

MemCalib: Benchmarking and Optimizing Memory Use in LLM Agents

Self-Healing Harness for Runtime Oversight of Agent Self-Modification

Jev-Mem: System-One-Controlled Agentic Memory for Efficient AI Agents

ScholarStack: Layered Research Asset Orchestration and Cross-Task Reuse for Scientific Agents

VibeMemBench: Evaluating Memory Systems for Coding Agents on Real Repository Coding Tasks

RPMem: Learning Long-Term Recurrent Parametric Memory Across Sessions for LLM Agents

Propose, Verify, Commit: Evidence-Grounded Memory for Long-Horizon Multi-Actor Conversations

Environment Steering: Using Data Flow Control to Improve Agent Utility and Safety

OptiSkill: A Hierarchical and Evolving SkillBank for LLM-Based Optimization Modeling

SelfOp: An Optimization Algorithm for Self-Improving Security Agents

Self-Organizing Agent Teams Learn to Reason Together

Designer-RSI: Evolving Procedural Memory from User Traffic for Agentic Graphic Design

AutoViewMem: Self-Configuring Orthogonal Views for Conversational Long-Term Memory

Verify, Don't Trust: Agentic Model Development for Video Discovery Retrieval at Scale

CoLearn: An Agentic Tutor that Learns its Learner in a Human--AI Co-Learning Loop

Loopjacking: Hijacking Human-in-the-Loop Approval

Scaling Discovery through Test-Time Communication

Coding Agents with an Obstacle-Aware Harness for Safe Robot Manipulation

An Empirical Study of Harness Design for Coding Agents

RAFT: A Stateful Retrieval-Augmented Framework for Troubleshooting Agents

Chronicle: Cut-Point Replay for Regression Testing of LLM Agents

SoL-Pi: Recursively Scaling Auto-Research Loops for Efficient Agent Harness

How Do Agent Harnesses Create Value? Planning Information and Release Control in Stateful LLM Agents

The Organization of Inference: Information, Resource Constraints, and AI Production

AdaRepair-Mem: Adaptive Experience Orchestration for Repository-Level Program Repair

JustMem: Just-Enough Memory Access for Long-Term Conversations

Long-horizon autoformalization of a core theorem underlying MIP* = RE

Rethinking Multi-Agent Collaboration: When More Is Less

LearnActCoder: Role-Aware Error Memory for Adaptive Clinical Coding Agents

FINSKILLOPS: A Self-Evolving Multi-Agent System for SEC Filing QA

ScientistTwo: Pioneering the Human Knowledge Frontier with Autonomous AI

When Hiring Becomes Agent-Mediated: Evaluating Access and Recurrence in Two-Agent Résumé Screening

AURORA: A Natural Language-Driven Agentic Framework for Understanding, Reasoning, and Orchestrating Reliable Air-Ground Co-Simulation

EconSkills: Studying Skill Transfer and Retrieval for Web Agents on Live Economic Data

An Architecture for Long-Horizon Agents: Levels, Ticks and Cascaded Intelligence

Reputation as Community Memory for the Agentic Web

Proxifield: Decentralized Multi-Agent Communication through Semantic Proximity

Closed-World Resolution Against Tool Hallucination in LLM Agents

From Rollout to Reset: A Graph-Based Harness for Autonomous Long-Horizon Manipulation Evaluation

MAGS: Multi-agent Auto-formalization Guarantees Safety for Agentic Outputs

GAVEL: Graph World Models for Verified and Efficient Long-Horizon LLM Task Planning

Compositional Policy Violations: When Step-Level Compliance Fails In Agentic AI Workflows

M-SQE: Multilingual Skill Quality Estimation for Enhancing Language Equality in Agentic Skill Use

Position: It is Time to Virtualize Foundation Models with a Self-evolving Operating System Layer

Bad Genius: Counterfactual-Guided Harness Evolution Beyond Task-Specific Shortcuts

Rollback the World, Keep the Reflection: Rollback-Induced Reflection for Long-Horizon LLM Agents

Symbolic Temporal Supervision of LLM Agents Using Contracts

Designing Agentic AI Workflow Portfolios under Imperfect Selection and Compute Cost

Agora: Git as Shared Memory for Collective AutoResearch

PrimeScientist: Strategic Allocation of Research Effort in Autonomous Research

Agentic Societies Need a Social Harness

Reflect, Revise, Reuse: Training-Free Skill Evolution for GUI Agents

Decomposition Buys Integrity, Not Yield

Mo' Models, Mo' Problems: How to best select model pools when designing Multi-Agent Systems

After the Party: Governing What a Viral Agent-Skill Ecosystem Left Behind

Grounding SWE-Agent Decisions in Architecture-0 Design: Navigating Unknown Unknowns through Physical Mapping

PaperDoctor: Evidence-Grounded and Actionable Feedback for Scientific Papers in Progress

EvolveTrade: Experience-Driven Policy Refinement for Self-Evolving LLM Trading Agents

RepoAtlas: Guiding Coding Agents via Evolving Multimodal Repository Views

EchoPath: Execution-Level Replayable Memory for GUI Agents

Protocol-Preserving Context Trimming for Agentic Workflows: Benefits, Failure Regimes, and Budget Guardrails

Spurious Tool Use: When RL Agents Learn the Wrong Reason to Act

Stellar Colosseum: A Many-Agent Harness for Long-Horizon Research in Mathematics and Theoretical Computer Science

The Router Within: Eliciting Native Skill Routing from a Frozen LLM

EvoOntology: A Self-Evolving Ontology Layer for Data Agents

RESKILL: Explicit Failure Attribution and Structured Repair for Interactive Language Agents

When Tool Calls Succeed but Workflows Fail: Anomalies at the Agent-Tool Boundary

SkillLift: Learning Dense Rubrics from Sparse Oracles for Efficient Skill Evolution

RSIAgent: Autonomous Exploration for Recursive Self-improvement in New Environments

OpenAI4S: Code as Action, Science as Sessions

BusMA: A Bus Communication Substrate for Multi-Agent Systems

CoMem: Collective-Individual Memory Synergy for Evolutionary Multi-Agent Systems

ModularRSI: Modular and Generalizable Recursive Harness Self-Improvement

The Stochastic Deputy: Structural Tenant Isolation for Tool-Using LLM Agents

Loop-Back Authority in LLM Agent Teams: A Paired Experiment on Flat and Hierarchical Coordination

Fabrication After Tool Failure: Tool-Augmented Agents Assert Values Their Tools Did Not Return

Question's Gambit: The First Move Matters in Agentic Deep Search

OmniHarness: Harnessing Generalizable Visual Generation via Symbolic Policy Learning

Retrieval-Driven Memory Reconsolidation for Long-Term LLM Agents

When Tools Get in the Way: The Effect of Unnecessary Tool Availability on LLM Answering

LIMBO: Lifelong Inference-Time Memory and Budget Optimization for LLM Agents

Confuse the Model, Control the Flow: Understanding and Mitigating Privacy Leakage from LLM Agents with Information Flow Control

Learning How Much to Collaborate: Difficulty-Aware Topology Selection for Multi-Agent Code Generation

When Malicious Instructions Persist: Persistent Memory Poisoning Attack on Harness-Based Agents

Do Not Restart: Residual Completion for Stateful Agent Handoffs

HarnessBandit: Joint Learnability-Transferability Scheduling for Multi-Harness Agentic Reinforcement Learning

AutoTailor: Automatic, User-Aligned Capability Selection and Adaptation for Web Agents

Asclepius: An Adaptive Harness for Long-Horizon Clinical Agents

Online Video Agent Harness for Long Video Understanding

What Drives Recovery in Agentic Text-to-Cypher? LAST-CQ: An LLM Agent Self-Refinement Framework

Skill Issue: Lessons from Optimizing Repository SKILLs for Coding Agents

LifeFuse-Mem: Lifecycle-Aware State Fusion Against Temporary Overwriting for Long-Term Memory

ORCH: Organizational Principles Enable Collective Intelligence in Embodied AI

When Agents Disagree: Bayesian Backward Reasoning as a Label-Free Anchor for Multi-Agent Collective Decision-Making

COBRA-Skills: Contextual Bandit-Guided Evolution for Agent Skill Optimization

Ecdysis: Efficient and Effective Training of Runtime Harnesses for LLM Agents

MAPLE: Memory-Augmented Planning with Language and Evolution

Agent-Integrated Software: Interaction Contracts and Continuous Assurance

Magenta: Closing the Loop Between Mathematical Reasoning and Lean Verification

Autonomous Chemical Mechanistic Discovery through Agentic Reasoning and Validation

Grounding Agent Memory: Environment-Probing Curation for Enterprise Agents

Is Bash All You Need? An Empirical Study of Tool Interfaces for Enterprise Digital Worker Agents

Auto-RecSys: Harnessing Autonomous Research Agents for Industry-Scale Recommender System

Show-Harness: Just a VLM Agent Can Play Robots

Fortunate Recall: Ontology-Driven Memory Lifecycle Management for Persistent Coherence in LLMs

Kernel-Managed Shared Memory for System-Wide Personalization

Belief-State Engine: Augmenting LLMs for Principled Planning Under Partial Observability

UnitBoost: Managing Compound LLM Systems with a Merge Operator, Not a Model

ROAM: Robust Organization of Atomic Memories for Agents through Semantic Relations

Introducing Consort: A Spec-First Agent Framework for Enforced, Test-Driven Development on Live Database Branches

RobustSGPO: Search-Space Control for Agent Harness Evolution

Harness or Model? Isolating the Harness Effect in Agentic Coding with a Contamination-Controlled Private Suite

Procedural Graphs: Self-Evolving Execution Structures for LLM Agents

Co-Evolving Harnesses and Models: On-Policy Correction Helps Weaker Models Catch Up Where Imitation Fails

ExecCritic: Learn to Test, Test to Improve for Coding Agents

SkillAdam: Stable and Efficient Skill Evolution for Agents

Closing the Consistency Gap: Self-Evolving Agents That Learn to Stay on Course

Beyond Agent Harnesses: Cross-Substrate Authority for Multi-Agent Systems

AttnCompress: Dynamic Attention-Guided Trajectory Compression for Software Engineering Agents

What Eviction Destroys: A Restore-Counterfactual Audit of Forgetting in Agent Memory

MemForest: Efficient Agent Memory Management via EventTree Partitioning and Progressive Merging

Agentic ML Exploration (A-MLE) for Ads Ranking

MoM: Memory of Memory

SkillAlign: Aligning Skill Interfaces for LLM-based Agents

Elastic Horizon: Discovering the Effective Interaction Frontier in Agentic Reinforcement Learning

PARSER: Read in Parallel, Reason in Depth for Long-Context LLM Agents

CHART: A Harness-Rotation Curriculum for Harness-Robust Search Agents

Does Your Agent's Memory Survive a Model Upgrade? A Controlled Study of Memory Portability

CONTINUITY: Security-Context Contracts for Composable LLM Agent Controls

Trace2Tower: Transition-Aware EigenTrace Induction of Multi-Level Skills for LLM Agents

TROVE: Adaptive Agent Skill Orchestration via Trace-Grounded Route Validation and Editing

Compact-Memory LLM Agents via Online Max-Member Clustering and Atom-Aware Packing

From Language Models to World-Acting Systems: Progress and Limits of Agentic AI across Digital, Social, Virtual, and Physical Environments

Forgetting Without Restarting: Execution-State Unlearning for Stateful LLM Agents

From Interaction Traces to Persistent Skills: Online Evolution for Computer-Use Agents

ESPO: Error-Structured Prompt Optimization via Diagnose, Diversify, and Stabilize

Interface-Induced Trajectory Censoring

Speak for Me: Giving LLMs the Situational Awareness to Participate in a Meeting

RuleMem: Active Rule Memory for Long-Term Conversational Agents

A Blind Trust, the Bloody Thrust: When Attacker-Controlled Hook Updates Steer AI Agent Harnesses towards Malicious Behaviors

Bioinfoysis Technical Report

SimSkill: A Lifelong Learning AI Agent for Autonomous Mastery of Traffic Simulation

Proactive Service Agents: A Unified Decision Framework, Methods, and Evaluation

What Do CAE Simulation Agents Really Need Beyond a Generic Harness?

Remember and Reweight: Enhancing Multi-Agent Debate with Experience Memory and Confidence Estimation

Plan Pointers and Record-Directive Form in Budgeted Verification of Inherited Agent Memory

Do GUI Agents Know When Not to Act? Enabling Conflict-Aware Termination for Multimodal GUI Agents

Dude: A Dual-Detection Multi-Agent System for Paper-Code Discrepancy Detection

EvalMem: An Operation-Level Diagnostic Framework for Long-Term Memory Systems

Fresh Memory, Stale Plans: Dependency-Scoped Validation for Distributed LLM-Agent Memory

Speculative Macro Commit for Faster Tool-Using Agents

SafeEvolve: Harness-Policy Co-Evolution from Agent Experience for Safety Alignment

Bilevel Coordinated Reflection: A Game-Theoretic Approach to Multi-Agent LLM Systems

Repo-To-Skill: Distilling GitHub Repositories Into AI4AI Skills

CORAL: An LLM-Native Harness for Production Recommender Systems

Toollery: Scaling LLM Agents to Thousands of Skills and Tools

Codebook Agent: Amortized Topology Design for LLM Multi-Agent Systems

SkillGLoW: Procedural-Family Skill Consolidation for Self-Improving Agents on Long-Horizon Task Streams

CHIME: Credit-Aware Hierarchical Memory Evolution for Long-Horizon Agentic Planning

Act More, Decide Less: Skill-Guided Adaptive Action Chunking for Long-Horizon LLM Agents

Defense-as-Skill: Evolving Runtime Guard Skill for Skill-Augmented Agents

Harness-of-Harness: Multi-Day Autonomous Software Development with Continual Improvement

Parsing the Stream: A Live Trace Model for Long-Horizon Agents and Their Observers

HarnessDev: Can LLMs Create and Evolve Their Own Agent Harness?

Where the Verifier Fails: A Category-Level Audit of Reward Signals in RLVR

Explore Before Committing: Hypothesis-Guided Search for Deep Research Agents

Making Prospective Memory SLM-Shaped: Typed Intention Stores for Small-Model Agents

MutMem-V2: Cryptographically Authorized Mutation in Persistent Agent Memory Portable Verification and Reproducible Evidence

AgentFactory: Towards Automated Agentic System Design and Optimization

HarnessEvolve: Learning from Reference Trajectories for Reliable Agent Self-Evolution

Skill Following: Evaluating Actual Skill Use in Retrieval-Enabled LLM Agents

Runtime-Independent Persistent Agents: Preserving Identity, Memory, and Code Across Models, Harnesses, and Servers

Token-Efficient Data Reasoning Agents via Adaptive Structuring of Unstructured Data

LOCI: A Locator-Critic with Refinement Loop

SkillZip Pro: Execution-Aware Dynamic Compression of Progressively Loaded Skills for Self-Evolving Agents

Agent Zero Memory: Provenance-Aware Long-Term Memory for LLM Agents

Can escalation channels redirect reward hacking toward defect disclosure?

Selective Forgetting: A Graph-Based Memory Framework for Long-Term LLM Agents

ContextPilot: Teaching Agents for Proactive Context Management via Fine-grained RL

EvoUndo: Recoverability-Constrained Self-Evolution for LLM Agent Harnesses

LoopArena: Benchmarking Models as Runtime Controllers for Loop Engineering

String: An Agentic OS Where Every App Is a Markdown File

openJiuwen: Beyond Static Harnesses for Long-Horizon Coding Agents

CURA: Certified Runtime Alarms for Computer-Use Agents

If Agents Were Angels, No Governance Would Be Necessary: Out-of-Band Policy Enforcement at a Trusted Tool Boundary

WikiSkill: Compiling Agent Experience into Persistent Knowledge for Skill Evolution

Naive Prompt Optimization: Rethinking the Need for Complex Prompt Search

Safety Does Not Compose: Non-Decaying Loop State for Autonomous LLM Agents

ASIL: Replacing Screenshot-and-Click with Structured State and Semantic Actions

Daydreaming: Stealing Hidden Agent Skills through Black-Box Task Interaction

PILOT in the Loop: Live Self-Improvement for Long-Horizon Agents

SKILL.state: Scalable Long-Horizon Agent Skills

Praxist: From Experimental Artifacts to Solution Lineages

Repair or Resample? Rethinking Failure Debugging in LLM Multi-Agent Systems

Same Model, Different Harness: Different Coding-Agent Results

From General Agents to RCA Experts: A Self-Evolving Harness for Root Cause Analysis

JIT-Agent: Scaling Harness Intelligence via Just-in-Time Harness Evolution

When Stale Constraints Go Unchecked: Budgeted Verification Failures in Inherited Agent Memory

A Few Pages of Markdown: Committed AI Configuration and Lower Quality Cost after Coding-Agent Adoption

Recursive Experiential-Working Memory Evolution for Long-Horizon Agent Harnesses

Paritok-4B: Intent-Conditioned Context Compression for Coding Agents

Prime Agent

The first of the night's three talks, and the clearest statement of the thesis. A persistent IPython REPL, recursive sub-agents that stay addressable after they finish, and continual refinement of prompts and skills. On identical weights it takes ARC-AGI-3 from 30% to 95.5%, which is the number the whole evening is arguing about.

What Compaction Destroys

Context Management as Code

Skill Lift

Bounded Loops: Pre-Run Spend Bounds, Proved Termination, and Verified Completion for Agent Harnesses

Success Leaves Detours: Learning Executable Walkthroughs for Long-Horizon Agents

Harness-Level Forgetting

Demystifying Agent Skills

Coding Agents are Strong Prompt Optimizers

The Skill Trigger Bottleneck

Harness-IF

Catastrophic Remembering

WatchPoint: Executable User Feedback for Real-World Agentic Web Development

Look Before You Leap: Pre-Action Verification for LLM Agents

Reason Wide, Not Deep

The Bitter Lesson of Tool Calling

ContinualSkillBench

Harness-R1

Lost in Compaction

Zero-Mem

Model or Harness

Rehearse

Filesystem Memory Audited

ACM

Silence Is Endorsement: Verification-Status Laundering in LLM Agent Pipelines

NOOA

PRO-LONG

Progressive Disclosure, Measured

From Memory to Skills

Bad Memory in Agents

Harness Handbook

Harness Evolution, Rethought

Failure as a Process

When Is Routing Meaningful

The Harness Effect

Agent Limitations Taxonomy

LLM-as-a-Verifier

Verification as a Scaling Axis

ReContext

Replicating ML Papers with Agents

BudgetBench: A Budget-Tiered Protocol and Pilot Harness for Memory Strategy Evaluation in Local Large Language Model Agents

AutoMem

ASPIRE

Generative Skill Composition

MCP Server Patterns

Always-On Agents

HORIZON

The Verification Horizon

Agent-Native Memory

Agent-as-a-Router

Sakana Fugu

AtomMem

Agent Communication Protocols

Skill-MAS

Compositional Skill Routing

PreAct

Beyond Domains

OpenClaw-Skill

SpatialClaw

Self-Harness

The Consistency Illusion

LEAP

State-Externalizing Harnesses

Do More Agents Help?

Reusable Context Engineering

Disentangling Agent Self-Evolution

Scaling Laws for Agent Harnesses

AutoScientists

Your Agents Are Aging Too

SkillOpt

Inductive Deductive Synthesis: Enabling AI to Generate Formally Verified Systems

The Efficiency Frontier

Adapting the Interface, Not the Model

Production Agent Architecture Methodology

Code as Agent Harness

MetaCogAgent

OpenJarvis: Personal AI, On Personal Devices

The same argument pointed at your laptop. Decompose the personal AI stack into five primitives, then let a frontier cloud model search over that spec while everything runs locally at inference. Roughly 800x lower marginal cost, with the harness rather than the model closing the gap.

Harnesses Are Not Uniformly Better

Is Grep All You Need?

Memory as a Model

Beyond Individual Intelligence

AEvo

LongMemEval-V2: Evaluating Long-Term Agent Memory Toward Experienced Colleagues

Continual Harness: Online Adaptation for Self-Improving Foundation Agents

The last step before online learning. Continual Harness keeps history, memory, skills, prompts, and sub-agent specs across trajectories and mutates them while the agent runs, then goes further and updates the weights DAgger-style from what just happened. The presenter calls test-time training the direction that matters most.

AI Co-Mathematician

Coordination as Architecture

Skills as Verifiable Artifacts

Contextual Agentic Memory is a Memo

When to Retrieve During Reasoning

OCR-Memory

RecursiveMAS

Agentic Harness Engineering

From Skill Text to Skill Structure

OneManCompany

Co-evolving Decisions and Skills

Stateless Decision Memory

MASS-RAG

Skill-RAG

Autogenesis

Memory Transfer Learning

WebXSkill

AiScientist

Memory Intelligence Agent (MIA)

Agent Skills in the Wild

The Universal Verifier for Agent Benchmarks

LightThinker++: From Reasoning Compression to Memory Management

Single-Agent LLMs vs. Multi-Agent Systems

MemFactory

Meta-Harness: End-to-End Optimization of Model Harnesses

The meta harness: a harness whose job is producing harnesses. A coding agent is handed the full search history, source, traces, and scores, and rewrites the retrieval, memory, and prompt-assembly code around a fixed model. State of the art on Terminal-Bench 2 without touching the weights.

On the Reliability Limits of LLM-Based Multi-Agent Planning

Natural-Language Agent Harnesses

MemCollab

Workflow Optimization for LLM Agents

Asynchronous Software Engineering Agents

BIGMAS

OpenDev

STRUCTUREDAGENT

Memex(RL)

Diagnosing Agent Memory

Auton Agentic AI Framework

SkillNet

Theory of Mind in Multi-Agent LLMs

ActionEngine

Codified Context

Learning to Rewrite Tool Descriptions

AgentConductor