Title: Agents-K1: Towards Agent-native Knowledge Orchestration

Source: http://arxiv.org/abs/2606.13669v1

Summary:

This work introduces a novel end-to-end pipeline for converting raw scientific documents into 'agent-native' knowledge graphs, moving beyond simple abstract-based retrieval to enable complex scientific reasoning. By establishing a unifying theoretical foundation for multimodal parsing and cross-document traversal, it provides a new architectural primitive for scientific knowledge orchestration in agentic systems.

Title: Ouroboros-Spatial: Closing the Data-Model Loop for Spatial Reasoning

Source: http://arxiv.org/abs/2606.11719v1

Summary:

This paper introduces a self-evolving training framework that enables models to co-evolve their training distribution with their own capabilities by acting as both proposer and solver. It represents a significant breakthrough in efficiency and reasoning by closing the data-model loop, allowing for substantial performance gains with an order of magnitude less training data.

Title: LLMs+Graphs: Toward Graph-Native, Synergistic AI Systems

Summary:

This paper defines the architectural principles for next-generation graph-native AI systems by synergizing LLMs with structured graph computation for grounded, multi-hop reasoning. It establishes a unified framework for LLM-augmented retrieval, knowledge graph integration, and graph-algorithm-strengthened agentic planning.Source: http://arxiv.org/abs/2606.11560v1

Title: A Five-Plane Reference Architecture for Runtime Governance of Production AI Agents

Source: http://arxiv.org/abs/2606.12320v1

Summary:

This paper proposes a formal Five-Plane Reference Architecture for agent governance, defining critical architectural primitives like reasoning planes and composite principals for autonomous systems. It provides the foundational structural framework necessary for the runtime management and secure deployment of production-level agentic infrastructure.

Title: The Power of Test-Time Training for Approximate Sampling

Source: http://arxiv.org/abs/2606.11437v1

Summary:

This work provides a foundational theoretical framework for Test-Time Training (TTT) by formalizing it as an approximate sampling problem and establishing query complexity lower bounds. It offers principled mathematical backing for the emerging paradigm of adapting model weights at inference time to solve complex reasoning tasks.

Title: Role-Agent: Bootstrapping LLM Agents via Dual-Role Evolution

Source: http://arxiv.org/abs/2606.10917v1

Summary:

The Role-Agent framework introduces a dual-role evolution cycle where an LLM functions as both agent and environment to bootstrap its own reasoning capabilities. This novel agentic framework addresses the limitations of static training data by enabling self-correcting co-evolution through internal state prediction and failure mode analysis.

Title: ReasonAlloc: Hierarchical Decoding-Time KV Cache Budget Allocation for Reasoning Models

Source: http://arxiv.org/abs/2606.11164v1

Summary:

ReasonAlloc introduces a hierarchical KV cache allocation strategy that significantly optimizes memory usage during the long chain-of-thought trajectories characteristic of modern reasoning models. By identifying "Reasoning Wave" demand patterns, this training-free framework provides a foundational primitive for scaling inference efficiency in complex reasoning tasks.

Title: Experience Makes Skillful: Enabling Generalizable Medical Agent Reasoning via Self-Evolving Skill Memory

Source: http://arxiv.org/abs/2606.09365v1

Summary:

This paper proposes SkeMex, a foundational architecture for self-evolving agent memory that enables the distillation and governance of procedural skills from interaction trajectories. It introduces a novel "Read-Write-Assess-Govern" reasoning loop that allows agents to continuously improve their capabilities post-deployment without the need for model retraining.

Title: End-to-End Context Compression at Scale

Source: http://arxiv.org/abs/2606.09659v1

Summary:

This paper introduces Latent Context Language Models (LCLMs), a novel architectural primitive that utilizes encoder-decoder compression to efficiently handle long-context sequences at scale. It establishes a new Pareto frontier for accuracy and efficiency, providing a foundational backbone for next-generation agents that require massive context windows.

Title: Capability-Aligned Hierarchical Learning for Tool-Augmented LLMs

Source: http://arxiv.org/abs/2606.09371v1

Summary:

This paper proposes Capability-Aligned Hierarchical Learning (CAHL), a novel framework that jointly optimizes high-level planning and low-level execution policies using reinforcement learning. It addresses the fundamental bottleneck of planner-executor misalignment, creating a more robust and foundational reasoning loop for tool-augmented agentic systems.