interprocess communication

To address challenges in LLM-agent interoperability—including fragmented tool integration, weak context sharing, and inefficient task coordination across heterogeneous systems—this paper systematically analyzes four emerging interoperability protocols: MCP, ACP, A2A, and ANP. We propose the first cross-platform evaluation framework covering interaction patterns, service discovery mechanisms, communication paradigms, and security models. Our contributions include: (1) an open, decentralized service discovery mechanism leveraging Decentralized Identifiers (DIDs) and JSON-LD; (2) declarative Agent Cards for standardized capability description and enterprise-grade task delegation; and (3) a multidimensional comparative analysis with a phased adoption roadmap. The framework advances scalability, security, and cross-domain standardization for LLM-agent ecosystems, providing practitioners and researchers with a rigorous, implementation-ready guideline for building interoperable intelligent agent systems.

This work addresses the challenges of low coordination efficiency, lack of auditability, and insufficient policy compliance in multi-agent systems for complex tasks by proposing a unified orchestration framework that integrates planning, policy execution, state management, and observability mechanisms. The framework introduces two novel protocols—the Model Context Protocol and Agent2Agent Protocol—to enable standardized, extensible communication among agents and between agents and external tools. By supporting policy governance and end-to-end traceability, the proposed architecture provides a practical, transparent, and accountable blueprint for deploying large-scale multi-agent systems within enterprise AI ecosystems.

Current agent communication protocols generally lack mechanisms for semantic alignment, clarification, and verification, shifting semantic responsibility onto prompts or application logic and thereby causing poor interoperability and high maintenance costs. This work proposes, for the first time, a human-inspired three-layer communication framework—comprising communication, syntactic, and semantic layers—and systematically analyzes 18 mainstream protocols to expose their structural deficiencies in semantic coordination. Through layered modeling, technical debt identification, and scenario mapping, the study not only derives a practical protocol selection guide but also advances agent communication beyond mere message passing toward a new paradigm of shared understanding, laying the foundation for building semantically robust, secure, and interoperable agent ecosystems.

This work proposes a lightweight, file system–based alternative to conventional multi-agent frameworks, which often incur redundant engineering overhead in sequential human-in-the-loop workflows. The approach represents workflow stages as numbered directories, stores prompts and contextual information in Markdown files, delegates non-AI tasks to local scripts, and employs a single agent to perform complex reasoning sequentially. Drawing inspiration from Unix pipelines, modular decomposition, multi-pass compilation, and literate programming, the architecture achieves high interpretability, readability, and maintainability. Released under the MIT license, this open-source solution substantially reduces the implementation complexity of sequential workflows while effectively replicating the core functionalities of multi-agent systems.

This work addresses the challenges faced by large language model (LLM) agents operating over flat tool registries—namely, combinatorial explosion in decision space, context saturation, and degraded routing accuracy. To overcome these limitations, the authors propose a skill-tree-based hierarchical architecture that separates routing logic at internal nodes from execution at leaf nodes. Inspired by pushdown automata, the framework incorporates a LIFO stack-frame memory model and a lazy capability discovery mechanism, enabling isolated execution paths and scalable context management. The approach supports manifest-driven single-step execution loops and formal state modeling, significantly improving routing accuracy while reducing memory footprint and prompt costs under conditions of tool proliferation, multi-step workflows, and prompt exposure. This design meets enterprise-grade requirements for isolation and scalability.

This work addresses the limitations of traditional workflow platforms, which rely on static, pre-defined processes and struggle to accommodate the dynamic data integration demands of distributed systems. To overcome this, the authors propose a configuration-driven runtime orchestration framework that dynamically constructs execution graphs at request time through dependency-aware scheduling and parallel task execution, thereby circumventing the constraints of fixed workflows. This approach enables rapid adaptation to evolving integration scenarios without requiring system redeployment, significantly reducing latency. Empirical evaluation in a real-world Customer 360 enterprise use case demonstrates that the framework offers substantial advantages in flexibility, scalability, and efficient data aggregation compared to conventional solutions.

This work addresses the challenge of accurately interpreting ambiguous or incomplete user intents in highly complex 6G networks, where existing intent-driven systems often fall short. To bridge this gap, the paper proposes an agent-based end-to-end orchestration framework that leverages domain-expert agents collaborating with the TM Forum knowledge ecosystem to iteratively refine high-level intents into precise, machine-readable service directives. The approach innovatively decouples cognition from execution and introduces a dual-layer memory mechanism to ensure consistency across multi-turn interactions. Furthermore, the authors develop the first open-source large language model evaluation prototype tailored for 6G Network-as-a-Service (NaaS). Experimental results demonstrate that while current open-source models exhibit strong compliance in instruction following, they still show a significant deficiency in accurately and hallucination-free mapping of fine-grained intents into valid, catalog-supported service orders.

This study addresses the challenge of ensuring traceability, controllability, and correctness of large language model–driven agents within business processes while preserving their autonomy. To this end, the work proposes the first multidimensional attribute classification framework specifically designed for agent orchestration, integrating principles from business process management to strike a balance between agent autonomy and system robustness. Complementing the framework, the authors introduce qualitative decision-making guidelines and quantitative evaluation metrics. The efficacy of the proposed approach is empirically validated through multi-agent experiments in a predictive light-sensing scenario, demonstrating its capacity to support both theoretical inquiry and practical deployment of orchestrated intelligent agents in real-world applications.

This work addresses the structural security risks posed by large language model–driven autonomous agents to traditional operating systems, whose “resource exposure plus permission check” model proves inadequate—once compromised, attackers can abuse low-level resources to perform privilege-escalated operations. To mitigate this, the paper proposes AgenticOS, an intent-centric secure operating system architecture that treats structured agent intents as the entry point for system calls. The kernel synthesizes a least-privilege execution environment and enforces mandatory mediation, end-to-end auditing, and information flow control. Built upon a four-layer design—comprising the Ghost Kernel, Logic Shutter, Agent Capsule, and Semantic Boundary Gateway—and leveraging an Intent ABI, Manifest-Only Runtime, and Weaver capability mechanism, AgenticOS redefines the OS role from resource manager to intent filter, enabling semantic-level security governance of AI behaviors and substantially reducing the risk of resource misuse following agent hijacking.

This study addresses the lack of systematic investigation into architectural design decisions for non-large language model components in current AI agent systems. The authors propose a protocol-guided, source code–driven empirical analysis method that enables, for the first time, transparent deconstruction of heterogeneous AI agent systems. Through cross-project qualitative coding and co-occurrence analysis of 70 open-source projects, they identify five core design dimensions—sub-agent architecture, context management, tooling systems, security mechanisms, and orchestration—and uncover their combinatorial patterns. Based on these findings, the study further distills five archetypal architectural patterns: lightweight tool-oriented, CLI framework–based, multi-agent orchestrator, enterprise system, and domain-specific vertical architectures.

This work addresses the limitations of existing 6G network assurance mechanisms, which lack precise understanding and proactive validation of user intent, thereby struggling to guarantee end-to-end SLA compliance. To bridge this gap, the paper proposes the first agent orchestration framework that integrates test-driven development (TDD) principles into 6G assurance. The framework leverages multi-agent collaboration to decompose user intent into auditable specifications and automatically generates verification tests prior to resource allocation, establishing a deterministic mapping from intent to test. Built upon the TM Forum information model, a message-driven architecture, the MCP protocol, and open-source large language models (LLMs), the prototype implementation reveals significant disparities among LLMs in tool-calling reliability and hallucination control, underscoring the critical role of domain knowledge integration in enhancing agent system effectiveness.