In the artificial intelligence space, as we transition from isolated large language models to autonomous agents capable of complex reasoning and tool use. While foundational architectures and local context management protocols have been established, the challenge of cross-platform, decentralized, and secure interaction remains a significant barrier to the realization of a truly Agentic Web. Building upon the foundations of AI agent architectures and the Model Context Protocol (MCP) for multi-agent coordination, this paper introduces the Agent Communication Protocol (ACP). ACP provides a standardized framework for Agent-to-Agent (AA) interaction, enabling heterogeneous agents to discover, negotiate, and execute collaborative workflows across disparate environments. We propose a federated orchestration model that integrates decentralized identity verification, semantic intent mapping, and automated service-level agreements. Our evaluation demonstrates that ACP reduces inter-agent communication latency by % while maintaining a zero-trust security posture. This work represents a critical advancement toward a scalable and interoperable ecosystem of autonomous digital entities

翻译：在人工智能领域，我们正从孤立的大型语言模型向具备复杂推理与工具使用能力的自主智能体过渡。尽管基础架构与本地上下文管理协议已初步建立，但跨平台、去中心化且安全的交互问题，仍是实现真正"智能体网络"（Agentic Web）的主要障碍。本文基于现有AI智能体架构与多智能体协调的模型上下文协议（MCP），提出了智能体通信协议（ACP）。ACP为智能体间（A2A）交互提供了标准化框架，使异构智能体能够在不同环境中实现服务发现、协商与协同工作流执行。我们提出一种联邦式编排模型，该模型集成了去中心化身份验证、语义意图映射与自动化服务等级协议。实验评估表明，ACP在维持零信任安全态势的同时，将智能体间通信延迟降低了%。这项工作标志着向可扩展、可互操作的自主数字实体生态系统迈出了关键一步。