A Survey of Agent Interoperability Protocols: Model Context Protocol (MCP), Agent Communication Protocol (ACP), Agent-to-Agent Protocol (A2A), and Agent Network Protocol (ANP)

Abul Ehtesham, Aditi Singh, Gaurav Kumar Gupta, Saket Kumar · 2025

Read A Survey of Agent Interoperability Protocols: Model Context Protocol (MCP), Agent Communication Protocol (ACP), Agent-to-Agent Protocol (A2A), and Agent Network Protocol (ANP) by Abul Ehtesham, Aditi Singh, Gaurav Kumar Gupta, Saket Kumar (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-02279.txt.

Introduction

Ehtesham, Singh, Gupta, and Kumar surveyed four protocols that let AI agents work together and use tools in a standard way. As agents built on large language models spread, ad-hoc links between them have become hard to scale and secure. The survey compares the Model Context Protocol, the Agent Communication Protocol, the Agent-to-Agent Protocol, and the Agent Network Protocol, each aimed at a different layer of cooperation.

The protocols differ in how agents find each other, pass messages, and stay secure. One focuses on safe tool access, another on general messaging over the web, a third on delegating tasks between peers, and the fourth on open, decentralized networks. The authors propose adopting them in phases, from tool access up to open agent marketplaces. They present the comparison as a base for building secure, interoperable agent systems.

Catalog record

CREATORAbul Ehtesham, Aditi Singh, Gaurav Kumar Gupta, Saket Kumar
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE66 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-02279

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A Survey of Agent Interoperability Protocols: Model Context Protocol (MCP), Agent Communication Protocol (ACP), Agent-to-Agent Protocol (A2A), and Agent Network Protocol (ANP) Abul Ehtesham, Aditi Singh, Gaurav Kumar Gupta, Saket Kumar Abstract Large language model (LLM)-powered autonomous agents demand robust, standardized protocols to integrate tools, share contextual data, and coordinate tasks across heterogeneous systems. Ad-hoc integrations are difficult to scale, secure, and generalize across domains. This survey examines four emerging agent communication protocols: Model Context Protocol (MCP), Agent Communication Protocol (ACP), Agent-to-Agent Protocol (A2A), and Agent Network Protocol (ANP), each addressing interoperability in distinct deployment contexts. MCP provides a JSON-RPC client-server interface for secure tool invocation and typed data exchange. ACP defines a general-purpose communication protocol over RESTful HTTP, supporting MIME-typed multipart messages and both synchronous and asynchronous interactions. Its lightweight and runtime-independent design enables scalable agent invocation, while features like session management, message routing, and integration with role-based and decentralized identifiers (DIDs).A2A enables peer-to-peer task delegation using capability-based Agent Cards, supporting secure and scalable collaboration across enterprise agent workflows. ANP supports open-network agent discovery and secure collaboration using W3C decentralized identifiers (DIDs) and JSON-LD graphs. The protocols are compared across multiple dimensions, including interaction modes, discovery mechanisms, communication patterns, and security models. Based on the comparative analysis, a phased adoption roadmap is proposed: beginning with MCP for tool access, followed by ACP for structured, multimodal messaging, session-aware interaction, and both online and offline agent discovery across scalable, HTTP-based deployments, A2A for collaborative task execution, and extending to ANP for decentralized agent marketplaces. This work provides a comprehensive foundation for designing secure, interoperable, and scalable ecosystems of LLM-powered agents. K eywords Large Language Models (LLMs), Agent Communication, Interoperability Protocols, Model Context Protocol (MCP), Agent Communication Protocol (ACP), Agent-to-Agent Protocol (A2A), Agent Network Protocol (ANP), Autonomous Agents, Multimodal Messaging, Decentralized Identity (DID) , peer-to-peer 1 Introduction Large Language Models (LLMs) have become central to modern artificial intelligence, powering autonomous agents that operate across cloud, edge, and desktop environments [ 1 , 2 ] . These agents [ 3 ] ingest contextual information, execute tasks, and interact with external services or tools. However, inconsistent and fragmented interoperability practices make it difficult to integrate, secure, and scale communication among LLM-driven agents [ 4 ] . Interoperability (the ability of distinct agents and systems to discover capabilities, exchange context, and coordinate actions seamlessly) is essential for modular, reusable, and resilient multi-agent [ 5 ] workflows. Standardized protocols reduce development overhead, improve security, and enable cross-platform collaboration. Clear, universally adopted standards remain nascent. This survey examines four emerging agent communication protocols, each targeting a different interoperability tier: • Model Context Protocol (MCP) : A JSON-RPC client–server interface for secure context ingestion and structured tool invocation. MCP streamline the integration of large language models (LLMs) with external data sources and tools. MCP addresses the challenges of fragmented and custom-built integrations by providing a universal, model-agnostic interface for AI systems to access and interact with diverse resources [ 6 , 7 , 8 ] . MCP was launched by Anthropic in Novemeber 2024. • Agent-to-Agent Protocol (A2A) : Enable secure, structured, and interoperable collaboration between AI agents across platforms, vendors, and environments. A2A is designed to support peer-to-peer agent interactions using capability-based representations known as Agent Cards, which describe what an agent can do and how it can be securely invoked. A2A supports asynchronous, event-driven communication through HTTP and Server-Sent Events (SSE), making it suitable for distributed, scalable agent ecosystems, [ 9 ] . A2A was launched by Google in April 2025. • Agent Communication Protocol (ACP) : ACP is a general-purpose protocol for agent communication that uses RESTful HTTP interfaces to support MIME-typed multipart messages and both synchronous and asynchronous interactions. This REST-based communication model enables lightweight, runtime-independent agent invocation, making ACP well-suited for scalable system integration. ACP includes structured session management, message routing, and a flexible authentication model that integrates with role-based access control (RBAC) and decentralized identity (DID) systems. It is compatible with a wide range of agent frameworks and deployment models, from lightweight stateless utilities to long-running, stateful services. ACP also supports agent discovery through runtime APIs, offline packaging, and manifest-based metadata. ACP was launched by IBM in March 2025 [ 10 ] . • Agent Network Protocol (ANP) : The Agent Network Protocol (ANP) is an open-source communication framework designed to enable secure, decentralized collaboration among AI agents across the open internet. Unlike traditional client-server architectures, ANP adopts a peer-to-peer (P2P) model, allowing agents to autonomously discover, authenticate, and interact with one another without centralized intermediaries [ 11 , 12 ] . Architectural details, integration approaches, communication patterns, and security considerations are reviewed for each protocol. A comparison highlights trade-offs in interaction modes, discovery mechanisms, communication models, and security frameworks. A phased adoption roadmap sequences MCP, A2A, ACP, and ANP to guide progressive deployment in real-world agent ecosystems. The remainder of the paper is organized as follows. Section 2 discusses challenges in agent interoperability. Section 3 reviews background and related work. Sections 4–7 describe the architectures of MCP, A2A, ACP, and ANP, respectively. Section 8 presents the comparative evaluation. Section 9 outlines the phased adoption roadmap. Section 10 concludes and suggests future research directions. 2 Challenges and Solutions in Agent Protocol Interoperability Despite the emergence of multiple open protocols like MCP, ACP, A2A, and ANP, achieving seamless agent interoperability in real-world AI systems remains a non-trivial task. This section identifies key challenges encountered in agent-based architectures and highlights how each protocol addresses them with purpose-built design principles. Lack of Context Standardization for LLMs: Large Language Models (LLMs) require contextual grounding to produce accurate outputs. However, existing application architectures provide no unified mechanism to deliver structured context to LLMs, leading to ad hoc tool integrations and unreliable behavior. Solution: The Model Context Protocol (MCP) addresses this by standardizing how applications deliver tools, datasets, and sampling instructions to LLMs, akin to a USB-C for AI. It supports flexible plug-and-play tools, safe infrastructure integration, and compatibility across LLM vendors. Figure 1: Protocol-aligned solution to challenges in agent communication Communication Barriers Between Heterogeneous Agents: Enterprise systems often consist of agents built using different stacks and frameworks, resulting in isolated behavior and poor collaboration. Solution: The Agent Communication Protocol (ACP) offers a RESTful, SDK-optional interface with open governance under the Linux Foundation. It enables asynchronous-first interactions, offline discovery, and vendor-neutral execution, bridging interoperability gaps at scale. Absence of Unified Agent Collaboration Standards: Even when agents communicate, there’s no shared framework for dynamic negotiation, capability sharing, and coordination. Solution: The Agent2Agent (A2A) protocol introduces a multimodal communication standard to unlock dynamic interaction between opaque, autonomous agents—regardless of framework. It simplifies enterprise integration and supports shared task management and user experience negotiation. Internet-Agnostic Agent Communication: The modern internet is optimized for human interaction but suboptimal for autonomous agents, which require low-latency, API-native communication and decentralized identity validation. Solution: The Agent Network Protocol (ANP) provides a layered protocol architecture incorporating decentralized identity (W3C DID), semantic web principles, and encrypted communication to facilitate cross-platform agent collaboration over the open internet. Together, these protocols aim to transform fragmented AI ecosystems into robust, secure, and interoperable agent networks scalable across organizational and vendor boundaries. See Table 7 for a detailed comparative overview. 3

Read the complete text