LLM Agent Communication Protocol (LACP) Requires Urgent Standardization: A Telecom-Inspired Protocol is Necessary

Basic Information

• Paper ID: 2510.13821
• Title: LLM Agent Communication Protocol (LACP) Requires Urgent Standardization: A Telecom-Inspired Protocol is Necessary
• Authors: Xin Li, Mengbing Liu, Chau Yuen (School of Electrical and Electronic Engineering, Nanyang Technological University)
• Classification: cs.NI (Network and Internet Architecture)
• Publication Venue/Conference: NeurIPS 2025 AI4NextG Workshop
• Paper Link: https://arxiv.org/abs/2510.13821

Abstract

This position paper argues that the LLM agent domain requires a unified, telecommunication-inspired communication protocol to ensure security, interoperability, and scalability, particularly in the context of next-generation (NextG) networks. Current ad-hoc communication methods are creating a fragmented ecosystem, similar to the "protocol wars" of early networking, which impedes innovation and poses significant risks. Drawing inspiration from modern telecommunications standardization and layered protocols, the authors propose the LLM Agent Communication Protocol (LACP). LACP establishes a three-layer architecture designed to ensure semantic clarity in communication, transactional integrity for complex tasks, and robust built-in security.

Research Background and Motivation

Core Challenges

Current LLM agent communication faces three critical challenges:

1. Interoperability Gap: Lack of universal standards necessitates customized, fragile integrations between different agent systems
2. Security Deficiency: Existing protocols treat security as an afterthought rather than a core component, exposing risks such as data tampering and agent spoofing
3. Monolithic Design: Communication logic is tightly coupled with agent core implementation, lacking atomic transaction support, which compromises reliability of complex multi-step operations

Problem Significance

With the rapid growth of LLM agents in autonomous systems, scientific discovery, and other domains, multi-agent collaboration capabilities have become critical for practical deployment. The current fragmented ecosystem mirrors the "protocol wars" of the 1970s-1990s, which hindered computer network development before TCP/IP became widely adopted.

Limitations of Existing Approaches

The paper analyzes seven existing protocols (OpenAI Function Calling, LangChain Agent Protocol, Anthropic MCP, ACP, ANP, Agora, Google A2A) and identifies common deficiencies:

• Lack of cross-framework interoperability
• Incomplete security mechanisms
• Absence of transactional integrity guarantees
• Homogeneous architectural design

Core Contributions

1. Risk Analysis: Identifies systemic risks arising from the current fragmented state of agent communication
2. Design Principles: Extracts core guiding principles from telecommunications history for building robust, interoperable agent protocols
3. Protocol Proposal: Proposes LACP—a layered, secure, and scalable LLM agent interoperability framework
4. Critical Discussion: Anticipates and refutes key objections to standardization

Methodology Details

Task Definition

Design a universal inter-agent communication protocol for LLM agents that achieves:

• Cross-framework semantic interoperability
• End-to-end security assurance
• Transactional integrity
• Scalable architectural design

LACP Three-Layer Architecture

1. Semantic Layer

• Function: Conveys communication intent
• Design: Defines a minimal set of universal message types, adopting the "narrow waist" principle
• Core Message Types:
  • PLAN: Expresses high-level intent (intent_id, role, natural_language, graph_ops)
  • ACT: Invokes external tools (intent_id, tool_call, params, deadline, cost_cap)
  • OBSERVE: Returns result status (intent_id, status, output, metrics)

2. Transactional Layer

• Function: Ensures communication reliability and integrity
• Key Mechanisms:
  • Message signing and verification
  • Sequence number management
  • Unique transaction IDs (idempotency guarantee)
  • Atomic transaction support (two-phase commit concept)
  • JWS (JSON Web Signature) encapsulation

3. Transport Layer

• Function: Efficient and secure message transmission
• Characteristics: Transport protocol agnostic, supports HTTP/2, QUIC, WebSockets, etc.
• Components: Binary frame headers, checksum mechanisms

Technical Innovations

1. Telecommunication-Inspired Layered Design: Borrows from the OSI model to achieve separation of concerns and independent layer evolution
2. "Narrow Waist" Principle: Provides universal core while allowing edge innovation
3. Security-First Approach: Integrates security as a core protocol feature rather than an add-on
4. Transactional Integrity: Built-in atomic operation support ensures reliable execution of complex tasks

Experimental Setup

Experimental Environment

• Platform: Local machine
• Server: Python 3.11 + Flask framework
• Cryptography Library: python-jose
• JSON Processing: orjson (high-performance)
• Client: Custom Python script based on requests library
• Interoperability Testing: langchain library (v0.1.20)

Experimental Design

Three core experiments validate LACP feasibility:

1. Performance Overhead Analysis: Quantifies latency and payload overhead of LACP relative to standard REST API
2. Interoperability Demonstration: Verifies LACP communication capability across different frameworks
3. Security Validation: Tests security guarantees of LACP's transactional layer

Experimental Results

Performance Overhead Analysis

Key Findings:

• Latency overhead is minimal, with absolute increases of only 0.03ms across all scenarios
• Payload size overhead decreases with message complexity, adding only 30% for large complex tasks
• Demonstrates LACP's feasibility for practical applications

Interoperability Demonstration

Experimental Setup:

• LACP tool server: Independent Flask server accepting only signed LACP messages
• LangChain agent client: Standard ReAct agent equipped with custom LACP tool

Results: LangChain agent successfully and transparently uses framework-agnostic external tools to complete computational tasks. The entire interaction is secured and structured by LACP without requiring framework-specific integration code.

Security Validation

Tampering Attack Test

• Attack Method: Modifying signed message content (amount changed from 100 to 10000)
• Protection Result: Server immediately detects signature mismatch, returns HTTP 403 Forbidden

Replay Attack Test

• Attack Method: Resending identical valid messages
• Protection Result: Server identifies duplicate transaction_id, returns HTTP 409 Conflict

Security Validation Conclusion: LACP provides application-layer security guarantees beyond what TLS can cover, including end-to-end message integrity and idempotency assurance.

Related Work

Existing Protocol Comparison

The paper provides detailed analysis of seven mainstream agent communication protocols:

(×: Not supported, ◦: Partially supported, •: Supported)

Telecommunications Protocol Evolution Insights

The paper traces protocol evolution from 1G to 6G:

• 1G: Analog cellular networks, basic channel-based voice
• 2G: Digital identity and security (GSM)
• 3G: Soft handover state machines (UMTS/WCDMA)
• 4G: Bearer abstraction and all-IP core (LTE)
• 5G: Service-based architecture
• 6G: Distributed cognitive protocols

Conclusions and Discussion

Main Conclusions

1. Standardization Urgency: Standardization of LLM agent communication is not merely beneficial but essential for realizing the potential of distributed AI
2. LACP Feasibility: Experiments validate LACP's feasibility in performance, security, and interoperability
3. Applicability of Telecommunications Experience: Successful telecommunications standardization experiences can effectively guide agent protocol design

Limitations

1. Experimental Scale: Current experiments conducted only in local environments, lacking large-scale distributed validation
2. Protocol Completeness: LACP remains a conceptual proposal requiring more detailed technical specifications
3. Ecosystem Adoption: Standardization requires broad industry collaboration and adoption

Future Directions

1. Standardization Organization Establishment: Requires collaborative institutions similar to ITU and 3GPP
2. Protocol Refinement: Detailed technical specifications and expanded message types
3. Large-Scale Validation: Testing protocol performance in real distributed environments
4. Security Enhancement: Further refinement of security mechanisms and threat models

In-Depth Evaluation

Strengths

1. Accurate Problem Identification: Clearly identifies core issues in current agent communication
2. Appropriate Historical Analogy: Telecommunications protocol evolution provides a compelling argumentative foundation
3. Scientific Design Principles: Layered architecture and security-first principles are validated through practice
4. Comprehensive Experimental Validation: Demonstrates feasibility through performance, interoperability, and security experiments
5. Strong Foresight: Provides a framework for agent communication in the 6G era

Weaknesses

1. Insufficient Implementation Details: LACP's specific technical specifications require further development
2. Incomplete Scalability Validation: Lacks performance data from large-scale deployments
3. Unclear Standardization Path: Strategy for promoting industry adoption of standards needs clarification
4. Insufficient Integration Consideration: Migration costs and compatibility issues with existing systems underexplored

Impact

1. Academic Value: Provides important theoretical foundation for agent communication standardization
2. Practical Value: LACP framework can guide actual protocol design
3. Industry Impact: May promote agent communication standardization efforts
4. Reproducibility: Provides complete experimental code and data

Applicable Scenarios

1. Security-Critical Applications: High-risk domains such as finance, healthcare, and autonomous systems
2. Large-Scale Multi-Agent Systems: Complex tasks requiring reliable coordination
3. Cross-Platform Integration: Interoperability of agent systems from different vendors
4. NextG Network Applications: Intelligent network services in the 6G era

References

The paper cites extensive references including:

• Telecommunications protocol standards (3GPP, ITU, etc.)
• Network protocol history (TCP/IP evolution)
• Current agent frameworks (OpenAI, LangChain, Anthropic, etc.)
• Wireless communication technology evolution (1G to 6G)

Overall Assessment: This is a position paper of significant practical importance that accurately identifies the urgent need for standardization of LLM agent communication and proposes a solution based on telecommunications protocol experience. The LACP protocol design is sound, experimental validation is comprehensive, and it provides valuable theoretical foundation and practical guidance for promoting agent communication standardization. Although implementation details and standardization pathways require further refinement, its foresight and practicality make it an important contribution to the field.