cs.AIOct 28, 2026

Agent Transfer Protocol: Server-Mediated Agent-to-Agent Communication Architecture for the Agentic Web

Authors: Yuqi Qiu, Zuyao Xu, Lu Sun, Xiang Li, Fasheng Miao, Fubin Wu, Haixin Duan, Zheli Liu

Organizations: Nankai University Tianjin, China · Tsinghua University Beijing, China · Quancheng Laboratory Jinan, China

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

The Agentic Web needs to support continuous collaboration among agents from different organizations. These collaborative relationships may span multiple tasks, network connections and execution instances. Such interactions require stable communication entries, message reception when instances are offline, and source verification after delayed reading. How to organize these responsibilities so that cross domain communication does not depend on specific execution instances being continuously online is a key issue for building long term agent interaction infrastructure. We propose ATP (Agent Transfer Protocol), a cross organization agent communication protocol adopting a domain mediated architecture. In this architecture, administrative domains maintain stable addresses and receiving entries for logical agents. The domain servers of both parties continuously undertake cross domain transmission. The receiving domain verifies and saves messages for execution instances to read later. The sending domain signs the complete message envelope. This allows the receiving instance to independently reverify the domain source and content integrity after the original connection ends, provided the original verification public key is still trusted and obtainable. We implemented an ATP prototype and evaluated it through a controlled cross organization LLM agent workflow, mechanism checks and latency measurements. The workflow completed subsequent business processing after instances went offline or were replaced. The independent verifier successfully verified the domain source and content integrity of saved messages. Latency measurements further quantified the communication and processing overhead introduced by the continuous mediation of domain servers.

Figures & tables

Figure 1. ATP cross-domain message path. The sending domain authenticates the local sender, signs the envelope, and discovers the receiving domain through DNS. The receiving domain verifies and stores the message for later retrieval by the destination agent. The sending agent submits a message to its domain server, which discovers and authenticates the receiving domain and delivers the message to the recipient inbox for later retrieval.
VariantOfflineMigration
Success5/55/5
Total Time (s)407.6406.7
Offline (s)360.7361.1
Model (s)35.034.3
Business Tool (s)0.180.18
ATP send/recv (s)0.13 / 0.080.13 / 0.10
Table 1. Travel coordination results (medians over five runs per variant).
Figure 2. Travel coordination timelines. Rows show offline and migration, with five runs each; columns show two time windows from the booking instruction. The interval from 90 to 375 seconds is omitted. Dark bars mark absence; bold rows identify the runs closest to each variant’s median total time. A two-by-two grid shows all five runs of each variant. The top row is offline and the bottom row is migration; the left column spans zero to ninety seconds and the right column spans three hundred seventy-five to four hundred seventy-five seconds. A shared legend appears below the panels.
CheckObservation Result for Each TimePass
Retrieve after offline (600 seconds)20/20 messages received, no missing, duplication or content error5/5
Address unchanged after migration20/20 new messages fetched on new IP, old batch not fetched repeatedly5/5
Independent reverification of signature20/20 signatures passed verification5/5
Tampering detection60/60 modified envelopes rejected5/5
Two agents on one server20/20 messages each, no inbox mixing5/5
Table 2. Mechanism checks and observation results, each item runs 5 times.
Figure 3. End-to-end latency CDFs at two loads, with shared logarithmic axes. Dotted guides mark p50 and p99. Two side-by-side CDF panels compare direct HTTPS, same-domain ATP, and cross-domain ATP at two and twenty messages per second. Both panels use identical logarithmic latency limits and share a path-color legend below.
Figure 4. Runtime latency breakdown. Open markers denote p50; segment ends denote p99. Two side-by-side panels show the median-to-ninety-ninth-percentile latency intervals for cross-domain path segments and receiving-server processing stages. A shared legend below identifies the two traffic loads.
Test CaseRejected by Which Check (Status Code)Pass
Forged signatureSignature check (403)3/3
Tampered contentSignature check (403)3/3
Use other domain keyDomain consistency (403)3/3
Unauthorized source addressATS policy (403)3/3
Forged source request headerATS policy (403)3/3
ReplayReplay prevention cache (400)3/3
Table 3. Security checks. Each test case runs 3 times, all are rejected by expected mechanisms, messages do not enter the inbox.
Figure 5. Delivery under receiver outages, with three runs per duration. The x-axis is send time relative to outage start. The staircase predicts delivery delay; crosses mark expired messages at the 290 s cutoff. The orange circle marks the single prediction mismatch. Shading indicates the outage. Three equal-height panels in one row across both columns show all messages during thirty-, two-hundred-forty-, and two-hundred-seventy-second outages. Predicted delay, recovery, expired messages, repeated runs, send timing, and the single mismatch are explained by a shared legend below the panels.
Appendix figures & tables3 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 6. ATP interaction flows: message delivery and custody transfer, request/response deadlines and matching, and the subscription lifecycle. Three equal-height diagrams appear side by side across both columns. They illustrate asynchronous message delivery, request and response with deadline checks, and event subscription with renewal, cancellation, and expiry.
ResponseMeaningSender action
202Written to the recipient’s inboxDelivered
Network error, 408, 429, 5xxTemporary failureRetry with backoff within the window
400 TIMESTAMP_OUT_OF_WINDOW , 504 DEADLINE_EXCEEDEDExpiredExpired
400 NONCE_REPLAYEDAlready acceptedDelivered if the last attempt was uncertain, else permanent failure
Other 4xx (e.g., 403 bad signature, 404 unknown recipient)Permanent failureStop
Appendix
Table 4. Responses of the receiving domain and handling by the sending domain.
CaseConstructionRejected byResponseInboxPassed
Forged signatureSign with a random keyATK signature verification403 ATK_SIGNATURE_INVALIDAbsent3/3
Tampered contentModify a captured payloadATK signature verification403 ATK_SIGNATURE_INVALIDNo new copy3/3
Sender/key domain mismatchSign noats.example as evil.exampleDomain alignment403 ATK_SIGNATURE_INVALIDAbsent3/3
Disallowed source addressReplay from attacker hostATS policy403 ATS_VALIDATION_FAILEDNo new copy3/3
Spoofed source headersForge X-Forwarded-For and X-Real-IPATS policy403 ATS_VALIDATION_FAILEDNo new copy3/3
ReplayResend the same sender/nonceReplay cache400 NONCE_REPLAYEDNo new copy3/3
Appendix
Table 5. Adversarial validation. Every case was run three times; the receiver’s inbox was checked after each rejected request.
More in this topic

Agent Communication Protocols

Agent communication protocols define how AI agents discover one another, exchange messages and state, coordinate tasks, and enforce trust.

See all 49 papers

Explore similar work

CardsList
Feb 24, 2026

AWCP: A Workspace Delegation Protocol for Deep-Engagement Collaboration across Remote Agents

The rapid evolution of Large Language Model (LLM)-based autonomous agents is reshaping the digital landscape toward an emerging Agentic Web, where increasingly specialized agents must collaborate to accomplish complex tasks. However, existing collaboration paradigms are constrained to message passing, leaving execution environments as isolated silos. This creates a context gap: agents cannot directly manipulate files or invoke tools in a peer's environment, and must instead resort to costly, error-prone environment reconstruction. We introduce the Agent Workspace Collaboration Protocol (AWCP), which bridges this gap through temporary workspace delegation inspired by the Unix philosophy that everything is a file. AWCP decouples a lightweight control plane from pluggable transport mechanisms, allowing a Delegator to project its workspace to a remote Executor, who then operates on the shared files directly with unmodified local toolchains. We provide a fully open-source reference implementation with MCP tool integration and validate the protocol through live demonstrations of asymmetric collaboration, where agents with complementary capabilities cooperate through delegated workspaces. By establishing the missing workspace layer in the agentic protocol stack, AWCP paves the way for a universally interoperable agent ecosystem in which collaboration transcends message boundaries. The protocol and reference implementation are publicly available at https://github.com/SII-Holos/awcp.

Xiaohang Nie, Zihan Guo, Youliang Chen +2

Harbin Institute of Technology · Shanghai Innovation Institute · Sun Yat-sen University +2

Multi-Agent CollaborationMulti-Agent LLM Systems
AskPDF
May 25, 2025

Position: Collaborative Agentic AI Needs Interoperability Across Ecosystems

Collaborative agentic AI is projected to transform entire industries by enabling AI-powered agents to autonomously perceive, plan, and act within digital environments. Yet, current solutions in this field are all built in isolation, and we are rapidly heading toward a landscape of fragmented, incompatible ecosystems. In this position paper, we argue that interoperability, achieved by the adoption of minimal standards, is essential to ensure open, secure, web-scale, and widely-adopted agentic ecosystems. To this end, we devise a minimal architectural foundation for collaborative agentic AI, named Web of Agents, which is composed of four building blocks: agent-to-agent messaging, interaction interoperability, state management, and agent discovery. Web of Agents adopts existing standards and reuses existing infrastructure where possible. With Web of Agents, we take the first but critical step toward interoperable agentic systems and offer a pragmatic path forward before ecosystem fragmentation becomes the norm.

Rishi Sharma, Martijn de Vos, Pradyumna Chari +2

1EPFL, Lausanne, Switzerland · 2Massachusetts Institute of Technology, Cambridge, MA, USA.

Agent Communication ProtocolsMulti-Agent CollaborationMulti-Agent SystemsAgentic AI
AskPDF
Jun 17, 2026

A Technical Taxonomy of LLM Agent Communication Protocols

As large language models (LLMs) advance and multi-agent systems aim to overcome the limits of standalone agents, robust communication protocols are becoming essential infrastructure for distributed agent networks. Nonetheless, the fragmented protocol landscape presents a significant interoperability challenge. This study develops a technical taxonomy to classify and analyze LLM agent communication protocols. Following an established iterative method, we defined the taxonomy's purpose, meta-characteristic, and ending conditions, then performed five iterations, three empirical-to-conceptual and two conceptual-to-empirical, on nine actively maintained open-source protocols with demonstrable adoption. The taxonomy comprises five dimensions: counterparty, payload, interaction state, discovery mechanism, and schema flexibility. Classification reveals recurring architectural patterns: all sampled agent-to-agent protocols combine hybrid payloads with session-state persistence; most protocols support multiple predefined schemas, and two negotiate schemas at runtime, indicating a trend toward schema flexibility; decentralized discovery remains rare. Analysis suggests short-term convergence pressure toward protocols unifying agent-to-agent and agent-to-context (tool and data) communication. Long-term, however, no single protocol is likely to maximize versatility, efficiency, and portability simultaneously. The field will more likely evolve toward a federated, layered protocol stack. The framework guides protocol selection and highlights open research gaps such as privacy and policy enforcement.}

Linus Sander, Habtom Kahsay Gidey, Alexander Lenz +1

Technische Universität München, Munich, Germany.

Agent Communication ProtocolsMulti-Agent Communication
AskPDF