Cross-domain mutual trust multi-agent cooperative system and method

Abstract

This application relates to a cross-domain, mutually trusted multi-agent collaborative system and method. The system includes: a cross-domain, mutually trusted decentralized agent identity identification and verification system, used to construct agent identity identifiers based on the decentralized identities of agents and migrate agent identities to decentralized identities; an agent interaction information storage system, used to hash-anchor agent data files, establish cross-agent task execution state consensus, and generate agent interaction information storage data; and a multi-agent collaborative task execution and dynamic access control system, used to execute interactive collaboration of multi-agent task execution process control, dynamic access permission control, and agent state management and maintenance according to task execution contracts, access control contracts, and agent management contracts. This solves the problems of existing technical solutions being unable to establish cross-enterprise and cross-domain agent identity mutual trust, and having serious limitations in system security and trustworthiness.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of multi-agent collaboration, and particularly relates to a multi-agent collaboration system and method capable of cross-domain mutual trust. BACKGROUND

[0002] Under the driving of large language models (LLMs), AI-based agents (i.e., software entities or systems integrated with AI technology, having autonomous decision-making, environment perception, goal-driven, and interactive capabilities) are accelerating the penetration of end-side user applications and enterprise-side production and service pipelines. Among them, the emergence of Agent-to-Agent (A2A) interaction protocols makes it possible to achieve complex task orchestration and execution based on multi-agent collaboration, driving the comprehensive automation of information collection and processing processes of agents and improving the efficiency of enterprise production and supply chains. However, the existing multi-agent collaboration protocol (such as Google A2A) is limited to the design of communication message specifications between agents (such as agent discovery mechanisms, agent capability description specifications, and message fragmentation specifications), and the security measures rely on traditional centralized identity authentication, HTTPS-based channel security, and other means. There is no (i) cross-enterprise unified agent identity identification and identity verification mechanism, (ii) tamper-proof evidence capability for multi-agent interaction processes and data, and (iii) context-based dynamic access control capability. There are serious trust and security bottlenecks, and the multi-agent framework is greatly limited in the application process of landing deployment.

[0003] In academia and industry, the most advanced multi-agent collaboration protocols and frameworks include: Google A2A: In 2025, Google proposed the Agent2Agent (A2A) protocol, which aims to enable communication and interoperability between AI-based agents developed by different frameworks, languages, or vendors. It supports agent capability discovery, interaction mode negotiation, collaboration task management, and secure information exchange. A2A protocol proposes standardized multi-agent collaboration concepts such as A2A client (requesting agent), A2A server (service-exposing agent), agent capability description card (Agent Card), agent message (Message), agent task (Task), message fragment (Part), and data fragment (Artifact). It also designs a JSON-RPC-based agent communication message specification. In addition, A2A protocol designs a streaming processing (SSE) mechanism and task execution result push notification mechanism for large-scale agent tasks. In terms of security measures, A2A relies on the TLS protocol to implement mutual authentication between agents, and developers can implement resource access authorization based on traditional mechanisms such as OAuth. The shortcomings of Google A2A protocol are: (1) Although the agent capability description card mechanism is proposed to define agent identity and capability, there is no standardized agent identity identification mechanism. Different companies and institutions still use customized and non-uniform agent identity identification methods, making it difficult to achieve cross-institutional agent identity trust; (2) Channel protection relying on TLS can only achieve mutual trust between message senders and receivers, but cannot achieve long-term tamper-proof agent interaction data evidence. In the event of agent system failure, it is difficult to locate the source of the failure and determine the relevant agent's rights and responsibilities; (3) The protocol itself does not provide specifications for resource authorization and access control security mechanisms. In actual multi-agent system development, the implementation of resource security mechanisms is determined by the developer, which further challenges multi-agent collaboration between different vendors and institutions, as inconsistent resource security mechanisms between different institutions make it difficult to interface and align.

[0004] ANP: Agent Network Protocol, is a multi-agent collaboration protocol that attempts to apply decentralized identity (DID) to agent identity in the early stage. This protocol builds a set of agent communication specifications based on DID, similar to Google A2A, and also supports agent identity description, agent discovery and encrypted communication. However, ANP does not provide a secure migration mechanism for converting traditional vendor-specific agent identity to DID. DID is completely independent of other agent identity identification solutions, so it is difficult to promote its use. In addition, ANP does not store evidence based on DID for agent interaction related data. DID is only used as an agent identity identifier, and no association is established between the agent and the related task and data. Therefore, DID cannot play a role in automatic resource circulation and realize traceable interaction history query. At the same time, ANP also lacks resource authorization and access control mechanisms.

[0005] Large language model agent development framework / SDK, such as CAMEL, LangGraph: This kind of framework / SDK focuses on building a large language model centered agent task execution flow, and designs a series of multi-agent collaboration mechanisms, for example: Supervisor-based architecture is managed by a supervisor agent that manages other agents with different capabilities. The supervisor agent distributes different tasks to specific agents and collects the execution results of each agent. Network / Graph-based architecture builds a specific agent network based on complex task requirements. Agents communicate and write based on this network topology to complete complex task execution. Although there is no standardized description of inter-agent communication protocols, this kind of framework defines the format of agent communication messages through SDK code. The problem is that different frameworks use completely different message formats, and the agents developed by each framework cannot interconnect and collaborate, severely limiting their range of use. In addition, this kind of framework also lacks a mechanism for storing evidence of agent interaction history, making it impossible to audit the interaction history, and the real-world feasibility is poor. Its security capabilities depend on system prompt word engineering or human intervention, and its scalability and robustness are poor, making it vulnerable to emerging malicious prompt word injection attacks.

[0006] In summary, although a series of protocols and development frameworks for AI-driven multi-agent collaboration have emerged in the industry and academia in recent years, existing frameworks focus on designing communication message specifications, multi-agent collaboration modes and architectures, and security mechanisms rely on traditional TLS, prompt engineering and other methods. However, in the face of highly automated and large-scale interactions of the multi-agent ecosystem, and the demand for massive resource access, existing technical solutions cannot establish cross-enterprise and cross-domain multi-agent identity trust, lack tamper-proof evidence of multi-agent interaction history (cannot support judicial audit requirements), and lack dynamic authorization and access control mechanism design, which has serious limitations in system security and trust level. SUMMARY

[0007] The present application provides a cross-domain trusted multi-agent collaboration system and method to solve the problems that existing technical solutions cannot establish cross-enterprise and cross-domain multi-agent identity trust, lack tamper-proof evidence of multi-agent interaction history (cannot support judicial audit requirements), and lack dynamic authorization and access control mechanism design, which has serious limitations in system security and trust level.

[0008] The first aspect embodiment of the present application provides a cross-domain trusted multi-agent collaboration system, comprising: a cross-domain trusted decentralized multi-agent identity identification and verification system, configured to construct a multi-agent identity identification satisfying a preset condition based on a decentralized identity of a multi-agent, and build a target verification mechanism of the decentralized identity according to the multi-agent identity identification, so as to migrate the identity of the multi-agent to the decentralized identity by using the target verification mechanism, and generate an identity compatibility result of the multi-agent; a multi-agent interaction information evidence system, configured to hash anchor a data file of the multi-agent based on the identity compatibility result, the decentralized identity and a distributed ledger, so as to establish a task execution state consensus across multi-agents, and generate interaction information evidence data of the multi-agent according to the task execution state consensus; and a multi-agent collaboration task execution and dynamic access control system, configured to execute multi-agent task execution process control, dynamic access permission control and multi-agent state management and maintenance interaction collaboration based on the interaction information evidence data, a task execution contract, an access control contract and a multi-agent management contract.

[0009] Optionally, in an embodiment of the present application, the cross-domain trust decentralized agent identity and verification system comprises: an agent identity registration module, configured to register a decentralized identity of the agent and a decentralized identity document bound to the decentralized identity; an agent message verification module, configured to generate a message verification result of the decentralized identity based on a message signature of the decentralized identity, the decentralized identity document and the target verification mechanism; an agent identity cross-domain module, configured to generate identity verification information of the agent according to cross-chain interoperability of the agent based on the message verification result; and an agent identity migration module, configured to migrate the identity of the agent to the decentralized identity based on the identity verification information to generate the identity compatibility result.

[0010] Optionally, in an embodiment of the present application, the agent identity registration module comprises: an initialization unit, configured to initialize the agent to generate a private key and a public key of the agent; a generation unit, configured to determine the decentralized identity by using a decentralized identity controller based on the private key and the public key of the agent; a construction unit, configured to construct the decentralized identity document bound to the decentralized identity; a request unit, configured to submit a hash value of the decentralized identity and the decentralized identity document to a decentralized identity storage smart contract on a block chain; and a verification unit, configured to verify whether the decentralized identity satisfies a preset legal condition based on the decentralized identity storage smart contract to generate a verification result of the agent.

[0011] Optionally, in an embodiment of the present application, the agent interaction information notarization system comprises: a task initialization module, configured to respectively implement on-chain anchoring and off-chain notarization on metadata and participant information of a task to generate an initialized task; a task state transition verification module, configured to use common signature of the multiple agents to confirm state transition information of the initialized task; a data anchoring module, configured to establish a binding relationship between off-chain storage and on-chain hash value of the data file based on the state transition information; and an off-chain trusted data access module, configured to collect a data source satisfying a preset reliable condition in the multi-agent cooperation based on the binding relationship to perform on-chain notarization on data to generate interaction information notarization data of the agent.

[0012] Optionally, in an embodiment of the present application, the data anchoring module comprises: a packaging unit configured to off-chain package the data file to generate encrypted data of the data file; a calculation unit configured to calculate a hash value of the data file based on the encrypted data; a storage unit configured to off-chain store the data file in a trusted manner to generate a file identifier of the data file; and a verification unit configured to upload the hash value of the data file, the file identifier, a timestamp and a file version to an on-chain smart contract of the agent to verify the data file.

[0013] Optionally, in an embodiment of the present application, the multi-agent collaborative task execution and dynamic access control system comprises: an agent resource access control module configured to define an access policy in the multi-agent interaction by using the access control contract, and to generate an access result by restricting identity, capability and access context information of an access subject according to the access policy; an interaction logic execution module configured to define a finite state machine of a task meeting a preset complex condition by using an interaction logic contract based on the access result, and to restrict a state transition rule according to the task finite state machine; and an agent governance module configured to maintain capability and state of the agent on-chain based on the state transition rule, to query a current state of the agent, and to supervise the agent according to the current state.

[0014] Optionally, in an embodiment of the present application, the agent resource access control module comprises: an access request unit configured to send an access request to the access control contract; a policy extraction unit configured to extract an access policy corresponding to a resource based on the access request; a policy evaluation unit configured to perform policy evaluation based on the access policy, a decentralized identity of the agent, a decentralized identity document and context information, to generate an evaluated policy, and to determine whether the evaluated policy meets a preset passing condition; a token generation unit configured to generate an access token of the agent in a case where the evaluated policy meets the preset passing condition; an access verification unit configured to request a target resource by using the token to verify whether the token meets a preset authenticity condition, and to determine whether the agent is allowed to access in a case where the token meets the preset authenticity condition; and an access confirmation unit configured to generate the access result in a case where the agent is allowed to access.

[0015] A second aspect of this application provides a cross-domain, mutually trusted multi-agent collaboration method, comprising the following steps: constructing an agent identity identifier that meets preset conditions based on the agent's decentralized identity, and establishing a target verification mechanism for the decentralized identity based on the agent identity identifier, so as to migrate the agent's identity to the decentralized identity using the target verification mechanism to generate an identity compatibility result for the agent; based on the identity compatibility result, the decentralized identity, and the distributed ledger, hashing and anchoring the agent's data file to establish a cross-agent task execution state consensus, and generating interactive information storage data for the agent based on the task execution state consensus; and based on the interactive information storage data, executing interactive collaboration of multi-agent task execution process control, dynamic access permission control, and agent state management and maintenance according to a task execution contract, an access control contract, and an agent management contract.

[0016] Optionally, in one embodiment of this application, the step of constructing an agent identity identifier that meets preset conditions, and building a target verification mechanism for the decentralized identity based on the agent identity identifier, so as to use the target verification mechanism to migrate the agent's identity to the decentralized identity and generate the agent's identity compatibility result, includes: registering the agent's decentralized identity and a decentralized identity document bound to the decentralized identity; generating a message verification result for the decentralized identity based on the message signature of the decentralized identity, the decentralized identity document, and the target verification mechanism; generating the agent's identity verification information based on the message verification result and the agent's cross-chain interoperability; and migrating the agent's identity to the decentralized identity based on the identity verification information to generate the identity compatibility result.

[0017] Optionally, in one embodiment of this application, registering the decentralized identity of the smart agent and the decentralized identity document bound to the decentralized identity includes: initializing the smart agent to generate the private key and public key of the smart agent; determining the decentralized identity using a decentralized identity controller based on the private key and public key of the smart agent; constructing the decentralized identity document bound to the decentralized identity; submitting the hash value of the decentralized identity and the decentralized identity document to a decentralized identity storage smart contract on the blockchain; and verifying whether the decentralized identity meets preset legal conditions based on the decentralized identity storage smart contract, and generating a verification result for the smart agent.

[0018] Optionally, in one embodiment of this application, the step of hash-anchoring the agent's data file based on the identity compatibility result, the decentralized identity, and the distributed ledger to establish a cross-agent task execution state consensus, and generating the agent's interaction information storage data according to the task execution state consensus, includes: implementing on-chain anchoring and off-chain storage for the task's metadata and participant information respectively to generate an initialized task; using the joint signature of the multiple agents to confirm the state transition information of the initialized task; establishing a binding relationship between the off-chain storage and on-chain hash value of the data file based on the state transition information; and collecting data sources that meet preset reliability conditions participating in the multi-agent collaboration based on the binding relationship to perform on-chain storage of the data and generate the agent's interaction information storage data.

[0019] Optionally, in one embodiment of this application, establishing the binding relationship between the off-chain storage and the on-chain hash value of the data file based on the state transition information includes: off-chain encapsulating the data file to generate encrypted data of the data file; calculating the hash value of the data file based on the encrypted data; off-chain trusted storage of the data file to generate a file identifier for the data file; and uploading the hash value, the file identifier, the timestamp, and the file version of the data file to the on-chain smart contract of the smart agent to verify the data file.

[0020] Optionally, in one embodiment of this application, the interactive collaboration of multi-agent task execution process control, dynamic access permission control, and agent state management and maintenance based on the interaction information storage data, according to the task execution contract, access control contract, and agent management contract, includes: defining access policies in the multi-agent interaction using the access control contract, and constraining the identity, capabilities, and access context information of the visitor according to the access policies to generate access results; defining a finite state machine of a task that satisfies preset complex conditions using the interaction logic contract, and constraining state transition rules according to the task finite state machine; and maintaining the capabilities and states of the agent on-chain based on the state transition rules to query the current state of the agent and supervise the agent according to the current state.

[0021] Optionally, in one embodiment of this application, the step of defining the access policy in the multi-agent interaction using the access control contract, and constraining the accessor's identity, capabilities, and access context information according to the access policy to generate an access result includes: sending an access request to the access control contract; extracting the access policy corresponding to the resource based on the access request; evaluating the policy based on the access policy, the agent's decentralized identity, decentralized identity document, and context information, generating an evaluated policy, and determining whether the evaluated policy meets a preset pass condition; generating an access token for the agent if the evaluated policy meets the preset pass condition; requesting the target resource using the token to verify whether the token meets a preset authenticity condition, and determining whether the agent allows access if the token meets the preset authenticity condition; and generating the access result if the agent allows access.

[0022] A third aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the cross-domain trust-based multi-agent cooperation method as described in the above embodiments.

[0023] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described cross-domain trustworthy multi-agent cooperation method.

[0024] This application's embodiments can support cross-enterprise and cross-domain intelligent agent authentication and mutual trust. Through the application of DID and blockchain cross-chain technology, it effectively solves the problem of interoperability difficulties caused by different identity identification schemes in existing frameworks. Simultaneously, by proposing a migration protocol from traditional identities to DID, it achieves compatibility with traditional identities, significantly improving the framework's feasibility. It binds the data, computation processes, and DID involved in intelligent agent interactions and hashes them to an immutable ledger, ensuring auditable and traceable data throughout the process, systematically addressing the trust shortcomings of existing solutions. Through automated configuration and execution of fine-grained access policies via smart contracts, it achieves globally unambiguous task flow state monitoring and management, and intelligent agent lifecycle management. Therefore, it solves the problems of existing technical solutions that cannot establish cross-enterprise and cross-domain intelligent agent identity mutual trust, lack immutable evidence of intelligent agent interaction history (unable to support judicial auditing requirements), and lack dynamic authorization and access control mechanisms, resulting in serious limitations in system security and trustworthiness.

[0025] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0026] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0027] Figure 1 This is a schematic diagram of the structure of a cross-domain trustworthy multi-agent cooperative system provided according to an embodiment of this application;

[0028] Figure 2 This is a schematic diagram of a trusted multi-agent cooperation framework according to an embodiment of this application;

[0029] Figure 3 This is a flowchart illustrating a cross-domain, mutually trusting multi-agent collaboration method provided according to an embodiment of this application.

[0030] Figure 4 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of this application. Detailed Implementation

[0031] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0032] The following description, with reference to the accompanying drawings, illustrates a cross-domain, mutually trustworthy multi-agent collaborative system and method according to embodiments of this application. Addressing the shortcomings of existing technologies mentioned in the background section—namely, the inability to establish cross-enterprise and cross-domain agent identity trust, the lack of immutable evidence of agent interaction history (failing to support judicial auditing requirements), and the lack of dynamic authorization and access control mechanisms—which result in severe limitations in system security and trustworthiness, this application provides a cross-domain, mutually trustworthy multi-agent collaborative system. In this system, the application of DID and blockchain cross-chain technology effectively resolves the interoperability difficulties of agents caused by differing identity identification schemes in existing frameworks. Furthermore, by proposing a migration protocol from traditional identities to DID, compatibility with traditional identities is achieved, significantly improving the framework's feasibility. The system binds the data, computation processes, and DID involved in agent interactions and hashes them to an tamper-proof ledger, ensuring auditable and traceable data throughout the process, systematically resolving the trust shortcomings of existing solutions. Finally, it achieves global, unambiguous task flow state monitoring and management, and agent lifecycle management through automated configuration and execution of fine-grained access policies via smart contracts. This solves the problems of existing technical solutions, such as the inability to establish cross-enterprise and cross-domain intelligent agent identity mutual trust, the lack of tamper-proof evidence of intelligent agent interaction history (which cannot support judicial auditing requirements), and the lack of dynamic authorization and access control mechanism design, resulting in serious limitations in system security and trustworthiness.

[0033] Specifically, Figure 1 This is a schematic diagram of the structure of a cross-domain trustworthy multi-agent cooperative system provided in an embodiment of this application.

[0034] like Figure 1 As shown, the cross-domain trust-based multi-agent collaboration system includes: a cross-domain trust-based decentralized agent identity identification and verification system 100, an agent interaction information storage system 200, and a multi-agent collaboration task execution and dynamic access control system 300.

[0035] Specifically, the cross-domain mutual trust decentralized intelligent agent identity identification and verification system 100 is used to construct an intelligent agent identity identifier that meets preset conditions based on the decentralized identity of the intelligent agent, and to build a target verification mechanism for the decentralized identity based on the intelligent agent identity identifier, so as to use the target verification mechanism to migrate the identity of the intelligent agent to the decentralized identity, so as to generate an identity-compatible result of the intelligent agent.

[0036] It is understood that the preset conditions in the embodiments of this application can be general and persistent conditions; the target verification mechanism can be a DID (Decentralized Identifier) ​​generation, circulation and verification mechanism, wherein DID does not rely on the identity authentication system of a centralized institution, and realizes autonomous control and secure verification of identity through blockchain technology. Here, the oracle is a middleware that connects the blockchain and the external world, and is responsible for securely and reliably transmitting off-chain data to smart contracts.

[0037] In actual implementation, the decentralized intelligent agent identity identification and verification system 100 with cross-domain mutual trust in this application embodiment can build a universal and persistent intelligent agent identity based on DID, establish a complete DID generation, circulation and verification mechanism, covering stages such as intelligent agent identity registration, identity certificate issuance, and certificate-based A2A message authenticity verification. At the same time, it supports cross-domain intelligent agent identity verification and the ability to securely migrate traditional identities to DID, so as to use the target verification mechanism to migrate the identity of the intelligent agent to the decentralized identity, so as to generate an identity-compatible result of the intelligent agent.

[0038] This application's embodiments greatly enhance interoperability by using a unified, globally verifiable agent identity, supporting cross-enterprise and cross-domain agent identity verification and mutual trust. Through the application of DID and blockchain cross-chain technology, it can effectively solve the problem of interoperability difficulties for agents caused by different identity identification schemes in existing frameworks. At the same time, by proposing a migration protocol from traditional identity to DID, it achieves compatibility with traditional identity and greatly improves the feasibility of the framework.

[0039] It should be noted that the preset conditions can be set by those skilled in the art according to the actual situation, and no specific restrictions are imposed here.

[0040] Optionally, in one embodiment of this application, the cross-domain mutual trust decentralized agent identity identification and verification system 100 includes: an agent identity registration module, used to register the decentralized identity of the agent and the decentralized identity document bound to the decentralized identity; an agent message verification module, used to generate a message verification result of the decentralized identity based on the message signature of the decentralized identity, the decentralized identity document, and the target verification mechanism; an agent identity cross-domain module, used to generate the agent's identity verification information based on the message verification result and the agent's cross-chain interoperability; and an agent identity migration module, used to migrate the agent's identity to the decentralized identity based on the identity verification information to generate an identity compatibility result.

[0041] It is understood that in the embodiments of this application, the agent identity registration module, agent message verification module, agent identity cross-domain module, and agent identity migration module correspond to the agent identity registration protocol, agent message verification protocol, agent identity cross-domain protocol, and traditional agent identity migration protocol, respectively.

[0042] Specifically, the protocol descriptions and protocol flows of each protocol in this application embodiment are as follows:

[0043] I. Intelligent Agent Identity Registration Protocol: This protocol generates a unique identity identifier for an intelligent agent, namely a decentralized identity (DID), and a DID document bound to the identity identifier. The document describes relevant information, and the document hash value and DID are bound together through on-chain storage on the blockchain.

[0044] II. Intelligent Agent Message Verification Protocol: This protocol provides a DID-based message signing and verification mechanism to generate decentralized identity message verification results, thereby achieving DID-driven message authenticity assurance.

[0045] (1) Message signing: Agent A signs its message m with the private key sk corresponding to DID to obtain σ, and sends the message and the signature together to Agent B;

[0046] (2) Public key extraction: Agent B queries the DID storage smart contract based on Agent A's DID. The smart contract can be an automated protocol stored on the blockchain. When the preset conditions are met, the contract terms are automatically executed to obtain H(D). At the same time, the DID document is extracted from the off-chain storage, and the consistency between the two is verified to determine the authenticity of the DID document. At the same time, Agent A's public key pk stored in the document is extracted.

[0047] (3) Message verification: Agent B verifies the authenticity of message m based on signature σ and pk.

[0048] III. Cross-Domain Identity Protocol for Smart Agents: When two smart agents register their DIDs through different blockchains, this protocol achieves identity verification between the two smart agents through cross-chain interoperability, generating identity verification information for the smart agents, thereby realizing cross-domain identity verification. In the protocol description below, let's assume the current goal is to transfer the identity of smart agent A in chain X to chain Y for verification by smart agent B in chain Y:

[0049] (1) Requesting cross-chain identity: Agent A submits a cross-chain identity request to the cross-chain bridge. This request includes the DID. A and H(D) A );

[0050] (2) Identity Verification: The cross-chain bridge verifies the DID based on the consensus mechanism of Chain X. A and H(D)A Verify the authenticity of the information; if the verification is successful, proceed to the next step.

[0051] (3) Trust anchor contract update: The cross-chain bridge updates the trust anchor smart contract in chain Y and stores it in DID. A and H(D) A The DID information includes the DID itself and related information (such as the identifier of the source chain X) to demonstrate that the DID information has been verified.

[0052] (4) DID document transfer: Agent A transfers DID documents off-chain. A Transmitted to agent B;

[0053] (5) Authenticity verification: Agent B verifies DID A and H(D) A Whether it is in a trust anchoring contract, to determine the authenticity of agent A's identity.

[0054] IV. Traditional Agent Identity Migration Protocol: The goal of this protocol is to securely migrate legitimate agent traditional identities (e.g., an agent from Google carries an identity provided by Google) to DID, thereby achieving framework compatibility with agents using traditional identities and generating identity-compatible results.

[0055] (1) Requesting traditional identity migration: Agent A or its identity manager submits an identity migration request to the Migration Committee (a node in the blockchain). The request includes (DID, DID document, traditional identity identifier, traditional identity source platform information) mediated by the Oracle service (i.e., the blockchain's trusted data source service).

[0056] (2) Traditional identity verification: Oracle verifies the authenticity of agent A's traditional identity based on the traditional identity source platform certificate. If it is authentic, Oracle signs the traditional identity and forwards it to the Migration Committee.

[0057] (3) Identity binding and certificate issuance: The Migration Committee verifies the Oracle signature. If the verification is successful, the Migration Committee records the binding relationship of (DID, DID document, traditional identity identifier) ​​on the chain and issues a Master certificate to agent A. The Master certificate contains (DID, DID document, traditional identity identifier, Migration Committee information, signature, timestamp).

[0058] (4) Context-specific certificate generation: The Migration Committee can issue a context-specific certificate to agent A that is bound to the Master certificate in order to partially disclose information related to the Master certificate and protect the privacy of agent A.

[0059] Optionally, in one embodiment of this application, the smart agent identity registration module includes: an initialization unit for initializing a smart agent to generate a private key and a public key for the smart agent; a generation unit for determining a decentralized identity based on the private key and public key of the smart agent using a decentralized identity controller; a construction unit for constructing a decentralized identity document bound to the decentralized identity; a request unit for submitting the hash value of the decentralized identity and the decentralized identity document to a decentralized identity storage smart contract on the blockchain; and a verification unit for verifying whether the decentralized identity meets preset legal conditions based on the decentralized identity storage smart contract and generating a verification result for the smart agent.

[0060] Specifically, the intelligent agent identity registration protocol in this application embodiment includes:

[0061] (1) Initialization: The agent generates a pair of private keys (sk) and public keys (pk);

[0062] (2) DID generation: The agent generates a valid DID through the DID controller as its identifier;

[0063] (3) DID document generation: The agent constructs its own DID document D through the DID controller, which contains DID, public key pk, service point (e.g., domain name + port), agent capabilities (e.g., code generation), verification method, and other metadata; the DID document is stored in IPFS off-chain or in a centralized database;

[0064] (4) Request on-chain anchoring: The agent initiates a session with the DID Registry SmartContract on the blockchain and submits its DID and H(D) (i.e., the DID document hash value);

[0065] (5) Smart contract verification: The DID storage smart contract verifies the legality of the DID. If it is legal, the DID and H(D) successfully establish an on-chain binding relationship.

[0066] The intelligent agent interaction information storage system 200 is used to hash-anchor intelligent agent data files based on identity compatibility results, decentralized identity and distributed ledger, in order to establish a consensus on task execution status across intelligent agents, and generate intelligent agent interaction information storage data based on the consensus on task execution status.

[0067] It is understood that the intelligent agent interaction information storage system 200 in the embodiments of this application can be a tamper-proof and traceable intelligent agent interaction information storage system.

[0068] In actual implementation, the agent interaction information storage system 200 in this application embodiment can achieve hash anchoring of important data files and important multi-party interaction milestone events through DID and distributed ledger, preventing relevant agents from tampering with and denying the interaction history. Furthermore, it effectively establishes a consensus on the task execution status across agents by combining a multi-party signature mechanism, and generates agent interaction information storage data based on the consensus on the task execution status, effectively clarifying the rights and responsibilities of each party in agent collaboration.

[0069] The embodiments of this application can perform judicial-grade evidence storage of intelligent agent interaction history, greatly improving credibility. It binds the data, calculation process and DID involved in intelligent agent interaction and hashes them to an anti-tampering ledger, ensuring that the data throughout the process is auditable and traceable, and systematically solves the trust shortcomings of existing solutions.

[0070] Optionally, in one embodiment of this application, the intelligent agent interaction information storage system 200 includes: a task initialization module, used to perform on-chain anchoring and off-chain storage of the task's metadata and participant information respectively, to generate an initialized task; a task state transition verification module, used to confirm the state transition information of the initialized task using joint signatures of multiple intelligent agents; a data anchoring module, used to establish a binding relationship between off-chain storage of data files and on-chain hash values ​​based on the state transition information; and an off-chain trusted data access module, used to collect data sources that meet preset reliability conditions in the multi-agent collaboration based on the binding relationship, to perform on-chain storage of the data, and generate intelligent agent interaction information storage data.

[0071] It is understood that the task initialization module, task state transition verification module, data anchoring module, and off-chain trusted data access module in the embodiments of this application correspond to the task initialization protocol, task state transition verification protocol, data anchoring protocol, and off-chain trusted data access protocol, respectively; the data source that meets the preset reliability conditions can be a reliable and authoritative data source.

[0072] Specifically, the functional descriptions and process descriptions of each protocol in this application embodiment are as follows:

[0073] I. Task Initialization Protocol: The goal of this protocol is for the initiator (agent) of a multi-agent collaborative task to anchor the task's metadata and participant information on-chain and store it off-chain to generate the initialized task.

[0074] (1) The agent that initiates the task collects task-related metadata off-chain, including the task initiator agent DID, task participant agent DIDs, detailed task description, and deadline timestamp.

[0075] (2) Metadata hash calculation: The task initiating agent concatenates the above metadata and calculates the task metadata hash value H. task ;

[0076] (3) On-chain anchoring: The task initiator will anchor H task Submit to the on-chain smart contract and record the task status as "Initiated";

[0077] (4) Off-chain storage: Complete task metadata is stored through off-chain storage (IPFS or centralized database);

[0078] (5) Confirmation: The on-chain smart contract and off-chain storage return a message indicating that the task initialization was successful.

[0079] II. Task State Transition Verification Protocol: In this framework, to prevent malicious agents from pushing false task state transition information and interfering with the multi-agent collaboration process, the framework requires multiple agents to jointly sign and confirm the task state transition information, thereby clarifying the task responsibilities of the agents and effectively resisting malicious agents.

[0080] (1) Multi-agent signature: The relevant agents perform BLS signatures on the latest task milestone (marking the task state transition);

[0081] (2) Signature aggregation: An aggregator (which can be one of multiple agents) aggregates the BLS signatures of each agent to obtain Σ;

[0082] (3) Uploading state transition information to the chain: The aggregator uploads the task hash H task The aggregated signature Σ, milestone event name, and relevant participant DIDs are uploaded to the on-chain smart contract;

[0083] (4) State transition verification: The smart contract extracts the public key corresponding to the DID of the task-related participants through the Oracle and verifies the BLS signature;

[0084] (5) Task status update: If the previous step of verification is successful, the smart contract updates the task status recorded on the chain and marks it as "Verified".

[0085] III. Data Anchoring Protocol: The goal of this protocol is to establish a binding relationship between off-chain storage and on-chain hash values ​​for large data files in order to achieve long-term data tamper-proofing.

[0086] IV. Off-chain Trusted Data Access Protocol: To ensure that the data collected from the multi-agent collaboration is from reliable and authoritative data sources and to guarantee the long-term tamper-proof nature of reliable data, the trusted data access protocol is based on a distributed trusted oracle to achieve off-chain trusted data access and further performs on-chain notarization of the data.

[0087] (1) Authentication and key exchange: The intelligent agent requesting access to the data, the Oracle, and the authoritative data source jointly execute the three-party TLS handshake protocol, in which the first two act as TLS clients and the latter acts as a TLS server to establish a session key;

[0088] (2) Secure data request: The agent and the Oracle construct an encrypted data request based on secure computation by both parties (the agent is required to have data access permissions);

[0089] (3) Data response and decryption: The authoritative data source server responds to the request and returns encrypted data. The agent and Oracle jointly decrypt the data and verify the data integrity through MAC.

[0090] (4) Data hashing on the chain: Oracle uses the data anchoring protocol to store data hashes on the chain, so as to achieve long-term data anti-tampering.

[0091] Optionally, in one embodiment of this application, the data anchoring module includes: an encapsulation unit for off-chain encapsulating a data file to generate encrypted data of the data file; a calculation unit for calculating the hash value of the data file based on the encrypted data; a storage unit for off-chain trusted storage of the data file to generate a file identifier for the data file; and a verification unit for uploading the hash value, file identifier, timestamp, and file version of the data file to the on-chain smart contract of the smart agent to verify the data file.

[0092] The data anchoring module in this embodiment includes:

[0093] (1) Off-chain data preparation: The agent encapsulates the data file to be sealed off-chain and performs necessary data operations such as encryption to generate encrypted data of the data file;

[0094] (2) Hash generation: The agent calculates the hash value of the data file based on the SHA-256 algorithm;

[0095] (3) Off-chain complete data storage: The agent stores the complete data file to an off-chain trusted storage such as IPFS and obtains a unique file identifier (CID);

[0096] (4) Hash on the chain: The agent uploads information such as file hash value H, file identifier CID, timestamp and file version to the on-chain smart contract. After that, file visitors can verify the integrity of the file by combining the on-chain hash value.

[0097] The multi-agent collaborative task execution and dynamic access control system 300 is used to perform interactive collaboration of multi-agent task execution process control, dynamic access permission control and agent state management and maintenance based on interactive information storage data, and according to task execution contract, access control contract and agent management contract.

[0098] It is understood that the multi-agent collaborative task execution and dynamic access control system 300 in the embodiments of this application can be an automated multi-agent collaborative task execution and dynamic access control system.

[0099] In actual implementation, the multi-agent collaborative task execution and dynamic access control system 300 in this application embodiment is based on interactive information storage data. Through the design of smart contracts such as task execution contracts, access control contracts, and agent management contracts, and corresponding interaction protocols, it achieves strict and automated task flow control and resource access control, enhancing security. Through the automated configuration and execution of fine-grained access policies by smart contracts, it realizes task flow status monitoring and management, agent lifecycle management, and ensures the realization of globally unambiguous and automated multi-agent task execution process control, context-aware dynamic access control, and agent state management and maintenance.

[0100] This application proposes a multi-agent collaboration framework, named BlockA2A, which features trusted identity, traceable interaction, and secure resources. It addresses the shortcomings of existing multi-agent collaboration protocols and frameworks in areas such as cross-domain identity trust, interaction history storage, and dynamic resource access control, providing a foundational trust for large-scale, cross-enterprise autonomous and secure agent collaboration. By organically combining and utilizing decentralized identity, distributed ledger, and smart contract technologies, BlockA2A comprehensively covers the security of the identity, interaction, and execution layers of multi-agent collaboration. The specific implementation scheme of the BlockA2A multi-agent collaboration framework is as follows: Figure 2 As shown, the protocol flow design of the multi-agent collaboration framework in the agent identity layer, interaction layer, and execution layer is introduced.

[0101] Optionally, in one embodiment of this application, the multi-agent collaborative task execution and dynamic access control system 300 includes: an agent resource access control module, used to define access policies in multi-agent interactions using access control contracts, and to constrain the identity, capabilities, and access context information of the accessor according to the access policies, and generate access results; an interaction logic execution module, used to define a finite state machine of a task that satisfies preset complex conditions using interaction logic contracts based on the access results, and to constrain state transition rules according to the task finite state machine; and an agent governance module, used to maintain the capabilities and states of agents on-chain based on state transition rules, to query the current state of agents, and to supervise agents according to the current state.

[0102] It is understood that the agent resource access control module, the interaction logic execution module, and the agent governance module in the embodiments of this application correspond to the agent resource access control protocol, the interaction logic execution protocol, and the agent governance protocol, respectively.

[0103] In actual implementation, the protocol descriptions and protocol flow descriptions of each protocol in this application embodiment are as follows:

[0104] I. Intelligent Agent Resource Access Control Protocol: The core of this protocol is the ACC (Access Control Contract), which defines the access policy for a specific resource involved in multi-agent interactions. The access policy constrains the accessor's identity, capabilities, and access context information. The access control contract is responsible for policy evaluation (i.e., determining whether access is currently permitted), access token issuance and verification, and generating access results.

[0105] II. Interaction Logic Execution Protocol: The core of this protocol is the Interaction Logic Contract (ILC), which defines a finite state machine for complex tasks and constrains state transition rules through the definition of state transition guards. The Interaction Logic Execution Protocol relies on ILC for automated task execution and task state management.

[0106] (1) ILC initialization and deployment: The task-related intelligent agent (or task administrator) formulates the ILC. The ILC defines the finite set of task states and the initial state in the form of member variables, and defines state transition guards and constrains state transition conditions through member functions. The ILC is deployed to the blockchain.

[0107] (2) Task state transition verification: The preconditions for task state transition are consistent with the aforementioned task state transition verification protocol. They are achieved through off-chain multi-party signature generation and on-chain signature verification. In addition, the evaluation of state transition guard conditions is added to constrain the time and context of state transition.

[0108] (3) Task status update and notification: If the task status transition verification is successful, ILC updates the task status and generates an on-chain event to notify the relevant smart agents;

[0109] (4) Task execution history storage: The task execution history is stored on the chain as an on-chain session (transaction) related to ILC deployment and task state transition, which can prevent tampering for a long time.

[0110] III. Intelligent Agent Governance Protocol: This protocol, centered on the Intelligent Agent's DID, maintains the agent's capabilities and state on-chain, supporting relevant participants in multi-agent collaboration to query the current state of a specific intelligent agent. Its goal is to effectively supervise intelligent agents.

[0111] (1) Registration: The agent registers its identity through the agent identity registration protocol. The identity information (DID) is synchronized to the corresponding agent governance protocol. The initial state is "Active".

[0112] (2) Capability update: The agent manager (constrained by the smart contract require syntax) can update the capabilities of the agent. The update is presented in the DID document and is synchronously updated on the on-chain storage.

[0113] (3) DID Revocation: When an agent is detected as malicious or high-risk, the agent's state is set to "Revoked" and an event is generated to notify all relevant participants;

[0114] (4) DID search: Supports searching based on DID pairs.

[0115] Optionally, in one embodiment of this application, the intelligent agent resource access control module includes: an access request unit, configured to send an access request to an access control contract; a policy extraction unit, configured to extract the access policy corresponding to the resource based on the access request; a policy evaluation unit, configured to evaluate the policy based on the access policy, the intelligent agent's decentralized identity, the decentralized identity document, and context information, generate an evaluated policy, and determine whether the evaluated policy meets preset access conditions; a token generation unit, configured to generate an access token for the intelligent agent if the evaluated policy meets the preset access conditions; an access verification unit, configured to use the token to request the target resource to verify whether the token meets preset authenticity conditions, and determine whether the intelligent agent is allowed to access if the token meets the preset authenticity conditions; and an access confirmation unit, configured to generate an access result if the intelligent agent allows access.

[0116] The intelligent agent resource access control module in this embodiment may include:

[0117] (1) Access request: Agent A, as the visitor, sends an access request to ACC, which includes its own DID, the operation (action) to be performed on the resource, and the identifier of the target resource;

[0118] (2) Policy extraction: ACC extracts the access policy corresponding to the resource based on the request;

[0119] (3) Policy evaluation: ACC evaluates the policy based on the agent’s DID, DID document (which can be obtained through Oracle), and context information to determine whether an access token can be generated.

[0120] (4) Token generation: If the strategy evaluation is passed, an access token is generated. The token specifies the visitor's DID, resource identifier, resource operation and token expiration time. The token is uniquely identified by random characters and stored on the chain. At the same time, it is fed back to agent A.

[0121] (5) Token-based resource access and verification: Agent A requests the target resource based on the token. The manager of the target resource can verify the authenticity of the token through the verification function of ACC to decide whether to allow access.

[0122] (6) Access confirmation: The target resource manager reports the access result to agent A.

[0123] The cross-domain, mutually trustworthy multi-agent collaborative system proposed in this application effectively solves the interoperability difficulties of existing frameworks caused by different identity identification schemes through the application of DID and blockchain cross-chain technology. Furthermore, by proposing a migration protocol from traditional identities to DID, it achieves compatibility with traditional identities, significantly improving the framework's feasibility. It binds the data, computation processes, and DIDs involved in agent interactions and hashes them to an immutable ledger, ensuring auditable and traceable data throughout the process, systematically addressing the trust shortcomings of existing solutions. Through automated configuration and execution of fine-grained access policies via smart contracts, it achieves globally unambiguous task flow state monitoring and management, and agent lifecycle management. Therefore, it solves the problems of existing technical solutions that cannot establish cross-enterprise and cross-domain agent identity trust, lack immutable evidence of agent interaction history (unable to support judicial auditing requirements), and lack dynamic authorization and access control mechanisms, resulting in serious limitations in system security and trustworthiness.

[0124] Next, referring to the accompanying drawings, a cross-domain trustworthy multi-agent cooperation method proposed according to an embodiment of this application is described.

[0125] Figure 3This is a flowchart illustrating a cross-domain, mutually trusting multi-agent collaboration method according to an embodiment of this application.

[0126] like Figure 3 As shown, this cross-domain trust-based multi-agent cooperation method includes the following steps:

[0127] In step S301, based on the decentralized identity of the agent, an agent identity identifier that meets preset conditions is constructed, and a target verification mechanism for the decentralized identity is built according to the agent identity identifier, so as to use the target verification mechanism to migrate the agent's identity to the decentralized identity, thereby generating an identity compatibility result for the agent.

[0128] In step S302, based on the identity compatibility result, decentralized identity, and distributed ledger, the data file of the intelligent agent is hashed to establish a consensus on the task execution state across intelligent agents, and interactive information storage data of the intelligent agents is generated according to the consensus on the task execution state.

[0129] In step S303, based on the interactive information storage data, and according to the task execution contract, access control contract, and agent management contract, the interactive collaboration of multi-agent task execution process control, dynamic access permission control, and agent state management and maintenance is performed.

[0130] Optionally, in one embodiment of this application, an agent identity identifier that meets preset conditions is constructed, and a decentralized identity target verification mechanism is built based on the agent identity identifier to migrate the agent's identity to the decentralized identity, thereby generating an identity-compatible result for the agent. This includes: registering the agent's decentralized identity and a decentralized identity document bound to the decentralized identity; generating a message verification result for the decentralized identity based on the message signature of the decentralized identity, the decentralized identity document, and the target verification mechanism; generating the agent's identity verification information based on the message verification result and the agent's cross-chain interoperability; and migrating the agent's identity to the decentralized identity based on the identity verification information to generate an identity-compatible result.

[0131] Optionally, in one embodiment of this application, registering a decentralized identity for a smart agent and a decentralized identity document bound to the decentralized identity includes: initializing the smart agent to generate a private key and a public key for the smart agent; determining the decentralized identity using a decentralized identity controller based on the private key and public key of the smart agent; constructing a decentralized identity document bound to the decentralized identity; submitting the hash values ​​of the decentralized identity and the decentralized identity document to a decentralized identity storage smart contract on the blockchain; and verifying whether the decentralized identity meets preset legal conditions based on the decentralized identity storage smart contract, and generating a verification result for the smart agent.

[0132] Optionally, in one embodiment of this application, based on identity compatibility results, decentralized identity, and distributed ledger, the data files of the intelligent agents are hash-anchored to establish a consensus on the task execution state across intelligent agents, and interactive information storage data of the intelligent agents is generated based on the consensus on the task execution state. This includes: implementing on-chain anchoring and off-chain storage for the metadata of the task and the information of the participants, respectively, to generate an initialized task; using the joint signature of multiple intelligent agents to confirm the state transition information of the initialized task; establishing a binding relationship between the off-chain storage of the data files and the on-chain hash value based on the state transition information; and collecting data sources that meet preset reliability conditions in the multi-agent collaboration based on the binding relationship to perform on-chain storage of the data and generate interactive information storage data of the intelligent agents.

[0133] Optionally, in one embodiment of this application, a binding relationship is established between off-chain storage of a data file and its on-chain hash value based on state transition information, including: off-chain encapsulating the data file to generate encrypted data of the data file; calculating the hash value of the data file based on the encrypted data; off-chain trusted storage of the data file to generate a file identifier for the data file; and uploading the hash value, file identifier, timestamp, and file version of the data file to the agent's on-chain smart contract to verify the data file.

[0134] Optionally, in one embodiment of this application, based on the interactive information storage data, and according to the task execution contract, access control contract, and agent management contract, the interactive collaboration of multi-agent task execution process control, dynamic access permission control, and agent state management and maintenance is performed, including: defining access policies in multi-agent interactions using the access control contract, and constraining the identity, capabilities, and access context information of the visitors according to the access policies to generate access results; based on the access results, defining a finite state machine of a task that meets preset complex conditions using the interaction logic contract, and constraining state transition rules according to the task finite state machine; and based on the state transition rules, maintaining the capabilities and states of agents on the chain to query the current state of agents and supervise agents according to the current state.

[0135] Optionally, in one embodiment of this application, an access control contract is used to define an access policy in multi-agent interaction, and the access policy constrains the identity, capabilities, and access context information of the visitor to generate an access result. This includes: sending an access request to the access control contract; extracting the access policy corresponding to the resource based on the access request; evaluating the policy based on the access policy, the agent's decentralized identity, the decentralized identity document, and context information, generating an evaluated policy, and determining whether the evaluated policy meets preset access conditions; generating an access token for the agent if the evaluated policy meets the preset access conditions; requesting the target resource using the token to verify whether the token meets preset authenticity conditions, and determining whether the agent allows access if the token meets the preset authenticity conditions; and generating an access result if the agent allows access.

[0136] It should be noted that the foregoing explanation of the embodiment of the cross-domain trustworthy multi-agent cooperative system also applies to the cross-domain trustworthy multi-agent cooperative method of this embodiment, and will not be repeated here.

[0137] The cross-domain, mutually trustworthy multi-agent collaboration method proposed in this application effectively solves the interoperability difficulties of intelligent agents caused by different identity identification schemes in existing frameworks through the application of DID and blockchain cross-chain technology. Simultaneously, by proposing a migration protocol from traditional identities to DID, compatibility with traditional identities is achieved, significantly improving the framework's feasibility. The method binds the data, computation processes, and DID involved in intelligent agent interactions and hashes them to an immutable ledger, ensuring auditable and traceable data throughout the process, systematically addressing the trust shortcomings of existing solutions. Through automated configuration and execution of fine-grained access policies via smart contracts, it achieves globally unambiguous task flow state monitoring and management, and intelligent agent lifecycle management. Therefore, it solves the problems of existing technical solutions failing to establish cross-enterprise and cross-domain intelligent agent identity mutual trust, lacking immutable evidence of intelligent agent interaction history (failing to support judicial auditing requirements), and lacking dynamic authorization and access control mechanisms, resulting in serious limitations in system security and trustworthiness.

[0138] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include:

[0139] The memory 401, the processor 402, and the computer program stored on the memory 401 and capable of running on the processor 402.

[0140] When the processor 402 executes the program, it implements the cross-domain mutual trust multi-agent cooperation method provided in the above embodiments.

[0141] Furthermore, electronic devices also include:

[0142] Communication interface 403 is used for communication between memory 401 and processor 402.

[0143] The memory 401 is used to store computer programs that can run on the processor 402.

[0144] The memory 401 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0145] If the memory 401, processor 402, and communication interface 403 are implemented independently, then the communication interface 403, memory 401, and processor 402 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be divided into address buses, data buses, control buses, etc. For ease of representation, Figure 4 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0146] Optionally, in a specific implementation, if the memory 401, processor 402, and communication interface 403 are integrated on a single chip, then the memory 401, processor 402, and communication interface 403 can communicate with each other through an internal interface.

[0147] Processor 402 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0148] This embodiment also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described cross-domain trustworthy multi-agent cooperation method.

[0149] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0150] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0151] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0152] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0153] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0154] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0155] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0156] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A multi-agent collaboration system that can be cross-domain trusted, characterized by, The application relates to a cross-domain trust decentralized intelligent agent identity and verification system, which is used for constructing an intelligent agent identity meeting preset conditions based on a decentralized identity of the intelligent agent, building a target verification mechanism of the decentralized identity according to the intelligent agent identity, migrating the identity of the intelligent agent to the decentralized identity by using the target verification mechanism, and generating an identity compatibility result of the intelligent agent. An intelligent agent interaction information storage system is used for hashing anchoring data files of the intelligent agent based on the identity compatibility result, the decentralized identity and a distributed ledger, establishing a cross-intelligent agent task execution state consensus, and generating interaction information storage data of the intelligent agent according to the task execution state consensus. A multi-intelligent agent cooperation task execution and dynamic access control system is used for executing multi-intelligent agent task execution process control, dynamic access permission control and intelligent agent state management and maintenance interaction cooperation according to a task execution contract, an access control contract and an intelligent agent management contract based on the interaction information storage data. The decentralized identity of the intelligent agent is constructed to meet preset conditions, and the target verification mechanism of the decentralized identity is built according to the intelligent agent identity, so as to migrate the identity of the intelligent agent to the decentralized identity by using the target verification mechanism, which comprises the following steps: generating a unique intelligent agent identity for the intelligent agent by using an intelligent agent identity registration protocol, and binding a decentralized identity document to the intelligent agent identity; providing message signature based on the decentralized identity, the decentralized identity document and the target verification mechanism by using an intelligent agent message verification protocol, and generating a message verification result of the decentralized identity; based on the message verification result, when two intelligent agents register the decentralized identity through different blockchains, realizing identity verification between the two intelligent agents by cross-chain interoperation through an intelligent agent identity cross-domain protocol, and generating identity verification information of the intelligent agents; based on the identity verification information, migrating a legal intelligent agent traditional identity to the decentralized identity by using a traditional intelligent agent identity migration protocol. The task execution state consensus across intelligent agents is established, and the interaction information storage data of the intelligent agents is generated according to the task execution state consensus, which comprises the following steps: on-chain anchoring and off-chain storage of metadata and participant information of the intelligent agent on a task are performed to generate an initialized task; the state transition information of the initialized task is confirmed by jointly signing the task state transition of the multi-intelligent agent; based on the state transition information, a binding relationship between off-chain storage of a large data file and on-chain hash value is established; based on the binding relationship, data sources meeting preset reliable conditions in the multi-intelligent agent cooperation are collected, off-chain trusted data access is realized based on a distributed trusted Oracle, and the data is further stored on chain to generate the interaction information storage data of the intelligent agents. ​ The interaction collaboration of the task execution contract, the access control contract and the agent management contract includes: defining an access policy of a resource involved in the multi-agent interaction through the access control contract, and restricting the identity, ability and access context information of an access according to the access policy, using the access control contract to be responsible for policy evaluation, access token issuance and token verification, and generating an access result; based on the access result, defining a finite state machine of a complex task using an interaction logic contract, and restricting state transition rules through the definition of state transition guards; based on the state transition rules, maintaining the ability and state of the agent on the chain with the agent decentralized identity as the core, supporting related parties in multi-agent collaboration to query the current state of a specific agent to effectively supervise the agent. 2.The cross-domain trustable multi-agent collaboration system of claim 1, wherein, The cross-domain mutual trust decentralized agent identity and verification system comprises: An agent identity registration module is configured to register a decentralized identity of the agent and a decentralized identity document bound to the decentralized identity. An agent message verification module is configured to generate a message verification result of the decentralized identity based on a message signature of the decentralized identity, the decentralized identity document and a target verification mechanism. An agent identity cross-domain module is configured to generate identity verification information of the agent based on the message verification result according to cross-chain interoperability of the agent. An agent identity migration module is configured to migrate the identity of the agent to the decentralized identity based on the identity verification information to generate an identity compatibility result. 3.The cross-domain trustable multi-agent collaboration system of claim 2, wherein, The agent identity registration module comprises: An initialization unit is configured to initialize the agent to generate a private key and a public key of the agent. A generation unit is configured to determine the decentralized identity using a decentralized identity controller based on the private key and the public key of the agent. A construction unit is configured to construct the decentralized identity document bound to the decentralized identity. A request unit is configured to submit a hash value of the decentralized identity and the decentralized identity document to a decentralized identity storage smart contract on a blockchain. A verification unit is configured to verify whether the decentralized identity satisfies a preset legal condition based on the decentralized identity storage smart contract to generate a verification result of the agent. 4.The cross-domain trustable multi-agent collaboration system of claim 1, wherein, The agent interaction information notarization system comprises: A task initialization module is configured to implement on-chain anchoring and off-chain notarization of metadata and participant information of a task respectively to generate an initialized task. A task state transition verification module is configured to use common signatures of the multi-agent to confirm state transition information of the initialized task. A data anchoring module is configured to establish a binding relationship between off-chain storage and on-chain hash values of the data file based on the state transition information. A chain-offline trusted data access module is configured to collect data sources that meet preset reliable conditions in the multi-agent cooperation based on the binding relationship, to store data on a chain to generate interaction information storage data of the agent. 5.The cross-domain trustable multi-agent collaboration system of claim 4, wherein, The data anchoring module includes: An encapsulation unit is configured to encapsulate the data file off-chain to generate encrypted data of the data file; A calculation unit is configured to calculate a hash value of the data file based on the encrypted data; A storage unit is configured to store the data file off-chain to generate a file identifier of the data file; An inspection unit is configured to upload the hash value of the data file, the file identifier, a timestamp, and a file version to a smart contract on a chain of the agent to inspect the data file. 6.The cross-domain trustable multi-intelligent entity collaboration system of claim 1, wherein, The multi-agent cooperation task execution and dynamic access control system includes: An agent resource access control module is configured to define an access strategy in the multi-agent interaction by using the access control contract, to constrain identity, capability, and access context information of an access person according to the access strategy, and to generate an access result; An interaction logic execution module is configured to define a finite state machine of a task that meets a preset complex condition by using an interaction logic contract based on the access result, and to constrain a state transition rule according to the task finite state machine; An agent governance module is configured to maintain capability and state of the agent on a chain based on the state transition rule, to query a current state of the agent, and to supervise the agent according to the current state.

7. The cross-domain trustable multi-intelligent entity collaboration system according to claim 6, characterized in that, The agent resource access control module includes: An access request unit is configured to send an access request to the access control contract; A strategy extraction unit is configured to extract an access strategy corresponding to a resource based on the access request; A strategy evaluation unit is configured to perform strategy evaluation based on the access strategy, a decentralized identity of the agent, a decentralized identity document, and context information, to generate an evaluated strategy, and to determine whether the evaluated strategy meets a preset passing condition; A token generation unit is configured to generate an access token of the agent in a case where the evaluated strategy meets the preset passing condition; An access verification unit is configured to request a target resource by using the token to determine whether the token meets a preset authenticity condition, and to determine whether the agent is allowed to access in a case where the token meets the preset authenticity condition; An access confirmation unit is configured to generate the access result in a case where the agent is allowed to access.

8. A multi-agent cooperation method capable of cross-domain mutual trust, characterized in that, The method includes the following steps: An agent identity that meets a preset condition is constructed based on a decentralized identity of the agent, a target verification mechanism of the decentralized identity is built according to the agent identity, the identity of the agent is migrated to the decentralized identity by using the target verification mechanism to generate an identity compatibility result of the agent; A data file of the agent is hash anchored based on the identity compatibility result, the decentralized identity, and a distributed ledger to establish a task execution state consensus across agents, and interaction information storage data of the agent is generated according to the task execution state consensus. Based on the interaction information evidence data, according to the task execution contract, the access control contract and the agent management contract, the task execution process control, the dynamic access permission control and the interaction cooperation of the agent state management maintenance of the multi-agent are executed; The decentralized identity based on the agent constructs an agent identity identifier meeting a preset condition, and builds a target verification mechanism of the decentralized identity according to the agent identity identifier, so as to migrate the identity of the agent to the decentralized identity by using the target verification mechanism, including: generating a unique agent identity identifier for the agent by using an agent identity registration protocol, and a decentralized identity document bound with the agent identity identifier; providing a message signature based on the decentralized identity and the target verification mechanism by using an agent message verification protocol to generate a message verification result of the decentralized identity; based on the message verification result, when two agents register the decentralized identity through different blockchains, realizing identity verification between the two agents by cross-chain interoperation through an agent identity cross-domain protocol to generate identity verification information of the agent; based on the identity verification information, migrating a legal agent traditional identity to the decentralized identity by using a traditional agent identity migration protocol; The task execution state consensus across the agents is established, and the interaction information evidence data of the agent is generated according to the task execution state consensus, including: anchoring and off-chain evidence of the metadata and participant information of the task by using the agent to generate an initialized task; confirming the state transition information of the initialized task by using the multi-agent to jointly sign the task state transition; based on the state transition information, establishing a binding relationship between off-chain storage of a large data file and on-chain hash value; based on the binding relationship, collecting data sources meeting a preset reliable condition in the multi-agent cooperation, realizing off-chain trusted data access based on a distributed trusted Oracle, and further storing the data on the chain to generate the interaction information evidence data of the agent; The interaction collaboration of the task execution contract, the access control contract and the agent management contract for performing the multi-agent task execution flow control, the dynamic access permission control and the agent state management maintenance includes: defining an access strategy of a certain resource involved in the multi-agent interaction through the access control contract, and performing constraint on an identity, an ability and access context information of an access person according to the access strategy, performing strategy evaluation, access token issuing and token verification by using the access control contract, and generating an access result; based on the access result, defining a finite state machine of a complex task by using an interaction logic contract, and performing constraint on a state transition rule by defining a state transition guard; based on the state transition rule, performing on-chain maintenance on the ability and the state of the agent with the agent decentralized identity as a core, and supporting a related participant in the multi-agent collaboration to query a current state of a specific agent, so as to effectively supervise the agent.

9. An electronic device, comprising: Comprising: A memory, a processor and a computer program stored on the memory and executable on the processor, the processor executing the program to implement the multi-agent collaboration method across domains and trusted.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the multi-agent collaboration method across domains and trusted.

Images