A full-link hybrid identity authentication system for intelligent agent scenarios

By using a full-chain hybrid identity authentication system, which leverages lightweight session IDs and full-chain identity orchestration, the system addresses the security risks and data redundancy issues in identity management within intelligent agent scenarios, enabling efficient identity tracking and real-time permission decision-making.

CN123122836APending Publication Date: 2026-11-17ASIAINFO TECH (CHENGDU) INC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610982181.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-02
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Traditional human identity management models cannot adapt to the dynamic behavioral characteristics of intelligent agents, leading to security and audit risks. Existing identity technology solutions cannot achieve efficient identity tracking, dynamic verification, and permission adjustment in intelligent agent scenarios, and also suffer from data redundancy and high latency issues.

Method used

Design a full-link hybrid identity authentication system, including a runtime plane module, a policy enforcement point module, and a control plane module. The system manages traffic through a gateway, uses lightweight session IDs for identity transmission, and establishes a full-link identity orchestration and link context index to achieve unified management and efficient tracking of human-machine hybrid identities.

Benefits of technology

It enables the transmission of identity tokens without identity throughout the process of intelligent agents, reducing data redundancy and computational overhead, supporting real-time authorization decisions with high concurrency and low latency, and can dynamically adjust permissions and perform efficient identity tracking and auditing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123122836A_ABST
    Figure CN123122836A_ABST
Patent Text Reader

Abstract

This invention discloses a full-link hybrid identity authentication system for intelligent agent scenarios, comprising a runtime plane module, a policy execution point module, and a control plane module. On one hand, this invention incorporates all traffic into the control plane for management, using the gateway as the policy execution point. Requests carry only a lightweight session ID and are indexed using a context data structure. The control plane stores the full context, and policy decisions employ minimal necessary context projection, enabling the intelligent agent to transmit identity tokens without any authentication tokens throughout the process. On the other hand, through the full-link identity orchestration and link context indexing of the control plane, human identity context, NHI context, and relationship and environmental context are incorporated into a unified data structure model, achieving unified management and efficient tracking of human-machine hybrid identities.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a full-link hybrid identity authentication system for intelligent agent scenarios. Background Technology

[0002] Traditional human identity management models are only for the static management of natural persons. They have a long lifecycle and cannot reflect the working mode of intelligent agents (AI). The human identity management model cannot adapt to the dynamic behavioral characteristics of non-human subjects represented by intelligent agents. For example, using human identities when managing intelligent agents' access to IT resources can cause security and audit risks.

[0003] When the concept and related methods of NHI (Neural Identity Management) were first proposed abroad, they only addressed the management of resources and did not consider "digital populations" such as intelligent agents. As intelligent agents play a more important role in production activities, traditional human identity management models and simple NHI management are no longer sufficient to address the aforementioned security and auditing risks. Currently, mainstream identity technologies in the industry are mainly divided into two categories:

[0004] (1) Human identity management model: One of the more typical methods is the static identity mapping system (OAuth2.0, OIDC), which pre-configures the permission relationship between users, devices and resources as fixed rules and uses tokens to realize identity transfer. This type of solution manages human identity and load / service identity separately, lacks the ability to uniformly model and orchestrate hybrid identities, and can only perform one-time static authorization at the initial stage of request initiation. It cannot perform secondary verification and permission adjustment for dynamically generated behaviors during the execution of intelligent agents. At the same time, full context transmission and long token chain parsing lead to inefficient identity tracking and data redundancy. Moreover, no audit and traceability mechanism for non-human subjects has been established, making it difficult to identify the logic of "legal permissions but out-of-bounds behavior". The first category is decentralized identity solutions such as DID+VC and blockchain, which achieve user self-management through asymmetric encryption. First, it relies on blockchain to complete identity parsing, credential verification and status update. In the high-frequency interaction scenario of intelligent agents, it will generate significant transaction delay and gas cost, and cannot support high concurrency and low latency real-time authorization decision-making. Second, this solution focuses on the decentralized storage of static identity attributes and credentials, lacks the ability to dynamically model runtime session context, delegation relationship and behavioral intent, and still needs to transmit full credential data, resulting in inefficient identity tracking, data redundancy, and difficulty in dynamically verifying behavioral intent during the execution of intelligent agents. (2) Resource-oriented NHI management mode: realizes the binary identity binding of "human user + intelligent agent". It combines the workload identity of the intelligent agent itself with the identity of the human user who initiates the operation to form a unified dynamic identity at runtime. Under this solution, all operations can be traced back to the human user. It only considers simple binding and context, has a simple structure and single functional scenario. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a full-link hybrid identity authentication system for intelligent agent scenarios.

[0006] The objective of this invention is achieved through the following technical solution: a full-link hybrid identity authentication system for intelligent agent scenarios, comprising a runtime plane module, a policy execution point module, and a control plane module;

[0007] The runtime plane module consists of human identities, intelligent agents, applications, tools, and resource entities, covering the entire business chain from natural persons or business logic initiating requests to accessing target resources. Each entity corresponds to its own identity identifier and behavioral context, providing a basic data source for the construction of hybrid identities.

[0008] The policy execution point module is deployed at each identity flow stage of the runtime plane module to perform traffic interception, session ID extraction and forwarding, control plane data interaction, and permission decision execution.

[0009] The control plane module consists of full-chain identity orchestration, link context index, and decision-making mechanism. It uniformly receives and processes identity and context data reported by each gateway, constructs a human-machine hybrid identity model through full-chain identity orchestration, and performs identity tracking based on link context index.

[0010] Preferably, the specific processing steps of the strategy execution point module are as follows:

[0011] A1: The entry gateway collects human identity, application identity, and authorization intent and reports them to the control plane. The control plane assigns a unique session ID to this request and, based on the current session identifier, merges human identity, NHI, and delegation relationship data to generate or update a full-link hybrid identity instance.

[0012] A2: Extract the session ID from the request, initiate a context query request to the management plane, and the management plane extracts the minimum set of information required for decision-making from all hybrid identities based on the permission policy requirements;

[0013] A3: The projection data is reported to the control plane, which responds in real time through the rapid decision-making layer. In high-risk scenarios, the enhanced decision-making layer is triggered to supplement the context as needed or to introduce manual review. The decision results are sent to the gateway to complete access control.

[0014] Preferably, the control plane module also includes a human identity context module, an NHI context module, a relationship and environment context module, and a hybrid identity data index module;

[0015] The human identity context module is used to uniquely identify a human subject and carries their identity attributes and session association information;

[0016] The NHI context module is used to uniquely identify intelligent agents and non-human entities, and carries their runtime identity and authorization credentials;

[0017] The Relationship and Context module is used to describe human-computer delegation relationships, access link contexts, and environmental constraints.

[0018] The hybrid identity data indexing module uses the session ID as the core index key to establish an associated index of human identity context, NHI context, and relationship and environment context.

[0019] This invention has the following advantages: It overcomes the limitations of existing solutions that rely on token-based identity transmission and resource-oriented NHI management models that only cover the binary identity of "human user + intelligent agent". On the one hand, it incorporates all traffic into the management plane, using the gateway as the policy execution point, allowing requests to carry only a lightweight session ID and establishing a context data structure index. The management plane stores the full context, and the minimum necessary context projection is used when making policy decisions, realizing the transmission of intelligent agents without identity tokens throughout the process. On the other hand, through the full-chain identity orchestration and link context index of the management plane, human identity context, NHI context, and relationship and environmental context are incorporated into a unified data structure model, realizing unified management and efficient tracking of human-machine hybrid identities. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of a full-link hybrid identity authentication system architecture;

[0021] Figure 2 A schematic diagram of the strategy execution point module processing flow;

[0022] Figure 3 A schematic diagram of the control plane module system architecture. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0024] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0025] It should be noted that, unless otherwise specified, the embodiments and features described in this invention can be combined with each other.

[0026] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0027] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use, or the orientation or positional relationship commonly understood by those skilled in the art. They are only used for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention. In addition, the terms "first," "second," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0028] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0029] In this embodiment, as Figure 1 As shown, a full-link hybrid identity authentication system for intelligent agent scenarios includes a runtime plane module, a policy execution point module, and a control plane module.

[0030] The runtime plane module consists of human identities, intelligent agents, applications, tools, and resource entities, covering the entire business chain from natural persons or business logic initiating requests to accessing target resources. Each entity corresponds to its own identity identifier and behavioral context, providing a basic data source for the construction of hybrid identities. Specifically, all interactions of entities are managed through the gateway deployed on the control plane. Intelligent agents can only access the session ID throughout the process and cannot obtain any identity tokens.

[0031] The policy execution point module is deployed at each identity flow stage of the runtime plane module to perform traffic interception, session ID extraction and forwarding, control plane data interaction, and permission decision execution.

[0032] The control plane module consists of full-chain identity orchestration, link context indexing, and a decision-making mechanism. It uniformly receives and processes identity and context data reported by each gateway, constructs a human-machine hybrid identity model through full-chain identity orchestration, and performs identity tracking based on the link context index. This invention overcomes the limitations of existing solutions that rely on token-based identity transmission and resource-oriented NHI management models that only cover the binary identity of "human user + agent." On the one hand, it incorporates all traffic into the control plane management, using gateways as policy execution points, allowing requests to carry only lightweight session IDs and establishing context data structure indexes. The control plane stores the full context, and uses minimum necessary context projection during policy decision-making, achieving tokenless transmission of identity throughout the agent process. On the other hand, through the full-chain identity orchestration and link context indexing of the control plane, human identity context, NHI context, and relationship and environmental context are incorporated into a unified data structure model, achieving unified management and efficient tracking of human-machine hybrid identities.

[0033] Furthermore, such as Figure 2 As shown, the specific processing steps of the strategy execution point module are as follows:

[0034] A1: The entry gateway collects human identity, application identity, and authorization intent and reports them to the control plane. The control plane assigns a unique session ID to this request and, based on the current session identifier, merges human identity, NHI, and delegation relationship data to generate or update a full-link hybrid identity instance.

[0035] A2: Extract the session ID from the request, initiate a context query request to the management plane, and the management plane extracts the minimum set of information required for decision-making from all hybrid identities based on the permission policy requirements;

[0036] A3: The projection data is reported to the control plane. The control plane responds in real time through the rapid decision-making layer. In high-risk scenarios, the enhanced decision-making layer is triggered to supplement the context as needed or introduce manual review. The decision result is sent to the gateway to complete access control. Specifically, the policy execution point module is distributed and deployed in various identity flow links of the runtime plane (key nodes such as entry, tool call, and resource pre-positioning). All control logic is uniformly implemented by the software layer, and permission verification logic is embedded in key nodes such as intelligent agents calling tools and applications accessing resources. Access control is executed based on the decision result sent by the control plane to achieve real-time interception and control of dynamic behavior. In step A1, the request is triggered by the entry gateway. The entry gateway collects data such as human identity, application identity, and authorization intent and reports it to the control plane. The control plane assigns a unique session ID to this request and, based on the current session identifier, integrates human identity, NHI, and delegation relationship data to generate or update a full-link hybrid identity instance, providing a unified identity model for subsequent authorization decisions; the gateway sends the session ID to the gateway. The information is attached to the request and forwarded without transmitting any identity token. In step A2, this step is executed by the corresponding node gateway. This step avoids the transmission of the full context and token chain, reducing data redundancy and computational overhead, and providing lightweight input for policy decision-making. In step A3, this step completes the permission determination based on the projected context. That is, the gateway reports the projected data to the control plane. The control plane first achieves real-time response through the fast decision layer. In high-risk scenarios, the enhanced decision layer is triggered to supplement the context as needed or introduce manual review. The decision result (allow / reject / mark) is sent to the gateway to complete access control. At the same time, the decision basis and execution result are recorded throughout the process and synchronized to the audit module to form a traceable evidence chain. All decision-making processes are based on the full context of the control plane. Neither the gateway nor the agent parses or transmits any identity token.

[0037] Furthermore, the control plane module also includes a human identity context module, an NHI context module, a relationship and environment context module, and a hybrid identity data index module;

[0038] The human identity context module is used to uniquely identify a human subject and carries their identity attributes and session association information;

[0039] The NHI context module is used to uniquely identify intelligent agents and non-human entities, and carries their runtime identity and authorization credentials;

[0040] The Relationship and Context module is used to describe human-computer delegation relationships, access link contexts, and environmental constraints.

[0041] The hybrid identity data indexing module uses the session ID as the core index key to establish an associated index of human identity context, NHI context, and relationship and environment context. Specifically, the human identity context module stores core fields such as natural person identifier, tenant / organization, role, attributes, and human-machine session identifier; the NHI context module stores fields such as subject identifier, instance identifier, description, attributes, credential / key reference, and session parameters; the relationship and environment context module stores fields such as delegation / representation edges, session / link / relationship, resource attributes, scope, and time information. The hybrid identity data indexing module uses the session ID as the core index key to establish an associated index of human identity context, NHI context, and relationship and environment context, enabling efficient identity query and end-to-end traceability, supporting gateway permission decisions and audit attribution. All index data is synchronized to the distributed cache of the control plane in real time to ensure query efficiency under high concurrency.

[0042] In this embodiment, taking a typical scenario of an agent calling financial MCP resources as an example: Human user U initiates a request through the application, authorizing agent A to only call tool 1 to read public resource R1. However, during execution, agent A is induced to generate an illegal plan and attempts to call tool 2 to access sensitive resource R2. If the resource-oriented NHI management mode is adopted, the agent can directly obtain and pass the token, which poses a risk of leakage. Moreover, it can only record the shallow authorization relationship of "user U → agent A" and cannot trace the complete operation chain of "application → agent A → tool 2 → resource R2". It is impossible to intercept unauthorized behavior in real time during the tool call process and impossible to locate the responsible party afterward.

[0043] After deploying the end-to-end hybrid identity authentication method of this invention, the ingress gateway collects data and reports it to the control plane when a request is initiated. The control plane assigns a unique session ID S12345 and generates a hybrid identity instance. Figure 3 The hybrid identity data structure uses S12345 as the core index, fully recording the human identity context, agent identity context, and delegation relationship, and the session-level identity chain is maintained by the link context index of the control plane; agent A can only access session ID S12345 and does not transmit any tokens. Figure 2In the gateway policy execution process, the tool call gateway and the financial resource front-end gateway sequentially extract S12345 and query the minimum necessary context from the control plane. Only decision fields such as "user authorization scope, tool type, and resource sensitivity" are extracted. The decision-making mechanism of the control plane makes a quick judgment based on the projection data and identifies that the behavior of agent A calling tool 2 to access R2 exceeds the authorization scope. The financial resource front-end gateway blocks the access in real time and marks it as abnormal. When security personnel investigate, they can trace back from resource R2 to user U through session ID S12345 to fully restore the operation link of "user U → application → agent A → tool 2 → resource R2". At the same time, since only session ID and minimum necessary context are transmitted, the system overhead is also low.

[0044] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A full-link hybrid identity authentication system for intelligent agent scenarios, characterized in that: This includes the runtime plane module, the policy execution point module, and the control plane module; The operational plane module consists of human identities, intelligent agents, applications, tools, and resource entities, covering the entire business chain from natural persons or business logic initiating requests to accessing target resources. Each entity corresponds to its own identity identifier and behavioral context, providing a basic data source for the construction of hybrid identities. The policy execution point module is deployed at each identity transfer stage of the running plane module to perform traffic interception, session ID extraction and forwarding, control plane data interaction and permission decision execution. The control plane module consists of full-chain identity orchestration, link context index, and decision mechanism. It uniformly receives and processes identity and context data reported by each gateway, constructs a human-machine hybrid identity model through full-chain identity orchestration, and performs identity tracking based on link context index.

2. The end-to-end hybrid identity authentication system for intelligent agent scenarios according to claim 1, characterized in that: The specific processing steps of the strategy execution point module are as follows: A1: The entry gateway collects human identity, application identity, and authorization intent and reports them to the control plane. The control plane assigns a unique session ID to this request and, based on the current session identifier, merges human identity, NHI, and delegation relationship data to generate or update a full-link hybrid identity instance. A2: Extract the session ID from the request, initiate a context query request to the management plane, and the management plane extracts the minimum set of information required for decision-making from all hybrid identities based on the permission policy requirements; A3: The projection data is reported to the control plane, which responds in real time through the rapid decision-making layer. In high-risk scenarios, the enhanced decision-making layer is triggered to supplement the context as needed or to introduce manual review. The decision results are sent to the gateway to complete access control.

3. The end-to-end hybrid identity authentication system for intelligent agent scenarios according to claim 2, characterized in that: The control plane module also includes a human identity context module, an NHI context module, a relationship and environment context module, and a hybrid identity data index module; The human identity context module is used to uniquely identify a human subject and carries their identity attributes and session association information; The NHI context module is used to uniquely identify intelligent agents and non-human entities, and carries their runtime identity and authorization credentials; The relationship and environment context module is used to describe the human-computer delegation relationship, access link context, and environmental constraints. The hybrid identity data indexing module uses the session ID as the core index key to establish an association index of human identity context, NHI context, and relationship and environment context.