Intelligent agent control method and device

Abstract

The embodiment of the invention discloses an agent control method and device, and the method enables a management and control strategy to be deployed in a security gateway, and achieves the unified management and control of different agent platforms through the security gateway. Meanwhile, according to the method, business logic decoupling of the management and control mechanism and the agent platform is achieved through safety section instrumentation, and the integration difficulty and the maintenance cost are reduced. The intelligent agent control device disclosed by the embodiment of the specification also has the above beneficial effects.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an intelligent body control method and device. Background Art

[0002] In an agent platform, agents usually have higher permissions than ordinary users. Attackers can construct prompt words to command agents to perform operations that they cannot directly access, thereby obtaining private data. This attack method is called an agent proxy attack. In order to combat agent proxy attacks, an independent security control solution is usually deployed on each agent platform. However, in a multi-agent platform scenario, this method is difficult to achieve unified control, and the upgrade and maintenance of the security control solution is time-consuming and labor-intensive. Summary of the invention

[0003] One or more embodiments of the present specification provide an intelligent agent control method and device to at least partially solve the above technical problems.

[0004] In a first aspect, an intelligent agent control method is provided, which is applicable to at least one intelligent agent platform, and the method comprises: For a target agent in the agent platform, determining an access control point for the target agent; configuring an access control policy for the access control point through a security gateway; Determine a point-of-cut implantation position based on the access control point, and implant a point-of-cut of a security aspect at the point-of-cut implantation position; When the intelligent agent platform runs to the cut-off point, the security cut-off point is injected into the cut-off point, the context information of the cut-off point is collected using the security cut-off point, and the access control policy is queried at the security gateway based on the context information, and the access control policy is executed.

[0005] As an optional implementation manner of the method of the first aspect, for a target agent in the agent platform, determining an access control point for the target agent specifically includes: At least one of the creation function, dialogue creation function, interface calling function and tool calling function of the target intelligent agent is used as an access control point for the target intelligent agent.

[0006] As an optional implementation manner of the method of the first aspect, the method further includes: Before implanting the cut point of the security cut plane at the cut point implantation position, the security cut plane is constructed according to the requirements of the access control policy.

[0007] As an optional implementation manner of the method of the first aspect, the method further includes: In response to the target agent executing the target operation, the security gateway allocates a unique link identifier to the operation link of the target operation, and performs link tracking on the operation link of the target operation based on the link identifier.

[0008] As an optional implementation manner of the method of the first aspect, the method further includes: The intelligent agent platform loads the security aspect SDK when it is started, and the security aspect SDK is used to implement the configuration and injection of the security aspect.

[0009] In a second aspect, an intelligent agent control device is provided, which is applicable to at least one intelligent agent platform, and the device comprises: A data acquisition module, used to acquire an access control point for a target agent in the agent platform; A security gateway, used to configure an access control policy for the access control point; A security aspect injection module is used to determine a point of entry implantation location based on the access control point, and implant a point of entry of the security aspect at the point of entry implantation location; when the intelligent agent platform runs to the point of entry, the security aspect is injected into the point of entry, the context information of the point of entry is collected using the security aspect, and the access control policy is queried at the security gateway based on the context information, and the access control policy is executed.

[0010] As an optional implementation of the device described in the second aspect, the data acquisition module is specifically used to use at least one of the creation function, dialogue creation function, interface calling function, and tool calling function of the target intelligent entity as an access control point for the target intelligent entity.

[0011] As an optional implementation manner of the device described in the second aspect, the security aspect injection module is also used to construct the security aspect according to the requirements of the access control policy before implanting the cut point of the security aspect at the cut point implantation position.

[0012] As an optional implementation of the device described in the second aspect, the security gateway is also used to allocate a unique link identifier to the operation link of the target operation in response to the target agent performing the target operation, and perform link tracking on the operation link of the target operation based on the link identifier.

[0013] As an optional implementation of the device described in the second aspect, the security aspect injection module is implemented through a security aspect SDK, and the intelligent agent platform loads the security aspect SDK when it starts.

[0014] According to a third aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed on an electronic device, the electronic device executes the above-mentioned intelligent agent control method.

[0015] In a fourth aspect, an electronic device is provided, including: at least one memory for storing a program; At least one processor is used to execute the program stored in the memory. When the program stored in the memory is executed, the processor is used to execute the above-mentioned intelligent body control method.

[0016] The beneficial effect of the intelligent agent control method described in the embodiment of this specification is that the method deploys the control strategy on the security gateway, and realizes the unified control of different intelligent agent platforms through the security gateway. At the same time, the method realizes the decoupling of the control mechanism and the business logic of the intelligent agent platform through the security aspect plugging, which reduces the integration difficulty and maintenance cost.

[0017] The intelligent body control device described in the embodiments of this specification also has the above-mentioned beneficial effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 The flowchart of an intelligent agent control method described in an embodiment of this specification is exemplarily shown.

[0020] Figure 2 The structural diagram of the intelligent agent control system described in the embodiment of this specification is exemplarily shown.

[0021] Figure 3 The schematic diagram of the structure of an intelligent body control device described in an embodiment of this specification is exemplarily shown.

[0022] Figure 4 The structural diagram of an electronic device described in an embodiment of this specification is exemplarily shown. DETAILED DESCRIPTION

[0023] First of all, it should be noted that the terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms "a", "the" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates otherwise.

[0024] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Therefore, it should be recognized by those skilled in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0025] It should be noted that: in other embodiments, the steps of the corresponding method are not necessarily performed in the order shown and described in this specification. In some other embodiments, the steps included in the method may be more or less than those described in this specification. In addition, a single step described in this specification may be decomposed into multiple steps for description in other embodiments; and multiple steps described in this specification may be combined into a single step for description in other embodiments.

[0026] AI Agent is a system composed of a large language model (LLM), memory, action planning, and tool usage (API calls). It completes most of the work through conditional automation, while humans are responsible for setting goals, providing resources, and supervising results. LLM acts as the brain of the agent, responsible for task splitting, tool selection and calling, and progress control, and finally ends the work autonomously after achieving the goal.

[0027] In practical applications, AI agents may be subject to agent proxy attacks. Agent proxy attacks refer to attackers using prompts to command AI agents to perform operations that they cannot directly access. Under normal circumstances, ordinary users cannot access sensitive or high-risk APIs, but AI agents may have higher permissions, such as access to sensitive APIs or data, and attackers can use this to obtain private data.

[0028] In order to resist such attacks, security control solutions are usually deployed on the intelligent agent platform. At present, the conventional practice is to deploy an independent security control solution for each intelligent agent platform. However, there is usually more than one intelligent agent platform in an enterprise. This method of deploying an independent security control solution for each intelligent agent platform makes it difficult to maintain the consistency of the security control solution. When the security control solution is upgraded, each intelligent agent platform needs to be maintained separately, which is time-consuming and labor-intensive.

[0029] In addition, current security control solutions usually only control the input and output of intelligent agents, which makes it difficult to cover the ever-changing attack methods.

[0030] In view of this, one or more embodiments of the present specification propose an intelligent agent control method and device, which can at least partially solve the above-mentioned technical problems and is particularly suitable for multi-agent centralized control scenarios.

[0031] The intelligent body control method and device described in one or more embodiments of this specification will be further described in detail below in conjunction with the drawings and specific embodiments of the specification, but this detailed description does not constitute a limitation on the embodiments of this specification.

[0032] Please refer to Figure 1 , Figure 1 This is a flow chart of an intelligent agent control method proposed in one or more embodiments of this specification. The method can be applicable to a single intelligent agent platform or to the centralized control of multiple intelligent agent platforms.

[0033] like Figure 1 As shown, the method may include steps S100 to S106: S100: For a target agent in the agent platform, determine an access control point for the target agent.

[0034] The above-mentioned intelligent agent platform refers to the platform on which the target intelligent agent is deployed. To realize a secure intelligent agent platform, it is necessary to meet the OVTP (Operator-Voucher-Traceable Paradigm) traceability paradigm, that is, to fully and accurately judge whether a network access is legal, it should be based on the access link information of the operator (Operator) of the access and the transmission link information of the voucher (Voucher). In this implementation, in order to make the intelligent agent platform meet the OVTP traceability paradigm, the full coverage of the target intelligent agent life cycle can be achieved by setting corresponding access control points. For example, the creation, dialogue creation, tool call, interface call and other events of the target intelligent agent can be used as nodes that need to be controlled, that is, the operations / events that may occur in the target intelligent agent are traversed, and some or all of them are selected for control as needed. Based on this, at least one of the creation function, dialogue creation function, interface call function and tool call function of the target intelligent agent can be used as an access control point for the target intelligent agent.

[0035] It should be noted that the above-mentioned access control points can be adaptively set according to needs, and this embodiment does not limit this.

[0036] S102: configuring an access control policy for an access control point through a security gateway.

[0037] This embodiment achieves unified management and control of all agent platforms by deploying access control policies for access control points of each target agent on the security gateway. In this way, the security team can directly configure the access control policies of each agent platform on the security gateway. For the centralized management and control scenarios of multiple agent platforms that need to maintain access control policy consistency, it is very convenient to achieve consistency of access control policies between different agent platforms.

[0038] The above-mentioned access control policies can be set according to the type of access control point. For example, for the target agent creating a dialogue event, you can set an access control policy to control whether the dialogue is allowed to be created. For tool or interface call events, you can set an access control policy to verify the target agent's operating permissions. For the interactive content between the target agent and the user, you can set an access control policy to detect whether there are keywords or patterns in the input content and / or output content of the target agent that may cause inappropriate behavior. You can also use keyword screening and behavior analysis to detect whether there are potential malicious activities in the input content and / or output content of the target agent, and initiate alarms and interventions when suspicious content is detected.

[0039] It should be noted that the above access control policy can be adaptively set according to needs, and this embodiment does not limit this.

[0040] S104: Determine a point-of-cut implantation position based on the access control point, and implant a point-of-cut of the security aspect at the point-of-cut implantation position.

[0041] After determining the access control point, you can build a security aspect according to the requirements of the access control policy, then determine the injection point from the access control point, and associate the injection point with the security aspect's point of contact, thereby implementing the security aspect's point of contact implantation.

[0042] Specifically, the implantation position of the pointcut can be set according to the parameters required by the access control policy corresponding to the access control point. For example, for the creation event of the target intelligent body, a pointcut can be set in the creation function of the target intelligent body, and the security aspect can be injected through the pointcut to collect the creator id, administrator information, model name, system instructions, and tool name and tool id of the target intelligent body. For the dialogue creation event, a pointcut can be set in the dialogue creation function, and the security aspect can be injected through the pointcut to collect the user id of the dialogue creation, the target intelligent body id and other information. For the tool / interface call event, a pointcut can be set in the tool / interface call function, and the security aspect can be injected through the pointcut to collect the name and id of the tool / interface and the operation permission credentials of the target intelligent body and other information.

[0043] S106: When the intelligent agent platform runs to the cut-off point, the security cut-off point is injected into the cut-off point, the context information of the cut-off point is collected using the security cut-off point, and the access control policy is queried in the security gateway based on the context information, and the access control policy is executed.

[0044] The above-mentioned security aspect is used to obtain the context information of the access control point, and then the control result of the access control policy can be queried based on the context information. For example, for the creation event of the target intelligent body, after the security aspect collects the creator id, administrator information, model name, system instructions, and tool name and tool id of the target intelligent body, it can upload this information to the security gateway, so as to realize the initialization of the target intelligent body in the security gateway, so that the security gateway can monitor the operation / access behavior of the target intelligent body. For the dialogue creation event, after the security aspect collects the user id and target intelligent body id of the dialogue creation, it queries the security gateway based on this information whether the dialogue is allowed to be created. If it is allowed, the security aspect releases the dialogue creation operation, and the security gateway can record the dialogue creation for link tracking. If it is not allowed, the security aspect intercepts the dialogue creation operation. For the tool / interface call event, after the security aspect collects the name and id of the tool / interface and the operation permission credentials of the target intelligent body, it can query the corresponding security control policy on the security gateway based on this information to determine whether the target intelligent body has the authority to call the tool / interface. If permission is granted, the tool / interface call operation is allowed, and the security gateway can record the tool / interface call information for link tracking; if permission is not granted, the security aspect intercepts the tool / interface call operation.

[0045] In some embodiments, in order to make the above-mentioned agent platform meet the traceability paradigm of OVTP (Operator-Voucher-Traceable Paradigm), it is also necessary to ensure the traceability of the agent operation link, so as to ensure the ability to respond quickly to emergencies. In this embodiment, a complete audit and logging mechanism can be established. When a security incident occurs, the specific operator, the target agent performing the operation, the operation and related credentials can be traced. Specifically, when the target agent performs a preset target operation, a unique link identifier can be assigned to the operation link of the target operation through the security gateway, and the operation link of the target operation can be tracked based on the link identifier, and the link tracking data can be recorded through the log. For example, in order to ensure that each call link can be traced, a link identifier can be added to the header of all http / gRPC requests to achieve link tracking.

[0046] The above-mentioned target operation can be adaptively set according to needs, for example, it can be a tool / function calling operation, an intranet database access operation, etc., which is not limited in this embodiment.

[0047] For the above-mentioned control method of an intelligent agent, in some implementations, the configuration and injection functions of the security aspect can be implemented through the security aspect SDK. Figure 2 , Figure 2 An intelligent agent control system for implementing the above-mentioned intelligent agent management and control method is shown, in which several intelligent agent platforms are included, and the security aspect SDK is loaded in these intelligent agent platforms. The access control policy corresponding to the security aspect SDK is uniformly managed by the security gateway. The security aspect SDK is started synchronously when the Java / Python application of the intelligent agent platform is started, and the security gateway is notified that the intelligent agent platform has been started through the intelligent agent platform identifier PlatformID. Assuming that the intelligent agent platform needs to be logged in, if no user logs in after startup, a default user with the user id of admin is initialized.

[0048] The above is the intelligent agent control method described in this embodiment. This method utilizes the plug-in capability of the security aspect to implant the security aspect into the key functions within the life cycle of the intelligent agent, and adds the logic of querying and executing the access control policy deployed by the security gateway to the enhanced logic of the security aspect, so as to achieve unified management and control of the intelligent agent. This solution can not only verify the input and output of the intelligent agent, but also manage the operation of the intelligent agent and API / tool ​​calls, thereby achieving security management and control of the entire life cycle of the intelligent agent. This solution is decoupled from the business logic of the intelligent agent platform, and the security mechanism is implemented through the security aspect and the security gateway. There is no need to modify the code of the original intelligent agent platform, which reduces the integration difficulty and maintenance cost.

[0049] Corresponding to the above-mentioned intelligent agent control method, this embodiment also provides an intelligent agent control device, which is applicable to at least one intelligent agent platform. Figure 3 , Figure 3 The schematic diagram of the structure of an intelligent agent control device is shown as an example, which can be used to implement the above intelligent agent control method. It should be noted that the intelligent agent control method described in one or more embodiments of the present application can rely on Figure 3 The intelligent agent control device shown is implemented, but not limited to this device.

[0050] like Figure 3 As shown, the intelligent body control device includes: The data acquisition module 301 is used to acquire the access control point for the target agent in the agent platform.

[0051] The security gateway 302 is used to configure access control policies for access control points.

[0052] The security aspect injection module 303 is used to determine the point-of-entry implantation location based on the access control point, and implant the point-of-entry of the security aspect at the point-of-entry implantation location. When the intelligent platform runs to the point-of-entry, the security aspect is injected into the point-of-entry, and the context information of the point-of-entry is collected using the security aspect. The access control policy is queried at the security gateway based on the context information, and the access control policy is executed.

[0053] With respect to the above-mentioned data acquisition module 301, the data acquisition module 301 can achieve comprehensive coverage of the target intelligent agent's life cycle by setting corresponding access control points, so that the intelligent agent platform meets the OVTP traceability paradigm. For example, the data acquisition module 301 can use events such as the creation, dialogue creation, tool call, interface call, etc. of the target intelligent agent as nodes that need to be managed, that is, traverse the operations / events that may occur in the target intelligent agent, and select some or all of them for management and control as needed. Based on this, the data acquisition module 301 can use at least one of the creation function, dialogue creation function, interface call function, and tool call function of the target intelligent agent as an access control point for the target intelligent agent.

[0054] For the above-mentioned security gateway 302, the access control policy configured by the security gateway 302 can be set according to the type of access control point. For example, for the target agent to create a dialogue event, the security gateway 302 can control whether the dialogue is allowed to be created by setting an access control policy. For tool or interface call events, the security gateway 302 can verify the target agent's operating authority by setting an access control policy. For the interactive content between the target agent and the user, the security gateway 302 can detect whether there are keywords or patterns that may cause improper behavior in the input content and / or output content of the target agent by setting an access control policy, and can also use keyword screening and behavior analysis to detect whether there are potential malicious activities in the input content and / or output content of the target agent, and initiate alarms and interventions when suspicious content is detected.

[0055] It should be noted that the access control policy deployed by the security gateway 302 can be adaptively set according to the requirements, and this embodiment does not limit this.

[0056] In some embodiments, the security gateway 302 can also be used to respond to the target agent executing a preset target operation, assign a unique link identifier to the operation link of the target operation, and perform link tracking on the operation link of the target operation based on the link identifier. The above-mentioned target operation can be adaptively set according to demand, for example, it can be a tool / function call operation, an intranet database access operation, etc., and this embodiment does not limit this.

[0057] With respect to the above-mentioned security aspect injection module 303, the security aspect injection module 303 can set the implantation position of the cut point according to the parameters required by the access control policy corresponding to the access control point obtained by the data acquisition module 301. Specifically, the security aspect injection module 303 can construct the security aspect according to the requirements of the access control policy, and then determine the injection point from the access control point, and associate the injection point with the cut point of the security aspect, so as to realize the implantation of the security aspect cut point.

[0058] The security aspect constructed by the security aspect injection module 303 is used to obtain the context information of the access control point, and then the control result of the access control policy can be queried based on the context information. For example, for the creation event of the target intelligent body, after the security aspect collects the creator id, administrator information, model name, system instructions, and tool name and tool id of the target intelligent body, the information can be uploaded to the security gateway, thereby realizing the initialization of the target intelligent body in the security gateway, so that the security gateway can monitor the operation / access behavior of the target intelligent body. For the dialogue creation event, after the security aspect collects the user id and target intelligent body id of the dialogue creation, the security aspect queries the security gateway based on the information whether the dialogue is allowed to be created. If allowed, the security aspect releases the dialogue creation operation, and the security gateway can record the dialogue creation for link tracking. If not allowed, the security aspect intercepts the dialogue creation operation. For the tool / interface call event, after the security aspect collects the name and id of the tool / interface and the operation permission certificate of the target intelligent body, the corresponding security control policy can be queried based on the information to determine whether the target intelligent body has the authority to call the tool / interface. If permission is granted, the tool / interface call operation is allowed, and the security gateway can record the tool / interface call information for link tracking; if permission is not granted, the security aspect intercepts the tool / interface call operation.

[0059] Correspondingly, for the creation event of the target intelligent body, the security aspect injection module 303 can set a cut-off point in the creation function of the target intelligent body, and inject the security aspect through the cut-off point to collect the creator id, administrator information, model name, system instructions, and tool name and tool id of the target intelligent body. For the dialogue creation event, the security aspect injection module 303 can set a cut-off point in the dialogue creation function, and inject the security aspect through the cut-off point to collect the user id of the dialogue creation, the target intelligent body id and other information. For the tool / interface call event, the security aspect injection module 303 can set a cut-off point in the tool / interface call function, and inject the security aspect through the cut-off point to collect the name and id of the tool / interface and the target intelligent body's operation permission credentials and other information.

[0060] In some implementations, the above-mentioned security aspect injection module 303 can be implemented through a security aspect SDK. The intelligent agent platform loads the security aspect SDK when it starts up to implement the function of the security aspect injection module 303.

[0061] For the above-mentioned intelligent body control device, taking the module as an example of a software functional unit, the data acquisition module 301 may include code running on a computing instance. Among them, the computing instance may include at least one of a physical host (computing device), a virtual machine, and a container. Further, the above-mentioned computing instance may be one or more. For example, the data acquisition module 301 may include code running on multiple hosts / virtual machines / containers. The multiple hosts / virtual machines / containers used to run the code may be distributed in the same region (region) or in different regions. Furthermore, the multiple hosts / virtual machines / containers used to run the code may be distributed in the same availability zone (AZ) or in different AZs, each AZ including one data center or multiple data centers with close geographical locations. Among them, usually a region may include multiple AZs.

[0062] Similarly, multiple hosts / virtual machines / containers used to run the code can be distributed in the same virtual private cloud (VPC) or in multiple VPCs. Usually, a VPC is set up in a region. For cross-region communication between two VPCs in the same region and between VPCs in different regions, a communication gateway needs to be set up in each VPC to achieve interconnection between VPCs through the communication gateway.

[0063] As an example of a hardware functional unit, the data acquisition module 301 may include at least one computing device, such as a server, etc. Alternatively, the data acquisition module 301 may also be a device implemented using an application-specific integrated circuit (ASIC) or a programmable logic device (PLD). The PLD may be a complex programmable logical device (CPLD), a field programmable gate array (FPGA), a generic array logic (GAL) or any combination thereof.

[0064] The multiple computing devices included in the data acquisition module 301 can be distributed in the same region or in different regions. The multiple computing devices included in the data acquisition module 301 can be distributed in the same AZ or in different AZs. Similarly, the multiple computing devices included in the data acquisition module 301 can be distributed in the same VPC or in multiple VPCs. The multiple computing devices can be any combination of computing devices such as servers, ASICs, PLDs, CPLDs, FPGAs, and GALs.

[0065] In other embodiments, the data acquisition module 301 can be used to execute any step in the above-mentioned intelligent agent control method, the security gateway 302 can be used to execute any step in the above-mentioned intelligent agent control method, and the security aspect injection module 303 can be used to execute any step in the above-mentioned intelligent agent control method. The steps that the data acquisition module 301, the security gateway 302 and the security aspect injection module 303 are responsible for implementing can be specified as needed, and the data acquisition module 301, the security gateway 302 and the security aspect injection module 303 respectively implement different steps in the above-mentioned intelligent agent control method to realize all the functions of the above-mentioned intelligent agent control device.

[0066] In this implementation, the intelligent body control device can also be applied to computing devices such as computers and servers, or to a computing device cluster including at least one computing device, to realize the intelligent body control function.

[0067] In some embodiments, an electronic device is also provided. Figure 4 , the electronic device includes: a bus 401, a processor 402, a memory 403 and a communication interface 404. The processor 402, the memory 403 and the communication interface 404 communicate with each other through the bus 401. The electronic device can be a server or a terminal device. It should be understood that the present application does not limit the number of processors and memories in the electronic device.

[0068] The bus 401 may be a peripheral component interconnect (PCI) bus, or an extended industry standard architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 The bus 401 is represented by only one line, but it does not mean that there is only one bus or one type of bus. The bus 401 may include a path for transmitting information between various components of the electronic device (for example, the processor 402, the memory 403 and the communication interface 404).

[0069] The processor 402 may include any one or more processors such as a CPU, a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP).

[0070] The memory 403 may include a volatile memory, such as a random access memory (RAM). The memory 403 may also include a non-volatile memory, such as a read-only memory (ROM), a flash memory, a hard disk drive (HDD), or a solid state drive (SSD).

[0071] The memory 403 stores executable program codes, and the processor 402 executes the executable program codes to respectively implement the functions of the aforementioned data acquisition module 301, the security gateway 302 and the security aspect injection module 303, that is, to implement the functions of the aforementioned intelligent body control device, thereby implementing the aforementioned intelligent body control method.

[0072] The communication interface 404 uses a transceiver module such as, but not limited to, a network interface card or a transceiver to implement communication between the electronic device and other devices or a communication network.

[0073] In some embodiments, a computer-readable storage medium is also provided, on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned intelligent agent control method is implemented.

[0074] The computer-readable storage medium may be any available medium that can be stored in an electronic device or a data storage device such as a data center that includes one or more available media. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk). The computer-readable storage medium includes instructions that instruct the electronic device to execute the above-mentioned intelligent agent control method.

[0075] It is to be understood that the structure illustrated in the embodiments of this specification does not constitute a specific limitation on the system of the embodiments of this specification. In other embodiments of the specification, the above system may include more or fewer components than shown in the figure, or combine some components, or split some components, or arrange the components differently. The components shown in the figure may be implemented in hardware, software, or a combination of software and hardware.

[0076] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0077] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0078] It should be noted that the above examples are only specific embodiments of the present invention, and the present invention is obviously not limited to the above examples, and there are many similar variations. All variations directly derived or associated from the contents disclosed by the technicians in this field should fall within the protection scope of the present invention.

Claims

1. An intelligent agent control method, applicable to at least one intelligent agent platform, the method comprising: For a target agent in the agent platform, determining an access control point for the target agent; configuring an access control policy for the access control point through a security gateway; Determine a point-of-cut implantation position based on the access control point, and implant a point-of-cut of a security aspect at the point-of-cut implantation position; When the intelligent agent platform runs to the cut-off point, the security cut-off point is injected into the cut-off point, the context information of the cut-off point is collected using the security cut-off point, and the access control policy is queried at the security gateway based on the context information, and the access control policy is executed.

2. The method according to claim 1, for a target agent in the agent platform, determining an access control point for the target agent, specifically comprising: At least one of the creation function, dialogue creation function, interface calling function and tool calling function of the target intelligent agent is used as an access control point for the target intelligent agent.

3. The method according to claim 1, further comprising: Before implanting the cut point of the security cut plane at the cut point implantation position, the security cut plane is constructed according to the requirements of the access control policy.

4. The method according to claim 1, further comprising: In response to the target agent executing the target operation, the security gateway assigns a unique link identifier to the operation link of the target operation, and performs link tracking on the operation link of the target operation based on the link identifier.

5. The method according to claim 1, further comprising: The intelligent agent platform loads the security aspect SDK when it is started, and the security aspect SDK is used to implement the configuration and injection of the security aspect.

6. An intelligent agent control device, applicable to at least one intelligent agent platform, the device comprising: A data acquisition module, used to acquire an access control point for a target agent in the agent platform; A security gateway, used to configure an access control policy for the access control point; A security aspect injection module is used to determine a point of entry implantation location based on the access control point, and implant a point of entry of the security aspect at the point of entry implantation location; when the intelligent agent platform runs to the point of entry, the security aspect is injected into the point of entry, the context information of the point of entry is collected using the security aspect, and the access control policy is queried at the security gateway based on the context information, and the access control policy is executed.

7. According to the device of claim 6, the data acquisition module is specifically used to use at least one of the creation function, dialogue creation function, interface calling function, and tool calling function of the target intelligent entity as an access control point for the target intelligent entity.

8. According to the device of claim 6, the security aspect injection module is also used to construct the security aspect according to the requirements of the access control policy before implanting the cut point of the security aspect at the cut point implantation position.

9. According to the device according to claim 6, the security gateway is also used to allocate a unique link identifier to the operation link of the target operation in response to the target agent performing the target operation, and perform link tracking on the operation link of the target operation based on the link identifier.

10. According to the device of claim 6, the security aspect injection module is implemented through the security aspect SDK, and the intelligent agent platform loads the security aspect SDK when starting.

11. A computer-readable storage medium storing a computer program, wherein when the computer program is executed on an electronic device, the electronic device executes the method according to any one of claims 1 to 5.

12. An electronic device comprising: at least one memory for storing a program; At least one processor is used to execute the program stored in the memory, and when the program stored in the memory is executed, the processor is used to execute the method according to any one of claims 1 to 5.

Images