An agent clustering collaboration and skill migration learning method and system

Abstract

This invention discloses a method and system for intelligent agent cluster collaboration and skill transfer learning. The method includes: supporting the deployment of multiple intelligent agent instances by department, project, or scenario and registering them to a cluster registry. The cluster registry contains instance identifiers, instance names, departments or projects, and online status. Instances within the cluster discover each other and negotiate task allocation through the registry. It adopts three collaborative modes: task delegation, result sharing, and parallel task allocation, corresponding to pipeline, parallel, and master-slave architectures. The participant types on the skill exchange platform are expanded to include staff and intelligent agent instances, with `participant_id` corresponding to employee or instance identifiers. Intelligent agents or employees publish "I want to learn" or "I can teach," and the platform matches them based on ability tags or skill identifiers. The requesting party initiates a migration request to the provider, who returns a skill definition containing skill identifiers, triggers, and processing logic. After review, this is included in the skill registry. Employees convert SOPs into skill definitions and publish "I can teach." After intelligent agents publish "I want to learn," they are matched with employees. The migration request and the employee return skill definitions, which are then reviewed to form executable skills, realizing human-teaching intelligent agents. This invention achieves cluster formation, feasible collaboration, exchangeable and transferable skills, and feasible implementation of human-teaching intelligent agents.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence and business-finance integration technology, specifically to a method and system for intelligent agent cluster collaboration and skill transfer learning. Background Technology

[0002] 2.1 Overview of Prior Technology In business and finance scenarios, various intelligent agents, such as those for intelligence, pricing, and compliance, need to collaborate, and their skills must be transferable and shareable within the enterprise. Current technologies often deploy intelligent agents as single instances, making it impossible to deploy multiple instances by department, project, or scenario to form a cluster. Intelligent agents lack collaborative models such as task delegation, result sharing, and parallel division of labor, hindering the formation of pipelined, parallel, or master-slave collaboration. Skill exchange platforms mostly only support employees posting "I want to learn" or "I can teach," preventing intelligent agents from joining as equal participants. Intelligent agents cannot proactively post learning needs or teachable skills, nor can they learn from other intelligent agents or employees, hindering skill transfer between agents. Skill transfer content is incomplete, lacking a complete skill definition including skill identifiers, triggers, and processing logic; there is no review mechanism after migration, preventing the formation of executable skills.

[0003] 2.2 Background Technology Pain Points Summary of the Invention

[0004] 3.1 Technical Issues The present invention aims to solve at least one of the following technical problems existing in the above-mentioned background technology: intelligent agents are mostly single instances, and cannot be deployed in multiple instances by department, project or scenario to form a cluster, and cannot discover each other or negotiate the division of labor; intelligent agents lack collaborative modes such as task delegation, result sharing, and parallel division of labor, and cannot form pipeline, parallel or master-slave collaboration; skill exchange platforms only support mutual assistance among employees, and intelligent agents cannot join as participants on an equal footing; intelligent agents cannot publish "I want to learn" or "I can teach", and cannot learn from other intelligent agents or employees, and skills cannot be transferred between intelligent agents; the content of skill transfer is incomplete, lacking a complete skill definition including skill identifiers, triggers and processing logic, and there is no review mechanism after migration, so it is impossible to form executable skills.

[0005] 3.2 Technical Solution This invention provides a method for intelligent agent cluster collaboration and skills transfer learning, including: supporting the deployment of multiple intelligent agent instances by department, project, or scenario; registering each instance to a cluster registry, which includes at least instance identifier, instance name, department or project, and online status; instances within the cluster discover each other and negotiate task allocation through the cluster registry; supporting collaboration between intelligent agent instances using three collaboration modes (task delegation, result sharing, and parallel task allocation), corresponding to pipeline, parallel, and master-slave; expanding the participant types of the skills exchange platform to two types: staff and intelligent agent instances, with participant_id corresponding to employee or instance identifier, allowing intelligent agents and employees to participate in skills exchange on an equal footing; intelligent agent instances or employees publish "I want to learn" or "I can teach" to the skills exchange platform, which matches based on ability tags or skill identifiers, and the requesting party initiates a skills transfer request to the provider, who returns a skills definition containing skill identifier, trigger, and processing logic; the requesting party submits the skills definition to the skills registry, which is then reviewed and included in the skills registry; employees can use enterprise SOPs... Alternatively, the operation instructions can be converted into skill definitions and published as "I can teach". After the agent instance publishes "I want to learn", the platform matches it with an employee. The agent initiates a skill migration request to the employee, who returns the skill definition. After review, the skill is formed into an executable skill, thus realizing the human-teaching agent.

[0006] 3.3 Beneficial Effects The beneficial effects of this invention include: cluster formation, with multi-instance deployment and a cluster registry enabling agent instances to discover each other and negotiate division of labor; collaborative implementation, with three modes—task delegation, result sharing, and parallel division of labor—supporting an intelligence-pricing-approval pipeline, multi-source parallelism, and master-slave division of labor; skill exchange, with expanded participant types enabling agents and employees to participate in skill exchange on an equal footing; skill transferability, with "I want to learn" and "I can teach" announcements, platform matching, migration requests, and returned skill definitions facilitating skill transfer between agents and between agents and employees; and skill execution, with complete skill definitions and registration after review enabling the formation of executable skills after migration, allowing human-taught agents to achieve this. Attached Figure Description

[0007] Figure 1 is a flowchart of the intelligent agent clustered collaboration and skill transfer learning method of the present invention. As shown in Figure 1, multiple instances are deployed and registered in the cluster, deployed by department, project or scenario. The cluster registry contains instance identifier, instance name, department or project, and online status. Cluster discovery and division of labor negotiation are carried out. Collaboration modes are defined, including task delegation, result sharing, and parallel division of labor, corresponding to pipeline, parallel, and master-slave. The types of participants in the skill exchange platform are expanded, with staff and agent_instance, and participant_id corresponding to employee or instance identifiers. The skill transfer process involves publishing "I want to learn" and "I can teach", platform matching, migration request, return of skill definition (skill identifier, trigger, processing logic), and inclusion in the skill registry after review. The human-teacher intelligent agent is used. Employee SOPs are converted into skill definitions and "I can teach" is published. After the intelligent agent publishes "I want to learn", it matches with employees, makes a migration request, and the employee returns the skill definition. After review, an executable skill is formed.

[0008] Figure 2 is a schematic diagram of the cluster registration and collaboration mode of the present invention. As shown in Figure 2, the cluster registration form includes instance identifier, instance name, department or project, online status, endpoint, and agentType; the three collaboration modes are task delegation (A delegates to B, B returns the result, corresponding to pipeline), result sharing (multiple instances share the result or context, corresponding to parallelism), and parallel division of labor (multiple instances execute in parallel, and the result is aggregated, corresponding to master-slave); business and finance examples include intelligence → pricing → approval pipeline, parallel capture of multi-source intelligence, and master control assigning tasks to intelligence / compliance / pricing sub-agents.

[0009] Figure 3 is a flowchart of the skill transfer and human-teaching intelligent agent of the present invention. As shown in Figure 3, the intelligent agent instance or employee publishes "I want to learn" or "I can teach"; the platform matches according to ability tags or skill identifiers, using tag similarity or identifier precise matching; a skill transfer session is established; the requesting party initiates a skill transfer request to the provider; the provider returns a skill definition, including skill identifier, trigger, and processing logic, which may include input parameters, output parameters, and preconditions; the requesting party submits a skill registry, which is reviewed by the administrator or system; if the review is successful, it is included in the skill registry to form an executable skill; if the review fails, modification suggestions are returned. Human-teaching intelligent agent: The employee converts the SOP into a skill definition and publishes "I can teach", the intelligent agent publishes "I want to learn", the platform matches the employee, the intelligent agent initiates a transfer request to the employee, the employee returns the skill definition, and after review, the intelligent agent obtains an executable skill.

[0010] Figure 4 is an architecture diagram of the intelligent agent clustered collaboration and skills transfer learning system of this invention. As shown in Figure 4, the cluster registration module supports multi-instance deployment, cluster registry, cluster discovery, and task negotiation; the collaboration mode definition module defines task delegation, result sharing, and parallel task allocation, corresponding to pipeline, parallel, and master-slave; the participant type extension module extends participant_type (staff, agent_instance), participant_id, and publishes "I want to learn" and "I can teach"; the skills transfer module implements platform matching, migration requests, returns skill definitions, and includes them in the skills registry after review; the human-teaching intelligent agent module supports employee SOP conversion, publishing "I can teach", intelligent agents publishing "I want to learn", matching, migration, and review. The modules are connected sequentially according to their connection relationships. Detailed Implementation

[0011] 5.1 Terminology Definition 5.2 Example 1: Typical Scenario (Intelligence → Pricing → Approval Flow Collaboration and Human-Educational Intelligent Agent) Scenario Description: A non-ferrous metals trading company deploys multiple instances of intelligent agents for intelligence, pricing, and compliance by department, forming a cluster; the intelligence → pricing → approval pipeline adopts a task delegation model; the pricing intelligent agent instance learns basis point pricing skills from employees, realizing human-teaching intelligent agents.

[0012] Step S1: As shown in Figure 1, the cluster registration module supports deploying multiple agent instances by department. The intelligence department deploys instance-intel-001 and instance-intel-002; the pricing department deploys instance-pricing-001; and the compliance department deploys instance-compliance-001. Each instance registers with the cluster registry, including the instance identifier, instance name, department (e.g., department: "intelligence"), and online status (online: true). Instances within the cluster discover each other through the cluster registry and negotiate their division of labor: the intelligence instance is responsible for data collection, the pricing instance is responsible for pricing suggestions, and the compliance instance is responsible for approval and pre-filling. The cluster registry also includes an endpoint field, such as instance-intel-001's endpoint being cluster.example.com / intel / 001 (example address).

[0013] Step S2: As shown in Figure 2, the collaboration mode definition module adopts the task delegation mode. The intelligence instance instance-intel-001 delegates the pricing subtask to instance-pricing-001 for execution, and instance-pricing-001 returns the result to instance-intel-001; instance-intel-001 then delegates the approval pre-filling subtask to instance-compliance-001. This forms a pipeline process of intelligence generation → pricing suggestion → approval pre-filling.

[0014] Step S3: The participant type extension module expands the participant types of the skills exchange platform to "staff" and "agent_instance". Employee Zhang San's participant_type is "staff", and participant_id is "staff-001"; the pricing agent instance's participant_type is "agent_instance", and participant_id is "instance-pricing-001". Agents and employees participate in skills exchange on an equal footing.

[0015] Step S4: As shown in Figure 3, the skill migration module receives the published skills. The pricing agent instance `instance-pricing-001` publishes "I want to learn," with the ability tag "basis point price"; employee Zhang San publishes "I can teach," with the skill identifier "skill-basis-point-001." The platform matches the ability tag "basis point price" with the skill identifier to establish a migration session. `instance-pricing-001` initiates a skill migration request to Zhang San, who returns the skill definition: `skillId` is `skill-basis-point-001`, `trigger` is the event trigger (when a new order arrives), and `logic` is the call to the basis API (Application Programming Interface) to calculate the point price suggestion. `instance-pricing-001` submits the skill definition to the skill registry. After administrator approval, it is included in the skill registry, forming an executable skill.

[0016] Step S5: As shown in Figure 4, the Human Education Intelligent Agent module supports employees in converting enterprise SOPs into skill definitions. Zhang San extracts the skill identifier, triggering conditions, and execution steps from the "Basis Price Point Operation Instructions" to form a skill definition containing skill identifiers, triggers, and processing logic, and publishes "I can teach". After instance-pricing-001 publishes "I want to learn", the platform matches Zhang San, initiates a migration request, Zhang San returns the skill definition, and after review, instance-pricing-001 obtains the executable skill, realizing the Human Education Intelligent Agent.

[0017] Beneficial effects: formation of multi-instance clusters; task delegation supports pipeline; expansion of participant types enables agents to participate in skill exchange; complete skill transfer process; implementation of human-centered intelligent agents.

[0018] 5.3 Example 2: Boundary Scenarios (Parallel Multi-Source Intelligence, Master-Slave Division of Labor, Differentiation by Participant_type) Trigger type) Scenario Description: The orchestration engine needs to crawl intelligence from multiple sources in parallel, adopting a result-sharing mode; the master control agent assigns sub-tasks to intelligence, compliance, and pricing sub-agents, adopting a parallel division of labor mode; the skill exchange platform distinguishes employees and agents by participant_type; skill definition triggers include three types: event, time, and condition.

[0019] Step S1: The cluster registration module supports deployment by project. The project "Copper Trading" deploys instance-intel-cu-001, instance-intel-cu-002, instance-pricing-cu-001, and instance-compliance-cu-001. The cluster registry also contains an agentType field; the agentType for instance-intel-cu-001 is "intelligence," used to filter instances participating in collaboration by type.

[0020] Step S2: The collaborative mode definition module adopts a result-sharing mode and a parallel division of labor mode. Multi-source intelligence parallel capture: instance-intel-cu-001 and instance-intel-cu-002 capture different intelligence sources in parallel, sharing the capture results or context. Master-slave division of labor: The master intelligent agent assigns sub-tasks to the intelligence, compliance, and pricing sub-intelligent agents; the sub-intelligent agents execute these tasks in parallel and then aggregate the results.

[0021] Step S3: The participant type extension module uses the participant_type field. The participant_type value can be either staff or agent_instance; when participant_type is staff, participant_id corresponds to the employee identifier; when participant_type is agent_instance, participant_id corresponds to the instance identifier.

[0022] Step S4: In the skills transfer module, the skill definition also includes input parameters, output parameters, and preconditions. Triggers include event triggers (when a specific event occurs), time triggers (according to a planned time), and condition triggers (when specified conditions are met). Review is performed by the administrator or the system; once approved, the skill is added to the skills registry; if the review fails, modification suggestions are returned to the requesting party. When the platform matches by ability tags or skill identifiers, it uses tag similarity or precise identifier matching; upon successful matching, a skills transfer session is established between the requesting party and the provider.

[0023] Boundary Description: Without multi-instance deployment and a cluster registry, a cluster cannot be formed. Without at least one of task delegation, result sharing, or parallel division of labor, pipelines, parallel or master-slave collaboration cannot be formed. If the participant type is not expanded to staff and agent instances, agents cannot participate in skill exchange. If the skill definition lacks skill identifiers, triggers, and processing logic, an executable skill cannot be formed. Without review and inclusion in the skill registry, skill quality cannot be guaranteed. Therefore, multi-instance deployment, a cluster registry, three collaboration modes, participant type expansion, complete skill definitions, and registration after review are essential technical features of this invention.

[0024] 5.4 Example 3: Comparison with Existing Technology Comparison: In existing technologies, intelligent agents are mostly single instances; there is no task delegation, result sharing, or parallel division of labor between intelligent agents; skill exchange platforms only support employees; intelligent agents cannot publish "I want to learn" or "I can teach"; skill transfer content is incomplete and there is no review mechanism.

[0025] Comparison results: Essential technical features explained: Without multi-instance deployment and a cluster registry, a cluster cannot be formed. Without the three collaboration modes, pipeline, parallel, or master-slave collaboration cannot be formed. If the participant type does not include staff and agent instances, agents cannot participate in skill exchange. If skill migration does not include publishing "I want to learn" or "I can teach," platform matching, migration requests, and returning skill definitions, skills cannot be migrated. If the skill definition does not include skill identifiers, triggers, and processing logic, executable skills cannot be formed. If skills are included in the skill registry without review, skill quality cannot be guaranteed. Therefore, multi-instance deployment, a cluster registry, three collaboration modes, expanded participant types, complete skill definitions, and registration after review are essential technical features of this invention.

[0026] 5.5 Explanation of differences from existing technologies This invention employs a technical solution involving multi-instance deployment and cluster registration, three collaborative modes, expanded participant types for the skills exchange platform, a skills migration process, and a human-teaching intelligent agent. All five elements are indispensable: without a multi-instance cluster, a cluster cannot be formed; without the three collaborative modes, pipelines, parallel processes, or master-slave collaboration cannot be established; without expanded participant types, intelligent agents cannot participate in skills exchange; without a skills migration process, skills cannot be transferred; and without a human-teaching intelligent agent, employee SOPs cannot be transformed into executable skills. No prior art has been found to integrate all five elements into a complete solution. This invention has creative extensions in the following aspects: (1) It extends the agent from a single instance to a multi-instance cluster, supporting deployment by department, project or scenario. The cluster registry supports instance discovery and division of labor negotiation, which is different from single instance deployment; (2) It defines three collaborative modes: task delegation, result sharing and parallel division of labor, corresponding to pipeline, parallel and master-slave, supporting business and financial pipelines such as intelligence → pricing → approval and multi-source parallel and master-slave division of labor; (3) It expands the types of participants in the skill exchange platform from only employees to staff and agent instances, so that agents and employees can participate in skill exchange on an equal footing; (4) The skill migration process includes the release of "I want to learn" and "I can teach", platform matching, migration request, return of skill definition containing skill identifier, trigger, processing logic, and inclusion in the skill registry after review; (5) The human-teaching agent realizes that employees convert SOP into skill definition and agents learn from employees to form executable skills. Therefore, this invention does not only implement the general agent collaboration concept, but is a technical extension and system implementation of clustered collaboration and skill migration learning for business and financial scenarios.

[0027] The difference from existing technologies: For a long time, intelligent agents in this field have mostly been single instances, skill exchange platforms only support employees, and intelligent agents cannot publish "I want to learn" or "I can teach," a model that has become industry practice. Those skilled in the art, based on existing technologies, find it difficult to readily conceive of a complete technical solution integrating multi-instance clusters, three collaboration modes, expanded participant types (staff and agent_instance), skill migration processes, and human-teaching intelligent agents. General intelligent agent collaboration solutions are mostly single-instance or simple multi-instance, lacking participant expansion for skill exchange platforms and the design of human-teaching intelligent agents. This invention breaks the old model of "single instance, employee-exclusive skill exchange," proposing a complete technical route of clustered collaboration, intelligent agent participation in skill exchange, and human-teaching intelligent agents. This route differs from general methods and has significant technical effects for business and finance scenarios.

[0028] The above description is merely a preferred embodiment of the present invention and does not limit the scope of patent protection of the present invention. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention. Equivalent substitutions, modifications, or variations made by those skilled in the art based on the technical solutions of the present invention, without inventive effort, should all fall within the scope of protection of the present invention.

Claims

1. A method for intelligent agent clustered collaboration and skill transfer learning, characterized in that, Includes the following steps: S1. Multi-instance deployment and cluster registration: Supports the deployment of multiple agent instances by department, project or scenario; registers each agent instance to the cluster registry; the cluster registry contains at least the instance identifier, instance name, department or project to which it belongs, and online status; agent instances within the cluster discover each other and negotiate the division of labor through the cluster registry; S2. Define Collaboration Modes: Three collaboration modes are adopted to support cooperation between agent instances; the three collaboration modes include: task delegation mode, where agent instance A delegates a subtask to agent instance B for execution, and B returns the result to A; result sharing mode, where multiple agent instances share the execution result or context; and parallel division of labor mode, where multiple agent instances execute different subtasks in parallel, and the results are converged; each collaboration mode corresponds to at least one of pipeline, parallel, and master-slave. S3. Expand the participant types of the skills exchange platform: Expand the participant types of the skills exchange platform to two types: staff and agent instances; staff represents employees, and participant_id corresponds to the employee identifier; agent instances represent agent instances, and participant_id corresponds to the instance identifier; agent instances participate in skills exchange on an equal footing with employees and can publish "I want to learn" or "I can teach"; S4. Skill Transfer Process: An agent instance or employee publishes "I want to learn" or "I can teach" to the skill exchange platform; the platform matches "I want to learn" and "I can teach" according to ability tags or skill identifiers; the requesting party, i.e., the agent instance or employee that published "I want to learn", initiates a skill transfer request to the provider, i.e., the agent instance or employee that published "I can teach"; the provider returns a skill definition; the skill definition includes at least a skill identifier, trigger, and processing logic; the requesting party submits the skill definition to the skill registry, which, after review, is included in the skill registry, forming an executable skill; S5. Human-Teaching Intelligent Agent: Employees convert enterprise SOPs or operating instructions into skill definitions and publish "I can teach" through the skill exchange platform; after the intelligent agent instance publishes "I want to learn", the platform matches the employee; the intelligent agent instance initiates a skill migration request to the employee, and the employee returns the skill definition; the intelligent agent instance submits the skill definition to the skill registry, and after review, it becomes an executable skill.

2. The method for intelligent agent clustered collaboration and skill transfer learning according to claim 1, characterized in that, The cluster registry also contains an endpoint field, which represents the URI of the task receiving interface of the agent instance.

3. The method for intelligent agent clustered collaboration and skill transfer learning according to claim 1, characterized in that, The cluster registry also includes an agentType field, which represents the agent type and is used to filter instances participating in collaboration by type.

4. The method for intelligent agent clustered collaboration and skill transfer learning according to claim 1, characterized in that, The task delegation mode corresponds to a pipeline-style collaboration, and the business and finance examples include a pipeline process of intelligence generation → pricing suggestions → approval of pre-filled information.

5. The method for intelligent agent clustered collaboration and skill transfer learning according to claim 1, characterized in that, The result sharing mode corresponds to parallel collaboration, and business examples include parallel crawling of multi-source intelligence and sharing of crawling results or context among multiple instances.

6. The method for intelligent agent clustered collaboration and skill transfer learning according to claim 1, characterized in that, The parallel division of labor mode corresponds to master-slave collaboration. An example of this is that the master intelligent agent assigns sub-tasks to intelligence, compliance, and pricing sub-intelligent agents, and the results are aggregated after the sub-intelligent agents execute them in parallel.

7. The method for intelligent agent clustered collaboration and skill transfer learning according to claim 1, characterized in that, The skills exchange platform uses the participant_type field to distinguish participant types. The participant_type value can be either staff or agent_instance. When the participant_type is staff, the participant_id corresponds to the employee identifier. When the participant_type is agent_instance, the participant_id corresponds to the instance identifier.

8. The method for intelligent agent clustered collaboration and skill transfer learning according to claim 1, characterized in that, The skill definition also includes at least one of input parameters, output parameters, and preconditions; wherein the input parameters represent the input required for skill execution, the output parameters represent the result of skill execution, and the preconditions represent the prerequisites for skill execution.

9. The method for intelligent agent clustered collaboration and skill transfer learning according to claim 1, characterized in that, The trigger includes at least one of event triggering, time triggering, and condition triggering; wherein event triggering means that it is triggered when a specific event occurs, time triggering means that it is triggered according to a planned time, and condition triggering means that it is triggered when a specified condition is met.

10. The method for intelligent agent clustered collaboration and skill transfer learning according to claim 1, characterized in that, The review is performed by the administrator or the system; once the review is passed, the skill definition is added to the skill registry; if the review fails, modification suggestions are returned to the requesting party.

11. The method for intelligent agent clustered collaboration and skill transfer learning according to claim 1, characterized in that, In the aforementioned human-teaching intelligent agent, when employees convert enterprise SOPs or operating instructions into skill definitions, at least the skill identifier, triggering conditions, and execution steps are extracted to form a skill definition containing skill identifiers, triggers, and processing logic.

12. The method for intelligent agent clustered collaboration and skill transfer learning according to claim 1, characterized in that, When the platform matches based on capability tags or skill identifiers, it uses tag similarity or identifier precision matching; after a successful match, a skill transfer session is established between the requester and the provider.

13. A system for intelligent agent cluster collaboration and skills transfer learning, characterized in that, include: The cluster registration module is used to support the deployment of multiple agent instances by department, project or scenario, and to register each agent instance to the cluster registry. The cluster registry contains at least an instance identifier, instance name, department or project to which it belongs, and online status; intelligent agent instances within the cluster discover each other and negotiate their division of labor through the cluster registry. A collaboration mode definition module, connected to the cluster registration module, is used to support collaboration between agent instances using three collaboration modes: task delegation mode, where agent instance A delegates a subtask to agent instance B for execution, and B returns the result to A; result sharing mode, where multiple agent instances share the execution result or context; and parallel division of labor mode, where multiple agent instances execute different subtasks in parallel, and the results are converged; each collaboration mode corresponds to at least one of pipeline, parallel, and master-slave. The participant type extension module, connected to the collaboration mode definition module, is used to extend the participant type of the skill exchange platform to two types: staff and agent instance. Staff represents employees, and participant_id corresponds to the employee identifier. Agent instance represents an agent instance, and participant_id corresponds to the instance identifier. Agent instances participate in skill exchange on an equal footing with employees and can publish "I want to learn" or "I can teach". The skills transfer module, connected to the participant type extension module, is used to receive "I want to learn" or "I can teach" requests published by agent instances or employees to the skills exchange platform; the platform matches "I want to learn" and "I can teach" according to ability tags or skill identifiers; the requesting party, i.e., the agent instance or employee that published "I want to learn", initiates a skills transfer request to the provider, i.e., the agent instance or employee that published "I can teach"; the provider returns a skill definition; the skill definition includes at least a skill identifier, a trigger, and processing logic; the requesting party submits the skill definition to the skills registry, which, after review, is included in the skills registry to form an executable skill; The Human Learning Agent Module, connected to the Skill Transfer Module, supports employees in converting enterprise SOPs or operating instructions into skill definitions and publishing "I can teach" through the skill exchange platform. After the agent instance publishes "I want to learn," the platform matches the employee. The agent instance initiates a skill transfer request to the employee, who returns the skill definition. The agent instance submits the skill definition to the skill registry, which, after review, forms an executable skill.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent agent clustered collaboration and skill transfer learning method as described in claim 1.

Images