60 results about "Planning agent" patented technology

The invention provides an industrial safety risk monitoring and early warning system and method for an autonomous planning intelligent agent, relates to the technical field of industrial safety production, and solves various limitation problems still existing in an existing safety risk monitoring scheme. In the system, an environment sensing layer is used for collecting multi-source sensing data in an industrial production environment in real time; the edge calculation layer is used for carrying out primary processing on the multi-source sensing data and executing data compression and key information extraction so as to generate key analysis data; the cloud analysis layer is used for performing deep intelligent analysis on the key analysis data, executing security risk assessment and risk diffusion prediction, and dynamically generating and adjusting an early warning decision through an autonomous planning agent; and the action execution layer is used for executing a corresponding safety action instruction according to the early warning decision and feeding back an execution result to the cloud analysis layer to realize system optimization. According to the invention, the risk early warning accuracy and the emergency disposal timeliness can be improved finally.

The invention discloses a property operation report generation method and device based on multiple agents and a related medium, and the method comprises the steps: calling an intention recognition agent to carry out intention analysis, and then calling a task planning agent to carry out path construction and scheme solving; determining an execution sequence of tasks in combination with the task planning data to obtain task processing data; calling a data collection agent to carry out data collection on the task processing data to obtain primary processing data; and calling a code calculation agent to carry out code generation and model operation on the primary processing data, calling an outline making agent to carry out outline arrangement and report generation processing, calling a report writing agent to complete report packaging, and outputting a property operation report. According to the invention, the structured analysis result obtained through calculation is subjected to outline arrangement and report generation processing, and the property operation report is packaged and output, so that manual report arrangement is not needed, the process is simplified, and the report output efficiency is improved.

The invention discloses a geological disaster emergency response path planning method based on reinforcement learning. The geological disaster emergency response path planning method comprises the following steps: S1, fusing and processing multi-source data; s2, constructing a four-dimensional space-time model based on the fused data; s3, multi-agent collaborative decision making is carried out, and a mixed agent system based on a Safe-PPO improved algorithm is deployed; a multi-objective optimization algorithm is adopted to balance the rescue efficiency and safety; deciding and outputting a leading path and a plurality of alternative paths; s4, performing conflict detection on the action of the path planning agent; s5, carrying out preferential selection on the paths subjected to conflict detection and global optimization; s6, issuing the path to a terminal for rescuing the action personnel; s7, dynamically monitoring the environment; s8, judging whether a path needs to be re-planned or not according to environment dynamic monitoring data; and S9, completing the task. According to the method, through reinforcement learning and multi-agent collaborative architecture, the risk penalty term is embedded in the reward function through the improved Safe-PPO algorithm, and the accuracy of path planning is improved.

The invention discloses a natural language driven CAD (Computer Aided Design) modeling method based on multi-agent collaboration. According to the method, a collaborative mechanism of a structure planning agent, a parameter completion agent and an execution verification agent is introduced, and the method comprises the steps that firstly, a structure planning agent analyzes natural language description and generates an abstract syntax tree structure meeting a CAD modeling rule; then, on the premise that the structure of the abstract syntax tree is kept unchanged, a parameter completion agent predicts and completes all symbol and numerical parameters, and a complete CAD modeling instruction sequence is generated; and finally, the execution verification agent performs structure and parameter dual verification on the generation instruction by using the geometric kernel, and feeds back and corrects if necessary. According to the method, a strategy of combining field adaptive full-parameter pre-training and task specific fine tuning is adopted in a training stage, and a structure-parameter decoupling modeling and closed loop verification mechanism is introduced, so that the structure accuracy, the parameter matching property and the geometric performability of a generated result are effectively improved.

The invention discloses a multi-agent resource scheduling method based on a large language model in an industrial scene, and the method comprises the following steps: 1, setting initial scheduling parameters and user input of a multi-agent scheduling system; 2, the planning agent performs intention recognition, feature value extraction and coding; and 3, the scheduling agent receives the scheduling coding information, selects a scheduling algorithm tool and generates a scheduling scheme. And 4, the fault detection agent receives the initial scheduling scheme and receives key parameters of the working agent. And 5, the fault detection agent judges that a fault exists and optimization exists, if yes, the scheduling agent is fed back, and the step 3 is executed again, and if not, an optimal scheduling result is output. According to the method, a multi-agent architecture driven by a large language model is adopted, and a dynamic algorithm tool selection mechanism and a prompt project are introduced. Minimization of task total time consumption and maximization of fault detection accuracy are taken as optimization objectives, and adaptivity of industrial scene resource scheduling and system reliability are remarkably improved.

The invention discloses a manufacturing process optimization method and system based on multi-agent autonomous collaboration, and the method is applied to a central coordinator, and comprises the steps: transmitting a received production demand to a task planning agent for the task planning agent to carry out the analysis and decomposition, and forming task information; receiving task information, wherein the task information comprises sub-tasks after the task planning intelligent agent analyzes and decomposes the production demand; the sub-tasks are sent to a plurality of intelligent agents, so that the intelligent agents perform bidding according to self capability evaluation functions and cost estimation to generate bidding information; and bidding information is received, the bidding information comprises a self-ability evaluation function and cost estimation of each agent for each sub-task, the sub-tasks are allocated to the multiple agents according to a total execution cost minimization principle, and a production plan is generated and executed. According to the method provided by the invention, planning and execution of the production task can be autonomously completed in a complex manufacturing scene, and the production efficiency and the product quality are improved.

The embodiment of the invention provides a task processing method, device and equipment for a multi-agent system. According to the scheme, the system comprises a planning agent, an execution agent, a context injection module and a memory storage module; wherein when a planning agent needs to perform task planning for a user request, a context acquisition request is sent to the context injection module, and the context acquisition request comprises task description information of the user request; afterwards, the context injection module responds to the context acquisition request and acquires a planning context related to the task planning from a memory storage module based on the task description information, and the memory storage module stores historical data generated when each agent in the multi-agent system executes a historical task; afterwards, the planning agent receives the planning context, and generates a calling instruction for calling the execution agent based on the task description information and the planning context;

The invention discloses a method for analyzing a specific economic crime task based on large model autonomous planning, which relates to the technical field of economic crime analysis, and comprises the following steps: S1, a planning agent receives a general task target of specific economic crime analysis, retrieves economic crime feature analysis technical and tactical methods stored in a knowledge base based on the general task target, and sends the economic crime feature analysis technical and tactical methods to the planning agent; generating a key milestone plan, and submitting the key milestone plan to a scheduling agent after manual rechecking and correction; according to the method, autonomous planning and step-by-step execution of an economic crime analysis process are realized by constructing an execution agent cluster and combining with a ReAct design mode of a scheduling agent, manual intervention is reduced, and the economic crime analysis process is completed by enabling the agents to learn expert technical and tactical methods and execute tasks according to professional work division and combining report compliance verification and an error repair mechanism. According to the method, subjectivity and errors of manual analysis are avoided, the context length of the main process is controllable based on modular design of the agent cluster, expandability and adaptability are achieved, and different types of specific economic crime analysis tasks can be coped with.

The invention discloses a data enhancement method and device based on multiple agents, and the method comprises the steps: a planning agent obtains task description and data description of sample data, and the data description comprises data distribution statistics and data feature details; the planning agent generates an initial zero sample thinking chain according to data distribution statistics and data feature details of the sample data, and the zero sample thinking chain comprises an agent type and an execution strategy called by each execution level; planning an execution strategy of the agent based on a zero sample thinking chain, determining an execution level of calling an enhancement strategy agent, and generating a new sample; and the planning agent evaluates the enhancement effect of the new sample on the downstream deep learning model, if the enhancement effect does not reach a set condition, the execution hierarchy sequence of the zero sample thinking chain and the data enhancement step in the execution hierarchy are adjusted, and the step of determining the execution hierarchy of the agent for calling the enhancement strategy based on the zero sample thinking chain is returned to the step of determining the execution hierarchy of the agent for calling the enhancement strategy based on the zero sample thinking chain. The efficiency of the deep learning model is improved.

The invention relates to the technical field of artificial intelligence, and discloses a research report generation method, device and equipment and a storage medium, and the method comprises the steps: responding to an input research type subject, carrying out research step planning through a planning agent based on the research type subject, and obtaining a multi-step research plan with a logic progressive relationship, gradually searching in a search engine through a research agent based on the multi-step research plan to obtain research related information corresponding to each research step, and generating a research report corresponding to the research type subject according to the research related information corresponding to each research step; due to the fact that the multi-step research plan with the logic progressive relation is generated through cooperative work of the multiple agents and step-by-step searching is carried out in the search engine to generate the research report, intelligent planning of research report generation is achieved, and then the research efficiency and the knowledge integration capacity are improved.

The invention belongs to the technical field of artificial intelligence and data analysis, and discloses a multi-agent data automatic analysis system based on a planning execution framework, which comprises a front-end interaction interface, a data abstract agent, a planning agent, a code execution agent, a report generation agent and an anti-review agent. The method has the beneficial effects that the intelligent automatic execution of the whole data analysis process is realized, and the analysis efficiency and the result reliability are remarkably improved; through multi-agent specialized division of labor and a'planning-execution 'framework, the capability limitation of a single agent in processing complex tasks is solved, and high-availability and high-quality data analysis service is provided for non-professional users.

The invention provides a multi-agent collaborative information processing method and device, equipment and a medium, relates to the technical field of information and communication, and is suitable for the field of financial science and technology and the field of medical health. The method comprises the following steps: the terminal responds to acquired target operation information, and performs screenshot on an original interface to obtain an original interface screenshot; sending the target operation information to a cloud planning agent to obtain a target planning scheme; performing action decision on the target planning scheme and the original interface screenshot through an edge execution agent to obtain a target planning action sequence; executing the first target planning action on the original interface to obtain a first interface screenshot; and performing interface verification on the first interface screenshot through the edge observation agent, and then selecting to update the original interface or update the first interface. According to the method, the three agents are uniformly coordinated to form a continuous self-adaptive, self-error correction and self-optimization multi-agent collaborative architecture, so that the information processing accuracy is improved.

The invention discloses a multi-agent collaborative urban community space planning method and device based on large model driving. The method comprises the following steps: converting original urban space data into a structured community space distribution diagram; a planning agent is constructed, the community space distribution diagram is understood and analyzed by using the planning agent, and an initial city planning scheme is output; dividing the initial city planning scheme into a plurality of community areas, and generating local optimization suggestions by a plurality of community planning reviewer agents; and fusing the community optimization suggestions by a high-level planning agent, carrying out global reasoning and coordination, and outputting a final urban space planning scheme. Through cooperative work of multiple types of agents, the intelligent level, cooperative efficiency and service fairness of urban space planning are improved, and the problems of lack of multivariate value expression in traditional generative planning, high labor cost and slow response in participative planning and the like are solved.

The application discloses a planning calibration method and system for an information analysis intelligent assistant, and comprises the following steps: in each planning round, a candidate strategy set is generated by a planning agent, the candidate strategy set contains an information retrieval target, a retrieval keyword combination and an information processing flow; the candidate strategy set is evaluated by multiple agents with heterogeneous private memories, a strategy selection module based on the agents is constructed for information consistency, and a strategy with the most stable evaluation result is selected as an optimal strategy of the round; a difference between a self-selected strategy of the planning agent and the selected optimal strategy of the round is recorded, a calibration constraint module guided by consistency is constructed, and the difference is converted into a cognitive calibration constraint and stored in the memory of the planning agent; strategy selection optimization is realized in single-round planning by relying on the strategy selection module based on information consistency; and the cognitive calibration constraint generated by the calibration constraint module guided by consistency is combined, so that the constraint is transmitted between different rounds and gradually accumulated, and a constraint is formed on strategy generation of the planning agent; and the trained agent is applied to an open domain webpage retrieval task to generate accurate retrieval results.

The invention provides a multi-agent data query method and device, electronic equipment and a storage medium, and relates to the technical field of data processing, by constructing metadata representation used for describing database table structure information, a target query instruction can be generated based on the metadata representation and a data query task input by a user in a natural language, and the target query instruction is sent to the user. A multi-agent cooperation system comprising a planning agent and an execution agent is configured, and task planning generation of the planning agent and query operation of the execution agent are iteratively executed until the planning agent judges that a task is completed; and after the task is completed, the planning agent carries out content induction, format adjustment and visual rendering on the query result and outputs a final result. The problems that in the prior art, flexibility and expansibility are insufficient due to dependence on a single agent or a fixed workflow, result processing only stays in original data, and systematic induction and visual presentation are lacked can be solved.

The application relates to the field of software engineering and discloses a natural language driven GUI test rehearsal and test script generation method based on an LLM, which comprises the following steps: S1: test operation design: a test personnel and a planning agent realize human-machine cyclic interaction, and convert a high-level test case described in natural language into a low-level and standardized test operation description; wherein the planning agent adopts a hybrid architecture based on retrieval enhancement generation, the test personnel reviews, adjusts and feeds back the test operation description, and the feedback information is stored in a context memory module of the planning agent as prior knowledge; S2: test automation rehearsal: an operation agent takes the test operation description generated in S1 as input. Through human-machine cyclic interaction between the planning agent and the test personnel, automatic conversion of a natural language test case into a standardized test operation is realized, so that a test personnel without a programming background can also participate in automatic test design.

The invention relates to an energy consumption analysis and energy-saving report generation method and device, and belongs to the technical field of energy consumption analysis and energy-saving report generation. The method comprises the steps that a planning agent analyzes user requirements and formulates an analysis process; the time and space analysis agents analyze the energy consumption data according to the analysis process to obtain a preliminary analysis result; the preliminary analysis results are fed back to the planning agent, and the planning agent combines the preliminary analysis results and submits the results to the questioning agent; the questioning agent questions the preliminary analysis result in combination with the content concerned by the user and the common pain point of energy consumption and feeds back the questions to the time and space analysis agent for enhanced analysis to obtain enhanced analysis results, and the planning agent combines the enhanced analysis results and submits the enhanced analysis results to the questioning agent; the questioning agent questions the enhanced analysis result until there is no problem, and the planning agent submits the final analysis result to the summarizing agent; and the summarizing agent generates an energy-saving report according to the final analysis result and the typical case data.

The application discloses a kind of production accident briefing generation method, system, equipment and medium based on agent, belong to industrial safety technical field, the method is: according to the multi-modal original data obtained, the accident graph of production accident is constructed;Search fragment recall is carried out in the preset historical accident briefing library, simultaneously according to the preset knowledge graph, graph reasoning fragment recall is carried out, the historical case fragment and the graph reasoning fragment are merged, obtain candidate fragment set;Through the preset planning agent and the accident graph, generate briefing outline, through the preset fact extraction agent and the candidate fragment set, additional fact data is obtained to the briefing outline Data outline, and then through the preset writing agent and the data outline, text generation is carried out, obtain first accident briefing, therefore, by implementing the application, intelligent production accident briefing can be automatically generated, to provide immediate accurate decision basis for safety supervision personnel.

The invention discloses a method and a system for realizing intelligent control of a tool body based on a multi-agent architecture, and belongs to the technical field of artificial intelligence and robots. The method aims to solve the problems of single task processing capability, poor robustness and difficulty in expansion of the existing single-body intelligent model. The core thought of the method is to introduce a multi-agent collaboration architecture, decouple'high-level task planning 'and'bottom-level skill execution', and decompose complex control tasks to specialized agents with different functions. The method comprises the following steps: receiving a high-order task instruction; decomposing the high-order task instruction into a plurality of subtasks by a planning agent; the planning agent assigns at least one corresponding execution agent to the subtask from a set comprising a plurality of specialized execution agents according to the attribute of each subtask; and the assigned execution agent cooperatively executes the allocated sub-task so as to complete the high-order task. According to the invention, the modularization, the expandability and the task execution robustness of the intelligent system are obviously improved, and the method is suitable for various intelligent scenes such as industrial manufacturing and home service.

There is provided a method comprising: extracting a performance indicator(s) from a software development environment, obtaining context of the software development environment, in response to the performance indicator(s) not meeting a corresponding guideline(s): automatically generating, by a planning agent, a plan for improving the performance indicator(s) based on the context, automatically selecting an implementation agent(s) based on the plan, generating a prompt for instructing the implementation agent(s) for automatic generation for improvement of the performance indicator(s) according to the plan, the prompt includes the context, feeding the prompt into the implementation agent(s), obtaining data elements generated by the implementation agent(s) in response to the prompt, and applying the data elements to the software development environment to generate an adapted software development environment, wherein the adapted software development environment improves software development efficiency by narrowing a gap with or meeting the corresponding guideline(s), and / or changing a trend of the performance indicator(s).

The application relates to the technical field of financial due diligence, and particularly discloses a financial due diligence multi-agent autonomous planning system, a financial due diligence multi-agent autonomous planning method, a financial due diligence multi-agent autonomous planning device and a financial due diligence multi-agent autonomous planning medium. The system comprises a master agent module, a planning agent module and an execution agent module. The master agent module is used for receiving user queries, performing task intention recognition and target information extraction. The planning agent module is used for generating a due diligence planning path according to the target information. The execution agent module is used for calling an associated specified query tool according to the due diligence planning path, executing a query task, and returning an execution result to the master agent module. The master agent module is also used for integrating the execution result returned by the execution agent module, and outputting a structured financial due diligence report according to a preset due diligence report template output structure. The system significantly improves the analysis depth, execution efficiency, conclusion reliability and business fitting degree of financial due diligence through multi-agent cooperation and automatic planning and execution.

The invention discloses a multi-step vulnerability utilization chain automatic generation method and device based on a knowledge graph, and belongs to the technical field of combination of cyberspace security and artificial intelligence. The implementation method comprises the following steps: 1, constructing a vulnerability knowledge graph; 2, performing full-text indexing and graph traversal on the vulnerability knowledge graph to obtain a structured context; 3, the Qwen3coder model constructs a penetration test spanning tree in the form of a tree structure by using the dependency relationship between penetration steps and hierarchical nodes, and obtains the current penetration test spanning tree state; 4, executing the analysis agent, the planning agent, the command generation agent, the execution agent and the evaluation agent in sequence in a circulation mode, and generating a complete vulnerability utilization chain plan; and 5, generating a penetration test report for defense by utilizing a Qwen3coder model based on vulnerability utilization chain planning. Compared with the prior art, in a complex network environment, automatic planning of a multi-step vulnerability utilization chain is achieved, and then vulnerability utilization defense is completed.

The invention discloses an intelligent advertisement putting service optimization system and a putting platform. The intelligent advertisement putting service optimization system comprises a Plan module, a Test module, an Optimization module, a Regeneration module and a Scale module, wherein the Plan module, the Test module, the Optimization module, the Regeneration module and the Scale module are connected in sequence and are subjected to loop iteration. The system adopts a multi-agent collaboration architecture, a planning agent, a test agent, an optimization agent, a regeneration agent and an expansion agent corresponding to each module are configured, and each agent shares information and makes a collaborative decision through a data flow pipeline to form a closed-loop iterative optimization mechanism. According to the system, intelligent collaborative management of the whole advertisement putting process is achieved, the problems of strategy making, effect verification, budget optimization, creativity regeneration, scale expansion and the like in advertisement putting are efficiently solved through module circulation and intelligent agent cooperation, the putting efficiency and effect are remarkably improved, and the full-link advertisement putting requirements of small and medium-sized enterprises are met.

The invention provides a game making system based on multi-Agent cooperation. The system comprises a planning Agent, a planning Agent, a code Agent and a scheduling Agent, the planning Agent is used for analyzing a natural language requirement input by a user based on a large language model LLM and generating a game planning book; the planning Agent is used for understanding the game planning book generated by the planning Agent, performing task decomposition on the game planning book and generating a game making plan, and the game making plan comprises a game resource generation plan and a code generation plan; the code Agent is used for generating codes based on the game planner and the code generation plan, generating game program framework codes firstly and then generating various functional codes step by step; and the scheduling Agent is used for monitoring the task state of each Agent, dynamically adjusting the execution sequence of each Agent and solving conflicts. The system can rapidly generate a game prototype through natural language description, improves the development efficiency of a game system, reduces the manual cooperation cost through an automatic assembly line, and shortens the game development period.

The invention discloses a plant protection unmanned aerial vehicle pesticide application path planning method and system based on deep learning, and relates to the technical field of smart agriculture, and the method comprises the steps: constructing meta-training task distribution of a virtual farmland scene in an offline stage, carrying out the architecture design and parameter optimization through a meta-learning framework, and obtaining a path planning agent; the architecture of the path planning agent is configured to extract time sequence invariant features from environment observation, generate a composite action instruction, and train the path planning agent by adopting a meta-reinforcement learning algorithm; deploying the trained path planning intelligent agent to the plant protection unmanned aerial vehicle, executing initial exploration flight on a target new farmland, and collecting a small amount of interaction data; and performing optimization and updating through feature extraction parameters of the path planning intelligent agent. According to the invention, collaborative improvement of plant protection unmanned aerial vehicle path planning in efficiency, adaptability and agricultural long-term benefits is realized.

The invention relates to the technical field of artificial intelligence, and can be applied to software code reconstruction scenes in financial and medical scenes. The invention provides a software reconstruction method, system and device and a storage medium, and the method comprises the steps: carrying out the context analysis of a target code entity through a planning agent, and generating a structured reconstruction plan based on an analysis result; generating a reconstructed code by the generation agent according to the reconstruction plan; the compilation agent performs compilation verification on the reconstructed code, and when compilation fails, compilation error information is fed back to the generation agent for iterative correction; and the test agent performs function test on the reconstructed code after successful compiling, and feeds back test failure information to the generation agent for iterative correction when the test fails. The method can be applied to financial / medical software code reconstruction scenes, and the success rate of software reconstruction is improved through multi-agent cooperation and double feedback closed loops of compiling and testing.

This invention discloses a method for generating SEO blogs based on a large language model and multi-agent collaboration. This method automatically executes the entire blog generation process by constructing a multi-agent system that coordinates a planning agent and a tool-using agent: the planning agent receives and parses user requirements, breaking down the blog generation goal into a multi-step execution plan; the tool-using agent calls external tools to execute specific tasks according to the plan; a dependency task judgment mechanism is introduced during task execution to check in real time whether the pre-dependencies of the tasks are completed, intelligently skipping tasks with failed dependencies to avoid redundant operations; after all tasks are completed, a large verification model is called to perform multi-dimensional SEO scoring on the generated content; if it is unsatisfactory, the planning agent is triggered to replan and iteratively execute based on the failure logs until a blog that meets the requirements is generated. This invention achieves end-to-end automation of SEO blog generation, significantly improving operational efficiency and content quality.

The application provides a search generation method, device and equipment based on multi-agent cooperation and a medium. The method comprises the following steps: receiving a user search demand and calling a planning agent to perform task planning to generate a tree-shaped task planning result; calling a decomposition agent to perform complex task fine-grained decomposition according to the tree-shaped task planning result to obtain corresponding planning tasks and the dependency relationship and parallel relationship between the planning tasks, calling multiple agents, and executing the corresponding planning tasks with the parallel relationship in parallel, and after the corresponding planning tasks are executed, calling an agent to execute the next planning task with the dependency relationship to obtain corresponding search results; calling a verification agent to perform multi-source information cross verification according to the corresponding search results, and generating a corresponding search report based on the verification. The application can ensure that the conclusion of the final report is based on solid and reliable facts, significantly improves the efficiency, and comprehensively guarantees the factual accuracy of the report.

The invention provides an enterprise production scheduling method and device based on multiple agents and electronic equipment, and relates to the technical field of manufacturing industry production, and the method comprises the steps: decomposing a long-term production plan through a plan agent, and generating a short-term production plan; the short-term production plan is a production plan within a preset number of days; after the short-term production plan is converted into a dispatch list of each process by using an execution agent, the dispatch list of each process is issued to a production informatization management system for execution; under the condition that the abnormal event is monitored, the abnormal event is processed based on a dynamic disturbance response mechanism, the short-term production plan is adjusted, and production scheduling is performed again based on the adjusted production plan. According to the multi-agent-based enterprise production scheduling method and device and the electronic equipment provided by the invention, by introducing a multi-agent technology and combining a multi-level treatment structure in an enterprise and actual data of each production line, intelligentization, flexibility and high efficiency of a production plan are realized.

The invention provides a game development automation method based on a multi-agent system. The game development automation method is used for providing a universal AI full-automatic auxiliary development end-to-end solution for game developers. The multi-agent system is composed of a planning agent (Planning Agent), a plurality of execution agents (Execution Agents) and a web page search agent (WebSearch Agent). The planning agent analyzes user demands, plans tasks, decomposes the tasks and allocates subtasks to the execution unit; the execution agent is provided with a well-designed professional field-based tool, and can help a large language model LLM to use a zero sample reasoning ability calling tool under the condition that pre-training and post-processing fine tuning are not performed, so that the problem of the professional field is solved; and the webpage search agent can dynamically obtain the Internet information based on the task execution requirement. The whole system is provided with a dynamic memory isolation mechanism, so that the knowledge illusion and the execution error rate of a large language model can be remarkably reduced, and the game development efficiency and the automation reliability are improved.