Test case generation and intelligent management system based on multi-agent collaboration

Abstract

The invention belongs to the technical field of artificial intelligence, and particularly relates to a test case generation and intelligent management system based on multi-agent collaboration. The system comprises an input and configuration module which is used for interacting with a user and receiving a demand document uploaded by the user and configuration information related to test case management; the agent cooperation module is used for generating and reviewing the test cases meeting the quality requirements through cooperative work of a plurality of agents on the basis of the demand document and the configuration information; and the RPA robot execution module is used for triggering automatic warehousing operation of the test cases after the test cases pass the review, and importing the test cases passing the review into the test case libraries of the corresponding items at the same time. The method has the characteristics that the manual test design and maintenance cost can be remarkably reduced while the quality and the stability of the test case can be ensured, and the overall efficiency and the delivery capability of the test work are greatly improved.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence technology, specifically relating to a test case generation and intelligent management system based on multi-agent collaboration. Background Technology

[0002] With the increasing scale, complexity, and delivery frequency of software systems, software testing has become a crucial link in ensuring software quality and system stability. Test cases, as the core carrier of software testing activities, directly affect test coverage and defect detection capabilities through their design quality. However, existing technologies still have the following significant technical shortcomings in the process of test case generation and management: (1) Test case design is highly dependent on human experience, resulting in poor quality stability. In the existing testing process, test cases are mainly manually analyzed and designed by testers based on requirements documents. Different testers have different levels of experience, comprehension ability and design habits, which makes it difficult to maintain consistency in the completeness, standardization and coverage depth of the generated test cases.

[0003] (2) The ability to automatically generate unstructured requirement documents is limited. Requirement documents are usually described in natural language, with loose structure and diverse expressions. Traditional rule-driven or template-based automation tools have difficulty accurately understanding the semantics of the requirements, thus failing to effectively generate high-quality test cases.

[0004] (3) Existing AI-based test case generation solutions lack effective quality review and feedback mechanisms. Some existing technologies attempt to introduce AI models to generate test cases, but they usually only perform one-time generation and lack automatic review, quality judgment and feedback regeneration mechanisms for the generated results, making it difficult to guarantee coverage of key scenarios, boundary conditions and high-priority test cases.

[0005] (4) Historical defects and testing experience are difficult to reuse systematically. In actual testing, a large number of discovered defects and their corresponding triggering scenarios, root causes and verification points are recorded in a scattered manner. Existing technologies cannot automatically link the above-mentioned historical defect experience to the new round of test case design process, which can easily lead to the recurrence of known defects in subsequent versions.

[0006] (5) The application capabilities of Robotic Process Automation (RPA) technology in test case management scenarios are limited. Existing RPA technology is mainly used to execute predefined fixed processes, and can complete operations such as logging into the system, data entry, and file upload. In test case management scenarios, RPA is usually only used as a data transfer and interface operation tool to import generated test cases into project management or test management systems. Its execution process lacks a collaborative mechanism with test case generation and review results, making it difficult to effectively control the import process based on the quality status, coverage, or review conclusions of the test cases.

[0007] In summary, existing technologies have not yet provided a technical solution that can introduce a multi-role AI review mechanism during the test case generation stage, dynamically supplement or optimize test cases by combining historical defect knowledge, and then collaboratively trigger the automated execution of the robot process after the test cases pass the quality review, so as to achieve end-to-end automation of test case generation, review and management.

[0008] Therefore, it is very important to design a test case generation and intelligent management system based on multi-agent collaboration that can ensure the quality and stability of test cases, significantly reduce the cost of manual test design and maintenance, and greatly improve the overall efficiency and delivery capability of testing work. Summary of the Invention

[0009] This invention aims to overcome the problems of existing test case design processes, such as reliance on human experience, low efficiency, unstable quality, and difficulty in continuous optimization. It provides a test case generation and intelligent management system based on multi-agent collaboration that can guarantee the quality and stability of test cases while significantly reducing the cost of manual test design and maintenance, and greatly improving the overall efficiency and delivery capability of testing work.

[0010] To achieve the above-mentioned objectives, the present invention adopts the following technical solution: A test case generation and intelligent management system based on multi-agent collaboration includes: The input and configuration module is used to interact with users and receive configuration information related to user-uploaded requirement documents and test case management. The intelligent agent collaboration module is used to generate and review test cases that meet quality requirements through the collaborative work of multiple intelligent agents, based on the requirements document and configuration information. The RPA robot execution module is used to trigger the automated import operation of test cases after the test cases have passed the review, and to import the reviewed test cases into the test case library of the corresponding project.

[0011] Preferably, the input and configuration module specifically includes the following processes: Receive user-uploaded requirement documents and user-inputted natural language query requests; Provide an interactive form interface for collecting configuration parameters related to test case management; The interactive form shall include at least the following fields: The target test management system includes selection fields, user account information, user password information, and the project identifier to which the test cases belong.

[0012] Preferably, the agent collaboration module includes: A test case intelligence agent is generated, which is built based on a large language model. It is used to perform semantic understanding and test point decomposition on the requirement document, and automatically generate initial test cases according to the preset structured test case template. The test case review agent, built based on a large language model, is used to automatically review the initial test cases according to preset review rules and output review conclusions. The review result judgment module is used to control the flow of the process based on the review conclusion: if the review fails, the test case generation agent is triggered to regenerate the test cases; if the review passes, the RPA robot execution module is triggered to perform automated data entry.

[0013] Preferably, the agent collaboration module further includes: A historical defect knowledge base is used to store structured historical defect data, which includes at least defect triggering scenario information; During the review process, the intelligent agent for the review test cases calls the historical defect knowledge base to compare the initial test cases with the historical defect data, and supplements or optimizes the test cases based on the comparison results.

[0014] Preferably, the process by which the review test case agent supplements or optimizes the test cases based on the comparison results is as follows: When it is detected that the initial test cases do not cover a certain defect triggering scenario in the historical defect knowledge base, supplementary test cases to cover that scenario are automatically generated. When it is detected that the initial test case has covered a defect triggering scenario in the historical defect knowledge base, the priority of the corresponding test case is automatically increased.

[0015] Preferably, the process of generating the use case agent includes the following steps: Receive the requirements document, parse the functional descriptions, business rules, input and output conditions and abnormal scenarios in the requirements document, and form a semantic representation of the function to be tested; By using preset prompts, the large language model can simulate the role of a senior test engineer; The results are generated using structured, example-based test case templates. The test case template includes at least: Test case title, preconditions, test steps, expected results, and priority fields; Generate test case data that meets the import requirements of the test management system according to the test case template, and output it in Excel format.

[0016] Preferably, the review use case agent specifically includes the following process: Receive test cases output by the test case generating agent; By using preset review prompts, the large language model is guided to simultaneously simulate the combined roles of senior test engineers and developers, and to perform multi-perspective test case reviews from the angles of test sufficiency, implementation rationality, and risk coverage. Based on a predefined review rule system, a systematic check of test case quality is performed; the review rule system includes at least: Does the test case format conform to the established specifications? Are the test cases complete, clearly described, and executable? Do the test cases cover the test scenarios for functional testing, performance testing, security testing, and boundary testing?

[0017] Preferably, the review result judgment module triggers the test case generation agent to regenerate test cases when the review fails, specifically including the following process: The review result judgment module feeds back the review comments or optimized prompts output by the review test case agent to the generating test case agent, which guides the generating test case agent to perform targeted regeneration, forming a closed-loop optimization process of "generation-review-feedback-regeneration".

[0018] Preferably, the RPA robot execution module specifically includes the following processes: Invoke Robotic Process Automation (RPA) processes; Populate the test case data into an Excel file; The RPA executor is controlled to sequentially complete the following steps: opening the target test management system webpage, entering the user account and password, selecting the target project, and calling the test case import function provided by the system. Test cases are automatically imported into the test case library of the corresponding project, enabling centralized management and maintenance of test cases.

[0019] As a preferred option, it also includes: The anomaly monitoring module is used to uniformly monitor abnormal situations during system operation, and specifically includes the following processes: Capture exceptions during the execution of each module; Perform automatic retry or abnormal termination operation based on the exception type.

[0020] Compared with the prior art, the beneficial effects of this invention are: (1) Significantly improves the standardization and consistency of test case generation: This invention constructs a test case generation agent based on a large language model and uses structured, example-based test case templates to constrain the generation process, effectively avoiding problems such as inconsistent formats and non-standard expressions caused by differences in personal experience in traditional manual design, so that the generated test cases maintain a high degree of consistency in structure, fields and expression methods, significantly improving the standardization and maintainability of test cases; (2) Improves the completeness and defect coverage of test cases: By introducing a multi-role intelligent review mechanism through the test case review agent, it simulates the roles of senior test engineers and developers. From a composite perspective, a systematic quality check is performed on test cases, and coverage assessment is conducted by combining multi-dimensional test scenarios such as functional testing, performance testing, security testing, and boundary testing. This can effectively identify problems such as missing test cases, missing test points, and unreasonable priority distribution, thereby significantly improving the ability of test cases to identify and cover potential defects; (3) Achieving continuous reuse and dynamic optimization of historical defects and testing experience: This invention uses a historical defect knowledge base to structure and store past defect data, and automatically associates historical defect triggering scenarios during the test case review stage to verify and supplement newly generated test cases, effectively preventing known defects from recurring in subsequent versions, and realizing the continuous reuse and dynamic optimization of testing experience. Automatic sedimentation, reuse and continuous optimization improve the stability of overall test quality; (4) Construct a closed-loop intelligent optimization mechanism for test case generation and review: This invention uses a review result judgment module to judge the quality of test case generation results and automatically triggers a regeneration process when the quality standards are not met, forming a closed-loop control mechanism of "generation-review-feedback-regeneration" to avoid the uncontrollable quality problem caused by one-time generation, so that the quality of test cases can be continuously improved with process iteration; (5) Realize the intelligent and automated collaboration of test case management process: This invention organically combines artificial intelligence agents and robotic process automation technology, and automatically triggers the process after the test case review is passed. RPA performs test case entry operations, making RPA no longer just a simple data transfer tool, but capable of intelligent control based on the quality status of test cases, thereby achieving end-to-end automation of test case generation, review and management; (6) Significantly improves test work efficiency and reduces labor costs: By automatically completing key links such as requirement analysis, test case generation, quality review, historical defect verification and test case entry, this invention greatly reduces the workload of manual participation in test case design and maintenance, shortens the test preparation cycle, improves test delivery efficiency, and effectively reduces labor costs while ensuring test quality, especially suitable for software development scenarios with frequent requirement changes and rapid iteration. Attached Figure Description

[0021] Figure 1This is a schematic diagram of an architecture for a test case generation and intelligent management system based on multi-agent collaboration according to the present invention. Figure 2 This is a flowchart illustrating the implementation of the test case generation and intelligent management system based on multi-agent collaboration of the present invention in a practical application. Detailed Implementation

[0022] To more clearly illustrate the embodiments of the present invention, specific implementation methods will be described below with reference to the accompanying drawings. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings and other implementation methods can be obtained based on these drawings without any creative effort.

[0023] like Figure 1 and Figure 2 As shown, this invention provides a test case generation and intelligent management system based on multi-agent collaboration, specifically including an input and configuration module, an agent collaboration module, an RPA robot execution module, and an anomaly monitoring module.

[0024] The specific functions of each module are as follows: 1. Input and Configuration Module: The input and configuration module is used to interact with the user and receive the input information required for test case generation and database entry, specifically including: Receive user-uploaded requirement documents; Receive natural language query requests from users; Provides an interactive form interface for collecting configuration parameters related to test case management.

[0025] The interactive form shall include at least the following fields: The selection field of the target test management system is used to select any one of the following systems: Cloud Effect, TAPD, ZenTao, and JIRA. User account information; User password information; The project identifier to which the test case belongs.

[0026] The input and configuration module sends the collected requirement documents and configuration information to the intelligent agent collaboration module to trigger subsequent automated processing.

[0027] 2. Agent Collaboration Module The intelligent agent collaboration module includes a test case generation intelligent agent, a test case review intelligent agent, a historical defect knowledge base, and a review result judgment. This module receives requirement documents and configuration information from the input and configuration module, and sequentially triggers the generation of the test case intelligent agent, the review of the test case intelligent agent, and the historical defect knowledge base according to a predefined process. It also controls the process branches based on the review results, realizing closed-loop control of generation -> review -> feedback, ensuring that the final output is complete, standardized, and comprehensive high-quality test cases.

[0028] 2-1. Generate use case agents The test case generation agent is built on a large language model service and is used to perform semantic understanding of requirement documents, decompose test points, and automatically generate structured test cases. Its working method includes at least the following steps: Receive the requirements document, parse the functional descriptions, business rules, input and output conditions and abnormal scenarios in the requirements document, and form a semantic representation of the function to be tested; By using preset prompts, the large language model can simulate the role of a senior test engineer; The results are generated using structured, example-based test case templates, which include at least the following: Test case title, preconditions, test steps, expected results, and priority fields; Generate test case data that meets the import requirements of the test management system according to the template, and output it in Excel format.

[0029] The above methods enable the automation and standardization of the test case design process by generating the intelligent agent module.

[0030] 2-2. Reviewing Use Case Agents The test case review agent is also built on the large language model service and is used to automate and intelligently review the generated test cases. Its functions include: Receive test cases output by the test case generating agent; By using preset review prompts, the large language model is guided to simultaneously simulate the combined roles of senior test engineers and developers, and to perform multi-perspective test case reviews from the angles of test sufficiency, implementation rationality, and risk coverage. Based on a predefined review rule system, a systematic check of test case quality is performed. The review rules include at least the following: Does the test case format conform to the established specifications? Are the test cases complete, clearly described, and executable? Are the number of P0-level test cases not less than the preset threshold? Are the number of P1-level test cases not less than the preset threshold? Do the test cases cover key test scenarios such as functional testing, performance testing, security testing, and boundary testing? Among them, P0, P1, and P2 usually represent the priority of test cases or defects. P0 test cases are used to verify core functions and critical business processes. Failure in these cases will render core system functions unavailable or prevent business operations from continuing. P1 test cases are used to verify important functions and critical scenarios. Problems in these cases will not render the system completely unavailable, but will severely impact user experience or business efficiency.

[0031] Based on the above review results, the review test case agent outputs structured review conclusions and problem labeling information, and feeds the review results back to the decision control module for subsequent test case optimization decisions and process control.

[0032] 2-3. Historical Defect Knowledge Base The historical defect knowledge base is used to store historical defects and testing experience data, and its data structure includes at least the following: Defect title; Belongs to the functional module; Defect triggering steps; Root cause classification; Fixed version information; The associated test case identifier.

[0033] During the review process, the review test case agent calls the historical defect knowledge base to check whether the newly generated test cases contain verification points for major historical defects.

[0034] When a test case is detected that does not cover a relevant defect scenario, the system automatically generates supplementary test cases; when the relevant scenario is covered, the system automatically increases the priority of the corresponding test cases, thereby enabling continuous reuse and optimization of testing experience.

[0035] 2-4. Judgment of Review Results The review result judgment is used to control the flow of the process based on the review results, and its functions include: When the test case review agent outputs a "fail" result, the control flow reverts to the agent generation module and triggers the regeneration of test cases; When the review results meet the preset quality standards, the control process enters the execution phase.

[0036] This module enables closed-loop optimization of test case generation and review.

[0037] 3. RPA Robot Execution Module The RPA robot execution module is used to trigger the automated storage and import of test cases after the test cases have passed the review. Its functions include: Invoke the Robotic Process Automation (RPA) process; Populate the test case data into an Excel file; The RPA executor is controlled to sequentially complete the following steps: opening the target test management system webpage, entering the user account and password, selecting the target project, and calling the test case import function provided by the system. Test cases are automatically imported into the test case library of the corresponding project, enabling centralized management and maintenance of test cases.

[0038] 4. Anomaly Monitoring Module The anomaly monitoring module is used to uniformly monitor abnormal situations during system operation, and its functions include: Capture exceptions during the execution of each module; Perform automatic retry or abnormal termination operation based on the exception type; Ensure the stability and reliability of the overall system process.

[0039] Based on the technical solution of this invention, the implementation process of this invention in practical applications will be illustrated through the following case scenarios. The specific application implementation scheme is as follows: refer to Figure 1 and Figure 2 This embodiment provides a system and method for automatic generation and storage of test cases based on multi-agent collaboration. Step 1: Standardization of User Interaction and Input Information After the system starts up, it first enters the user interaction phase. Users can upload the requirements document of the system to be tested through the input and configuration module, and can also enter natural language query requests to describe the test focus or test objectives.

[0040] Simultaneously, the system displays a configuration form to the user, used to collect configuration information related to test case management. This configuration information includes the target test management system type, user account information, user password information, and the project identifier to which the test case belongs. The input and configuration module performs integrity and format checks on the above information and standardizes user input.

[0041] When missing, invalid, or unrecognizable input information is detected, the input and configuration module sends the exception information to the exception monitoring module, which then performs prompting, retrying, or termination operations. Once the input information passes verification, the system proceeds to the next processing stage.

[0042] Example: Upload the "Taobao Login Function Requirements Document.doc" and enter the query "The login function needs to cover Chrome, Firefox, and Edge browsers, with a focus on compatibility test point discovery".

[0043] Step 2: Generate Test Cases The intelligent agent automatically generates test cases. After validating the user input information, the system sends the requirements document to the test case generation agent.

[0044] The generated test case agent is built based on the large language model service. First, the requirements document is semantically understood and structuredly parsed to extract the functional description, business rules, input and output conditions and exception handling logic, and then a semantic representation of the function to be tested is formed.

[0045] Subsequently, the system guides the large language model, simulating the role of a senior test engineer, through preset prompts to decompose the semantic representation into test points and generate structured test cases according to predefined test case templates. The test case templates include at least a test case title, preconditions, test steps, expected results, and priority fields, ensuring that the generated results conform to the import specifications of the test management system.

[0046] Prompt word reference: "You are a senior software testing engineer with 10 years of experience, skilled in writing comprehensive, detailed, and executable test cases. Please write test cases for me based on the following requirements."

[0047] **[Background of the Needs]** **[Test Objective]** **

Functional Requirements Description

[0048] 2. Test cases need to cover: **Functional Testing**: Normal Stream, Abnormal Stream **Performance Testing: Performance Consumption Caused by Complex Processes** **Scenario Testing:** A wide range of scenarios are available to cover the accuracy of intent recognition and step breakdown. 3. Provide at least 15 tests. Use cases, of which there are no fewer than 5 P0 level (core function) use cases.

[0049] 4. Output in Excel format. Once generated, the test case generator outputs the test case data and passes it to the test case review generator.

[0050] Step 3: Multi-perspective intelligent review of test case agents After receiving the test cases output by the test case generating agent, the review test case agent uses preset review prompts to enable the large language model to simultaneously simulate the combined roles of a senior test engineer and a developer, and review the test cases from the perspectives of test sufficiency, implementation rationality, and potential risk coverage.

[0051] Prompt word reference: "Please play the roles of both a senior test engineer and a senior development engineer, reviewing the input test cases. The goal is to ensure the test cases are standardized, complete, and comprehensive, and to use the historical defect knowledge base to identify and address any gaps and adjust priorities."

[0052] Review requirements: 1. Standardized inspection Is the use case structure complete (preconditions / steps / expected results)? Are there any issues with ambiguous wording, lack of verifiability, or unreasonable granularity? 2. Coverage Assessment Does it cover normal processes, abnormal processes, and boundary conditions? Have any high-risk scenarios been overlooked (abnormal input, state changes, permissions, concurrency, repetitive operations, etc.)? There should be no fewer than 5 P0-level (core functionality) test cases. The total number of use cases shall not be less than 15. 3. Historical Defect Correlation Analysis (Key Point) By comparing with historical defects, determine whether the current use case covers the corresponding triggering scenario. Uncovered defects: Identify missing defects and suggest additional test points. Covered defects: Identify the corresponding use cases and assess whether the coverage is adequate. 4. Priority Adjustment Suggestions For use cases related to historically high-frequency / high-severity defects, it is recommended to increase their priority. Key business processes and defect regression test cases need to be highlighted. 5. Output requirements: Review conclusion (Pass / Fail) If the request fails, regenerate the prompt word. Upon successful completion, output the complete test cases in Excel format. Please deliver the review results in a professional and actionable manner. During the review process, the intelligent agent for reviewing test cases synchronously calls the historical defect knowledge base to compare and analyze whether the newly generated test cases contain verification points targeting major historical defects.

[0053] Step 4: Enhance historical defect knowledge and optimize test cases The test case review agent calls the historical defect knowledge base. When it detects that the generated test case does not cover a certain historical defect triggering scenario, the system automatically generates the corresponding supplementary test case. When it detects that the test case has covered the relevant defect scenario, the system automatically increases the priority of the corresponding test case.

[0054] By using the above methods, historical defects and testing experience can be automatically reused and dynamically optimized in the new round of test case design, thus avoiding the recurrence of known defects.

[0055] Step 5: Review Result Judgment and Closed-Loop Control After the review is completed, the review case agent outputs a structured review conclusion and transmits the review result to the review result judgment module.

[0056] When the review result is "not passed", the review result judgment module control flow reverts to the test case generation agent, inputs the optimized prompt words from the review test case agent to the test case generation agent, triggers the regeneration of test cases, and forms a closed-loop optimization process of "generation - review - feedback - regeneration"; when the review result is passed, the test cases in Excel format are output, and the system enters the test case storage stage.

[0057] Step 6: RPA robot executes test cases and automatically imports them into the database. After the test cases pass the review, the system triggers the RPA robot execution module. The RPA robot outputs the test cases in Excel format and saves them to a local Excel file. Then, based on the target test management system type configured by the user, the system automatically opens the corresponding web interface, allows the user to enter their username and password, select the target project, and calls the system's test case import function to upload the Excel file, thus realizing the test cases being added to the database.

[0058] Step 7: Anomaly Monitoring and Process Completion During the execution of the above steps, the anomaly monitoring module performs unified monitoring of anomalies in the generation of test case agents, review of test case agents, historical defect knowledge base calls, and RPA execution. Based on the anomaly type, it performs automatic retry, rollback, or termination operations to ensure the overall stability and reliability of the system.

[0059] Once the test cases are successfully imported into the target test management system, the process described in this embodiment ends.

[0060] The innovative aspects of this invention are as follows: (1) A test case generation and review architecture for large language models based on multi-agent collaboration is proposed. This invention is the first to decouple test case generation and test case review into independent intelligent agent modules, constructing separate intelligent agents for generating test cases and reviewing test cases, and using a decision control module to uniformly schedule and control the process of the two, forming an automated test case processing architecture with multiple intelligent agents working together, which is different from the single-model, single-stage test case generation method in the prior art.

[0061] (2) Construct a test case generation mechanism for large language models based on structured template constraints This invention constrains the output format, field structure, and content organization of large language models by predefining structured, example-based test case templates, ensuring that the generated test cases strictly conform to established testing specifications. This technically solves the problem of unstable test case format and uncontrollable structure caused by the free generation of large language models.

[0062] (3) A multi-role intelligent review method based on a single large language model is proposed. In the intelligent agent for reviewing test cases, this invention enables the same large language model to simultaneously simulate two different technical roles, senior test engineer and developer, through preset prompt words. It executes test case review logic based on multiple role perspectives, realizing an automated review method that integrates cross-role cognition, which is different from traditional technical solutions that rely on a single test perspective or manual review.

[0063] (4) Introduce an automatic test case quality review mechanism based on rule constraints. This invention predefines a set of test case quality review rules, which are automatically executed by a test case review agent to perform quality verification. The rules include at least test case format standardization, content completeness, reasonableness of test case priority distribution, and test type coverage, thereby technically realizing automated and rule-based judgment of test case quality.

[0064] (5) Propose a test case review and completion mechanism driven by historical defects. This invention constructs a historical defect knowledge base, which stores historical defect triggering conditions and test scenarios in a structured manner, and automatically associates historical defect information with newly generated test cases during the test case review stage, thereby verifying and supplementing the newly generated test cases. This realizes a technical mechanism for the participation of historical defect knowledge in test case quality control.

[0065] (6) Construct a closed-loop control method for test case generation based on review results This invention uses a decision control module to judge the review results. When a test case does not meet the preset quality standards, it automatically triggers the test case generation agent to regenerate the test case, forming a closed-loop control process of "generation-review-feedback-regeneration". This introduces a quality gate and automatic rollback mechanism from a technical perspective.

[0066] (7) A collaborative control mechanism for RPA driven by agent review results is proposed. This invention uses the test case review results as the trigger condition for Robotic Process Automation (RPA). The test case is automatically added to the database only after it has passed the agent's review. This achieves state-driven collaborative control between the AI ​​agent and RPA, which is different from the passive execution of fixed processes by RPA in the prior art.

[0067] The above description is merely a detailed explanation of preferred embodiments and principles of the present invention. For those skilled in the art, there may be changes in specific implementation methods based on the ideas provided by the present invention, and these changes should also be considered within the scope of protection of the present invention.

Claims

1. A test case generation and intelligent management system based on multi-agent collaboration, characterized in that, include: The input and configuration module is used to interact with users and receive configuration information related to user-uploaded requirement documents and test case management. The intelligent agent collaboration module is used to generate and review test cases that meet quality requirements through the collaborative work of multiple intelligent agents, based on the requirements document and configuration information. The RPA robot execution module is used to trigger the automated import operation of test cases after the test cases have passed the review, and to import the reviewed test cases into the test case library of the corresponding project.

2. The test case generation and intelligent management system based on multi-agent collaboration as described in claim 1, characterized in that, The input and configuration module specifically includes the following processes: Receive user-uploaded requirement documents and user-inputted natural language query requests; Provide an interactive form interface for collecting configuration parameters related to test case management; The interactive form shall include at least the following fields: The target test management system includes selection fields, user account information, user password information, and the project identifier to which the test cases belong.

3. The test case generation and intelligent management system based on multi-agent collaboration according to claim 2, characterized in that, The intelligent agent collaboration module includes: A test case intelligence agent is generated, which is built based on a large language model. It is used to perform semantic understanding and test point decomposition on the requirement document, and automatically generate initial test cases according to the preset structured test case template. The test case review agent, built based on a large language model, is used to automatically review the initial test cases according to preset review rules and output review conclusions. The review result judgment module is used to control the flow of the process based on the review conclusion: if the review fails, the test case generation agent is triggered to regenerate the test cases; if the review passes, the RPA robot execution module is triggered to perform automated data entry.

4. The test case generation and intelligent management system based on multi-agent collaboration according to claim 3, characterized in that, The agent collaboration module also includes: A historical defect knowledge base is used to store structured historical defect data, which includes at least defect triggering scenario information; During the review process, the intelligent agent for the review test cases calls the historical defect knowledge base to compare the initial test cases with the historical defect data, and supplements or optimizes the test cases based on the comparison results.

5. The test case generation and intelligent management system based on multi-agent collaboration according to claim 4, characterized in that... The process by which the review test case agent supplements or optimizes the test cases based on the comparison results is as follows: When it is detected that the initial test cases do not cover a certain defect triggering scenario in the historical defect knowledge base, supplementary test cases to cover that scenario are automatically generated. When it is detected that the initial test case has covered a defect triggering scenario in the historical defect knowledge base, the priority of the corresponding test case is automatically increased.

6. The test case generation and intelligent management system based on multi-agent collaboration according to claim 5, characterized in that, The process of generating the use case agent specifically includes the following steps: Receive the requirements document, parse the functional descriptions, business rules, input and output conditions and abnormal scenarios in the requirements document, and form a semantic representation of the function to be tested; By using preset prompts, the large language model can simulate the role of a senior test engineer; The results are generated using structured, example-based test case templates. The test case template includes at least: Test case title, preconditions, test steps, expected results, and priority fields; Generate test case data that meets the import requirements of the test management system according to the test case template, and output it in Excel format.

7. The test case generation and intelligent management system based on multi-agent collaboration according to claim 6, characterized in that, The review case agent specifically includes the following process: Receive test cases output by the test case generating agent; By using preset review prompts, the large language model is guided to simultaneously simulate the combined roles of senior test engineers and developers, and to perform multi-perspective test case reviews from the angles of test sufficiency, implementation rationality, and risk coverage. Based on a predefined review rule system, a systematic check is performed on the quality of test cases; The review rule system includes at least the following: Does the test case format conform to the established specifications? Are the test cases complete, clearly described, and executable? Do the test cases cover the test scenarios for functional testing, performance testing, security testing, and boundary testing? 8. The test case generation and intelligent management system based on multi-agent collaboration according to claim 7, characterized in that, When the review result judgment module fails the review, it triggers the test case generation agent to regenerate the test cases, which specifically includes the following process: The review result judgment module feeds back the review comments or optimized prompts output by the review test case agent to the generating test case agent, which guides the generating test case agent to perform targeted regeneration, forming a closed-loop optimization process of "generation-review-feedback-regeneration".

9. The test case generation and intelligent management system based on multi-agent collaboration according to claim 8, characterized in that, The RPA robot execution module specifically includes the following processes: Invoke Robotic Process Automation (RPA) processes; Populate the test case data into an Excel file; The RPA executor is controlled to sequentially complete the following steps: opening the target test management system webpage, entering the user account and password, selecting the target project, and calling the test case import function provided by the system. Test cases are automatically imported into the test case library of the corresponding project, enabling centralized management and maintenance of test cases.

10. The test case generation and intelligent management system based on multi-agent collaboration according to claim 9, characterized in that, Also includes: The anomaly monitoring module is used to uniformly monitor abnormal situations during system operation, and specifically includes the following processes: Capture exceptions during the execution of each module; Perform automatic retry or abnormal termination operation based on the exception type.

Images