Task flow automatic generation method and device, computer equipment and storage medium

Abstract

The invention relates to the technical field of multi-model collaboration, and discloses a task process automatic generation method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining multi-modal input information; extracting semantic elements of the multi-modal input information, inputting the semantic elements into a preset story template, and injecting domain knowledge into the story template to output a story framework; performing structured processing and compliance and integrity verification on the story framework; inputting the structured user story which is verified to be qualified into a project management system to generate a final user story; and based on the final user story, updating the knowledge base to form an executable automatic task process. Through structural processing, knowledge enhancement and compliance integrity verification, in combination with an automatic access program and object storage-based knowledge base support, the problems of low efficiency and insufficient automation of traditional manual writing are solved, the compliance and speciality of user stories are guaranteed, and the convenient development requirements in the financial and medical fields are met.

Description

Technical Field

[0001] This invention relates to the field of multi-model collaboration technology, and in particular to a method, apparatus, computer equipment, and storage medium for automatic generation of task flows. Background Technology

[0002] In the software development process, user stories serve as a core vehicle for agile development, accurately describing user needs, functional scenarios, and business value. They are a crucial bridge connecting product managers, business stakeholders, and the development team, and their quality and workflow efficiency directly determine the progress of software development and the adaptability of the final product. Traditionally, the creation and management of user stories relied heavily on manual operations. Product managers or business personnel would manually write them based on experience, transmit them to the development team via unstructured methods such as emails and documents, and then manually enter them into the project management system.

[0003] However, this traditional approach has several pressing issues: First, writing user stories requires deep involvement from business personnel, which not only consumes significant time but also easily leads to vague descriptions of requirements and missing core semantic elements due to differences in individual experience and expression habits, causing misunderstandings in subsequent development work. Second, user stories written by different authors vary significantly in structure, fields, and expression, making it difficult to adapt to automated parsing and workflow requirements, severely impacting the collaboration efficiency of development teams. Third, existing tools mostly only support the storage and display of user stories, failing to achieve end-to-end automation from requirement input, structured generation, compliance verification to system interaction, resulting in excessive manual intervention and hindering the iteration speed of agile development. Therefore, it cannot meet the needs of automated generation and management of user stories in complex business scenarios, especially in fields such as finance and healthcare where compliance, professionalism, and security requirements are extremely high. For example, in the financial sector, stringent compliance requirements necessitate strict adherence to regulations such as anti-money laundering laws and data privacy protection rules. Traditionally handwritten user stories are prone to overlooking compliance clauses and using non-standard industry terminology, leading to compliance risks in generated requirements. Similarly, in the healthcare sector, medical procedures are highly specialized, involving diagnostic and treatment guidelines, equipment operating standards, and medical insurance policies. Handwritten user stories are susceptible to errors in terminology and missing core business logic, potentially resulting in healthcare systems that do not meet clinical needs. Furthermore, the healthcare sector demands a high degree of completeness in user stories; any omission can affect the safety and continuity of medical processes. Current technologies lack targeted professional knowledge enhancement and precise verification mechanisms, making it difficult to guarantee the professionalism and completeness of user stories. Moreover, the low efficiency of requirement transmission becomes even more pronounced during cross-hospital and cross-departmental collaborative development, hindering the deployment speed of healthcare information systems.

[0004] Therefore, there is an urgent need for a compliant and highly professional method, device, computer equipment, and storage medium for automatically generating end-to-end task processes. Summary of the Invention

[0005] This invention provides a method, apparatus, computer equipment, and storage medium for automatically generating task flows, in order to solve the technical problems of existing user story generation and management technologies, such as strong reliance on manual labor, low degree of automation, and insufficient compliance and professionalism.

[0006] Firstly, a method for automatically generating task flows is provided, including: Acquire multimodal input information; Semantic elements of the multimodal input information are extracted, and these semantic elements are input into a preset story template. Domain knowledge is then injected into the story template to output a story framework. The domain knowledge is stored in a knowledge base and includes pre-collected industry terms and compliance requirements. The story framework is then structured to generate structured user stories; Perform compliance and integrity checks on the structured user stories; Based on the verified structured user story, a preset automatic access program is triggered to input the verified structured user story into the project management system to generate the final user story. Based on the end-user story, the knowledge base is updated to form an executable automated task flow.

[0007] Secondly, it provides a task flow automatic generation device, including: The acquisition module is used to acquire multimodal input information; An extraction module is used to extract semantic elements from the multimodal input information, input the semantic elements into a preset story template, and inject domain knowledge into the story template to output a story framework; wherein, the domain knowledge is stored in a knowledge base, and the domain knowledge includes pre-collected industry terms and compliance requirements; A structuring module is used to perform structuring processing on the story framework to generate structured user stories; The verification module is used to perform compliance and integrity verification on the structured user stories; The generation module is used to trigger a preset automatic access program based on the verified structured user story, and input the verified structured user story into the project management system to generate the final user story; The management module is used to update the knowledge base based on the end-user stories, forming an executable automated task flow.

[0008] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described automatic task flow generation method.

[0009] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the above-described automatic task flow generation method.

[0010] The beneficial effects of this invention compared to existing technologies are as follows: This invention achieves automatic collection and processing of user stories through multimodal input adaptation, structured processing, knowledge enhancement, and compliance integrity verification. Combined with automatic access programs and object-based knowledge base support, it achieves automatic construction and management of user stories, effectively solving the problems of low efficiency and insufficient automation in traditional manual writing. It ensures the compliance and professionalism of user stories, realizes full automation of the user story generation and management task process, improves collaboration and development efficiency, and adapts to the convenient development needs of the financial and medical fields.

[0011] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other objects, features and advantages of the present invention more obvious and understandable, preferred embodiments are described in detail below. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of an application environment for an automatic task flow generation method according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating a method for automatically generating task flows according to an embodiment of the present invention; Figure 3 yes Figure 2 A schematic diagram of a specific implementation method for step S20; Figure 4 yes Figure 2 A schematic diagram of a specific implementation of step S40; Figure 5 yes Figure 2 A schematic diagram of a specific implementation method for step S50; Figure 6 yes Figure 2 A schematic diagram of a specific implementation method for step S60; Figure 7 This is a schematic diagram of a task flow automatic generation device in one embodiment of the present invention; Figure 8 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention; Figure 9 This is another structural schematic diagram of a computer device according to one embodiment of the present invention. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The technical solutions in the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0014] It should be understood that, when used in this specification and the appended claims, the terms “comprising” and “including” indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0015] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0016] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0017] Please see Figure 1 and Figure 2 , Figure 1This diagram illustrates an application scenario of the automatic task flow generation method provided in this invention. Users can input user story information via a client. The client communicates with the server via a network; the server obtains multimodal input information via the network; based on semantic element extraction and knowledge enhancement, the input information is structured to generate a structured user story; the structured user story undergoes compliance and integrity verification; through API and UI integration, the verified structured user story interacts with the project management system; and an executable automated task flow is formed by constructing, updating, and calling a knowledge base using object storage. The client can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be implemented using a standalone server or a server cluster consisting of multiple servers. The invention will be described in detail below through specific embodiments.

[0018] Please see Figure 2 As shown, Figure 2 This is a schematic flowchart illustrating a method for automatically generating task flows according to an embodiment of the present invention. The method includes the following steps: S10: Obtain multimodal input information.

[0019] In this embodiment, multimodal input information includes, but is not limited to, various forms of input requirements such as voice, multilingual text, Word / PDF attachments, AI prompts, and human-generated text. Specifically, in the financial sector, users can upload PDF documents outlining their cross-border payment compliance requirements; in the medical field, doctors can input functional requirements for the electronic medical record system via voice, which the system then automatically converts to text.

[0020] For step S10, the user story-supported multimodal input module achieves adaptation, which can meet the input habits of different users, break the limitation of a single input form, reduce the user's operation threshold, and at the same time realize the standardized conversion of various inputs, laying the foundation for subsequent structured processing and improving the efficiency of demand input.

[0021] To achieve a fully automated workflow for user story generation and management, this embodiment employs a Python-based automated user story creation and management toolkit using Playwright. Its core architecture includes a multimodal input module. This module integrates the SpeechRecognition library for speech-to-text conversion, supports multilingual parsing through the Transformer model, parses attachments using a Docx / PDF Parser tool, and is compatible with both AI-driven natural language input prompts (such as "Users can view the order list after logging in") and human text input. It also supports automatic formatting of directly input user story content.

[0022] Python Playwright is an open-source UI automation testing tool from Microsoft. Supporting programming languages ​​such as Python, it enables full-scene automated control of the browser, simulating all human interactions (clicks, input, page navigation). It supports cross-browser and cross-platform operation and boasts stable element location, operation recording, and playback capabilities. Furthermore, it can be combined with OCR plugins for operation result verification. In this embodiment, when API calls to the project management system fail, it automatically switches to the Playwright UI automation workflow to perform visual user story entry. Python Playwright is also used to collect historical user story data from the project management system through automated workflows to update the knowledge base. It can also work with visual recognition plugins to verify the accuracy of UI automation entry results, ensuring error-free operation.

[0023] The SpeechRecognition library is a mature speech recognition development library in the Python ecosystem. It provides a unified interface to adapt to various speech recognition engines, supports the conversion of audio signals (speech) into text data, and is compatible with various speech sources such as real-time microphone input and audio file parsing. It has cross-platform compatibility and high recognition accuracy. In this embodiment, after the user submits a voice input request, the SpeechRecognition library converts the speech signal into standardized text, which is then passed to the subsequent semantic understanding and structured generation modules to complete the unified adaptation of multimodal input.

[0024] The Transformer model is a deep learning model based on self-attention mechanism and is a core foundational model in modern Natural Language Processing (NLP). It breaks the sequence dependency limitations of traditional Recurrent Neural Networks (RNNs) and possesses powerful capabilities in contextual semantic understanding, language conversion, and text generation. It excels particularly in cross-language translation, semantic alignment, and long text parsing, serving as a core architectural foundation for multilingual processing and Large Language Models (LLMs). In this embodiment, when a user inputs a natural language request that is not in Chinese, the Transformer model uses its cross-language parsing capabilities to convert the request text in different languages ​​into a unified, standardized text. This ensures that subsequent semantic element extraction and structured generation modules can process the request correctly, achieving automated generation of cross-language user stories.

[0025] S20: Extract the semantic elements of the multimodal input information, input the semantic elements into a preset story template, and inject domain knowledge into the story template to output a story framework; wherein, the domain knowledge is stored in a knowledge base, and the domain knowledge includes pre-collected industry terms and compliance requirements.

[0026] In this embodiment, the structured processing of input information is achieved through multi-model collaboration based on the MCP technology framework. MCP (Model Collaboration Platform) is a distributed AI model collaboration platform based on a microservice architecture. It can integrate AI models with different capabilities and automate complex tasks through unified orchestration and knowledge sharing. It features a multi-model orchestration engine, a shared knowledge base system, an API gateway, standardized output, and visual orchestration tools. In this embodiment, it is used to parse natural language requirements, extract roles, functions, and value elements, generate standardized user stories, and inject industry terminology and compliance requirements.

[0027] The story template serves as the data structure for generating user stories, including fields, field types, and whether fields are required. In practice, a structured generation model generates the basic structure of a JSON Schema based on requirements, project type, and industry scenario. This JSON Schema includes multiple fields for storing semantic elements, industry terminology, compliance requirements, etc. Preferably, this embodiment uses the Qwen-Turbo model as the structured generation model to quickly build the story framework. The Qwen-Turbo model is a lightweight large language model developed by Alibaba Cloud, featuring fast response, structured generation, context understanding, and low-cost deployment capabilities. It can generate JSON Schema constraints based on MCP and complete the framework fields according to preset templates. JSON Schema is a metadata specification for describing JSON data formats. Its core functions include: data validation (defining field types, such as string / number), required fields, and format constraints (such as email, URL); structured constraints (describing complex data models through nested structures, such as nested objects, arrays); and automated validation (supporting automatic validation of JSON data conforming to the specification using tools such as Ajv, jsonschema).

[0028] For step S22, by unifying the basic structure of user stories, the problem of large differences in writing format among different people is solved, providing a unified carrier for subsequent knowledge enhancement and format verification, and improving the compatibility of automated processing.

[0029] It is also understood that, in this embodiment, semantic element extraction refers to extracting core information such as roles, functions, and values ​​from the requirements; injecting pre-collected and stored industry terms and compliance requirements into the story framework, that is, enhancing the knowledge of the user stories to be generated, expanding them from simple semantic elements into descriptions with professionalism and business details. Through step S20, vague and non-standardized requirements can be transformed into logically clear, complete, and professional story frameworks, solving the problem of chaotic traditional requirement descriptions and providing adaptable data for subsequent verification and system interaction.

[0030] In some embodiments of the present invention, such as Figure 3 As shown, a specific input information processing scheme is provided. In S20, the semantic elements of the multimodal input information are extracted, the semantic elements are input into a preset story template, and domain knowledge is injected into the story template to output a story framework. Specifically, it includes the following steps S21-S24.

[0031] S21: Use a semantic understanding model to parse the natural language requirements of the input information and extract the core semantic elements.

[0032] The semantic understanding model refers to a large language model with natural language parsing capabilities. Core semantic elements may include user roles, core functions, and business value. Preferably, this embodiment uses the Qwen-Max model as the semantic understanding model to accurately parse the essence of the requirements. The Qwen-Max model is a large language model (LLM) with hundreds of billions of parameters developed by Alibaba Cloud. Based on the Transformer architecture, it features complex semantic understanding, multimodal processing, high-precision generation, and zero-shot learning capabilities.

[0033] For step S21, by extracting key information from the requirements, avoiding the omission of core semantic elements, and solving the problems of low efficiency and large deviation in manual extraction, core basis is provided for the generation of story framework.

[0034] S22: Input the core semantic elements into the preset story template.

[0035] For step S22, automated input is achieved through the structured output control module of MCP. Specifically, relying on the preset story templates defined by JSONSchema (the story templates are pre-classified and stored in object storage according to project type, industry scenario, etc., and support dynamic loading), combined with MCP's field mapping engine, the automatic matching and filling of core semantic elements with template fields is completed. The structure of the preset story template follows the constraint rule of field name-field type-mandatory field.

[0036] Therefore, step S22, through the strong constraints of the JSON Schema template, can uniformly fill scattered core semantic elements, such as "role" and "function," into fixed fields of the story template, avoiding format differences caused by different user inputs. This provides a unified data foundation for subsequent structured processing and compliance verification, solving the problem of chaotic formatting in traditional user stories. Furthermore, by leveraging MCP's field mapping engine, it achieves automatic matching between elements and template fields, reducing manual intervention. At the same time, through "mandatory" constraints, it identifies missing core fields in advance, such as acceptance standards and compliance requirements, avoiding rework in subsequent processes due to missing fields. Additionally, it supports dynamically loading corresponding story templates based on project type and business scenario, ensuring that user stories from different domains can match exclusive fields, laying the foundation for accurate injection of domain knowledge in the future.

[0037] S23: By calling the domain knowledge enhancement model through the model collaboration platform, the domain knowledge enhancement model, based on the knowledge base, injects industry terms and compliance requirements into the story framework by field to obtain a complete story framework.

[0038] In this field, the automated user story creation and management tool suite incorporates the aforementioned semantic understanding model, structured generation model, and domain knowledge enhancement model. The domain knowledge enhancement model is used to inject industry terms and compliance requirements (such as anti-money laundering and data privacy) into the generation logic. Industry terms and compliance requirements refer to professional vocabulary and regulatory constraints in specific fields.

[0039] In practical implementation, for example, in the financial sector, compliance requirements such as "anti-money laundering verification" and "customer identity authentication" are incorporated into the framework, along with industry terminology such as "T+0 settlement." Similarly, in the medical field, requirements such as "encrypted storage of patient privacy" and "adaptation to medical insurance policies" are incorporated, along with terminology such as "standardized electronic prescription circulation." In other words, compliance requirements and industry terminology from financial anti-money laundering regulations and medical data security laws are precisely integrated into the framework, enhancing the specific application of the technology within its domain.

[0040] For step S23, by improving the industry adaptability and compliance of user stories, the problem of general stories lacking professional depth is solved, and deviations in the implementation of requirements due to a lack of industry knowledge are avoided.

[0041] S24: Based on the completed story framework, establish a logical relationship network among the core semantic elements, generate a standardized story framework, and output it.

[0042] It is understandable that a logical association network refers to a network of causal and subordinate relationships between core semantic elements, such as the correspondence between roles and functions, and functions and compliance requirements. This embodiment constructs this network based on knowledge graph technology. Through semantic association analysis between elements, it ensures that the connection of each core semantic element conforms to industry business process specifications and generates standardized user stories with logical closed loops.

[0043] For step S24, by establishing a logical connection network among core semantic elements, the logical coherence and integrity of the user story are ensured, the problems of element fragmentation and logical contradictions are solved, and the story can accurately reflect the essence of the business process.

[0044] S30: The story framework is structured to generate a structured user story.

[0045] For step S30, in specific implementation, the data format for generating user stories is preset so that the output is in accordance with the preset data format during the structured processing. In this embodiment, the preset data format refers to the JSON Schema format that conforms to the project type, and the structured user story refers to a standardized requirement carrier that strictly follows this format. In specific implementation, for example, in the financial field, stories containing mandatory fields such as "transaction type" and "compliance verification node" are generated according to the JSON Schema of the financial transaction system; another example is in the medical field, stories containing fields such as "diagnosis and treatment module" and "data security level" are generated according to the Schema of the medical information system.

[0046] Preferably, in specific implementation, the Structured Output Plugin of MCP can be used to dynamically load the corresponding JSON Schema template according to the project type, strictly restrict the field type and format to ensure that the output format is uniform, so that user stories can be directly adapted to automated parsing and system interaction, solve the problem of low flow efficiency caused by format incompatibility, and provide data support for subsequent verification and writing to the system.

[0047] In some embodiments of the present invention, a specific structured user story generation scheme is provided. In S30, that is, the story framework is structured to generate a structured user story, which specifically includes the following steps: The standardized story framework is restricted in terms of field types and formats using a structured output plugin. The standardized story framework after output constraints is the structured user story.

[0048] It is understandable that the Structured Output Plugin refers to the MCP's Structured Output Plugin, which is used to constrain the type (such as number, string), length, and format of fields. The plugin intervenes in the validation during the generation stage, and performs real-time format validation on each field based on the preset JSON Schema template to ensure that there are no issues such as type errors or excessive length, avoids automatic processing failures caused by field format errors, solves the problem of inconsistent formats, and improves the usability and compatibility of data.

[0049] In practice, for example, in the financial field, the "transaction amount" field is restricted to a numeric type and the "customer number" field is restricted to a 10-digit string; in the medical field, the "medical record number" field is restricted to a 12-digit string and the "vital signs value" field is restricted to a numeric type.

[0050] It is also understood that a structured user story refers to a standardized requirements document that fully conforms to JSON Schema requirements after field restrictions; in this embodiment, it serves as the core data carrier for subsequent validation and system interaction. Preferably, the output structured user story can be accompanied by a format validation log, allowing traceability of the field validation process, and the format is fully compatible with the interface requirements of the enterprise-level project management system, supporting direct transmission.

[0051] The standardized story framework with output constraints, that is, the output of standardized data that can be directly reused, can reduce the cost of subsequent manual correction and ensure that user stories can be seamlessly connected to the project management system interface.

[0052] S40: Perform compliance and integrity checks on the structured user stories.

[0053] In this embodiment, compliance verification refers to checking whether the story complies with industry regulations and policy requirements, while integrity verification refers to checking whether it covers core business logic. This embodiment adopts a hybrid intelligent verification mechanism of MCP and rule engine for dual verification to identify compliance risks and logical vulnerabilities in user stories in advance, solving the problems of low efficiency and high risk of omission in manual verification, and ensuring the security and feasibility of requirement implementation.

[0054] In some embodiments of the present invention, such as Figure 4 As shown, a specific structured user story verification scheme is provided. In S40, the structured user story is verified for compliance and integrity, which specifically includes the following steps S41-S44.

[0055] S41: Use the model collaboration platform to call the classification model and check whether the acceptance criteria for formatted user stories cover the core business logic.

[0056] In this embodiment, the classification model is the NLU (Natural Language Understanding) model built into MCP. The classification model is trained based on historical user story cases in the knowledge base, which can accurately identify the key nodes of the core business logic in different domains, ensuring that there are no core omissions in the acceptance criteria. The core business logic refers to the key process to realize the value of the requirement, which is used to judge the completeness of the acceptance criteria, ensure that the user story is acceptable, avoid the development results not matching the requirements due to the lack of core logic, and reduce rework costs.

[0057] In practice, for example, in the financial sector, the acceptance criteria for cross-border transfers include "transaction limit verification" and "arrival time confirmation"; in the medical field, the criteria for electronic prescriptions include "drug dosage verification" and "prescription validity period verification".

[0058] S42: Use the anomaly detection model through the model collaboration platform to identify compliance risks or logical vulnerabilities in formatted user stories.

[0059] Among them, the anomaly detection model refers to an AI model that can identify illegal content and logical contradictions. In this embodiment, the anomaly detection model is a dedicated detection model built into MCP. It can combine the compliance case library in the object storage to update the risk identification rules in real time and accurately identify hidden risks in areas such as financial anti-money laundering and medical privacy protection. Compliance risks refer to content that violates industry regulations, and logical vulnerabilities refer to contradictions in the relationship between elements.

[0060] Step S42 proactively mitigates project compliance risks and logical errors, addresses the issue of hidden risks that are difficult to identify manually, and ensures the compliance and stability of software development.

[0061] It is understandable that there may be compliance risks or logical loopholes in the specific implementation. For example, in the financial field, there is a compliance loophole in the identification story that "customer identity authentication is not mentioned"; and in the medical field, there are privacy risks in the identification of "patient data not being stored in an encrypted manner" and logical loopholes in the identification of "prescriptions not requiring doctor authorization".

[0062] S43: If both the classification model and the anomaly detection model fail the verification, output a correction suggestion to regenerate the user story.

[0063] In this embodiment, the correction suggestion refers to the targeted modification suggestions generated by AI, including missing core logic, violations, and directions for supplementation. It is automatically generated based on the model analysis results, providing users with clear optimization directions, reducing the blindness of manual modification, improving the efficiency of user story iteration, and ensuring accurate rectification of problems.

[0064] In practice, the correction suggestions can clearly indicate the problem type (compliance deficiency / logic loophole), specific location and supplementary basis. After output, it will automatically return to step S21 to restart the story framework output process, so as to achieve closed-loop optimization.

[0065] S44: If both the classification model and the anomaly detection model pass the verification, output the qualified structured user story.

[0066] Among them, a qualified structured user story refers to a requirement document that has no compliance risks, complete core logic, and standard format; in this embodiment, it is used as qualified data input for subsequent system interactions.

[0067] For step S44, output high-quality, directly implementable user stories to provide reliable data support for system interaction and ensure the accuracy and compliance of requirement transmission.

[0068] S50: Based on the verified structured user story, a preset automatic access program is triggered to input the verified structured user story into the project management system to generate the final user story.

[0069] In this embodiment, qualified structured user stories are automatically connected to the project management system via an automated access program to achieve automatic interaction with the system. Specifically, the automated access program integrates MCP's API gateway technology and Python Playwright UI automated operation processes. The API and UI integration method prioritizes interaction via the API interface, switching to a hybrid interaction mode that uses UI automation if the API fails. The project management system refers to an enterprise-level agile development management platform. In a preferred embodiment, such as this one, the project management system is the Shenbing system, specifically the Shenbing R&D Management Platform (Wizard), independently developed by Ping An Technology. This platform is a one-stop R&D management and continuous delivery platform for internet finance. In this embodiment, the Shenbing system serves as the core interface target, acting as a hub for requirements management, process execution, and knowledge base integration.

[0070] In some embodiments of the present invention, such as Figure 5 As shown, a specific interaction scheme is provided. In S50, a preset automatic access program is triggered based on the verified structured user story to input the verified structured user story into the project management system to generate the final user story. Specifically, it includes the following steps S51-S53.

[0071] S51: Obtain the verification result of the structured user story. If the verification passes, obtain the access credentials of the project management system through the identity authentication system of the project management system.

[0072] It is understandable that the identity authentication system of the project management system is a security mechanism for verifying the legitimacy of visitors. In this embodiment, the identity authentication system is specifically the OAuth2.0 authentication system of the Shenbing system, which confirms the operation permissions of the access subject through a standardized authorization process; the access credential is a temporary authorization token (such as a Token) generated by identity authentication, containing core information such as the scope of access permissions and the validity period. Among them, the OAuth2.0 authentication system is a commonly used third-party authorization login mechanism. The OAuth2.0 authentication system is used to achieve seamless login without the need for manual input of account passwords, ensuring login security. In this embodiment, the identity authentication system provides a secure access barrier for the interaction between the tool suite and the project management system, avoiding the leakage of business data due to unauthorized access; the access credential serves as the legitimate basis for subsequent API calls and automatic login, ensuring that all system operations are traceable and have compliant permissions.

[0073] For step S51, access credentials are obtained through a standardized identity authentication process, which ensures the security of the interaction between the tool suite and the project management system from the source. This solves the problems of chaotic permissions, incomplete operation records, and high risk of password leakage that exist in traditional manual login. At the same time, obtaining access credentials provides the necessary prerequisite for subsequent API calls and automatic system login, laying a solid security foundation for full-process automation and avoiding process interruption due to permission issues.

[0074] S52: Call the API interface and send the validated structured user story to the project management system based on the access credentials.

[0075] In this context, calling the API interface refers to the interactive method of data transmission through application programming interfaces (APIs). In this embodiment, based on a validated structured user story in JSON format, it directly connects to the data interface of the project management system. The transmission process is encrypted to ensure data security and accuracy. Calling the API interface requires the use of API gateway technology. API gateway technology is middleware that can dynamically adapt to different interface protocols, including those of different versions of the project management system.

[0076] For step S52, the synergy of API and identity authentication system solves the problem of incompatibility between interfaces of different versions of project management system, realizes seamless login and rapid transmission and entry of user stories, reduces manual intervention, eliminates the need for users to manually operate graphical interactive pages (such as web pages, APP, etc.) to log in, submit, etc., avoids the tedious operation of manual login, solves the problems of low efficiency and error-proneness of traditional manual entry, and improves the speed of agile development iteration.

[0077] In some embodiments of the present invention, another project management system interaction scheme is also provided. In S52, that is, calling the API interface and sending the verified structured user story to the project management system based on the access credentials, the scheme further includes the following steps: If the API call fails, a fault-tolerant switching mechanism is initiated, through which the validated structured user story is sent to the project management system.

[0078] This embodiment implements fault-tolerant backup based on Python Playwright. After an API call fails, it switches to the automated operation process of the interface, i.e., Playwright UI, which can ensure that the interaction process is not interrupted, solve the problem that a single API call is prone to failure due to interface failure, and improve the stability and reliability of the entire process.

[0079] In practice, the fault-tolerant switching mechanism is automatically triggered by the MCP collaboration module. After switching, the PythonPlaywright interface automation operation process is started, the input operation is performed according to the preset script, and the result is verified by MCP-OCR.

[0080] In some embodiments of the present invention, a specific API call failure handling scheme is provided. That is, if the API interface call fails, a fault-tolerant switching mechanism is entered, and the validated structured user story is sent to the project management system through the fault-tolerant switching mechanism. The specific steps include the following: Trigger the automated operation scripts of the project management system interface; Based on the interface automation operation script, switch to the interface automation operation process and perform a visual operation for user story entry to simulate manual operation by the user. Call the visual recognition plugin to verify the accuracy of the visual operation results; If the visualization operation result is verified to be accurate, the verified structured user story is sent to the project management system. If the visualization operation result verification is inaccurate, the fault tolerance switching mechanism will be re-executed until the visualization operation result verification is accurate.

[0081] Understandably, the Python Playwright tool supports simulating various interface interactions, pre-sets operation scripts adapted to the project management system interface, and can accurately locate input fields and automatically populate structured user story data. The interface automation process, namely the Playwright UI automation process, is a visual operation process that simulates manual clicks and input through the Python Playwright tool. Visual operations refer to simulating user interactions on the system interface, including but not limited to entering access credentials, clicking the login button, and other login operations, as well as entering validated structured user stories and clicking the generate button. In this embodiment, the MCP-OCR (Model Collaboration Platform Optical Character Recognition) plugin is used as the visual recognition plugin. MCP-OCR is a plugin module in the MCP framework specifically designed for optical character recognition (OCR) and semantic enhancement. It generates a comparison result by comparing the character content of the input interface with the corresponding fields of the validated story. If the error exceeds a threshold, it is judged as inaccurate. The visual operation result refers to the user story content entered in the interface automation process. Creation and configuration refers to generating user story entries in the project management system and setting attributes such as priority and module. In this embodiment, it is used to complete the final implementation of user stories, realize full automation from generation to configuration, avoid the tedious operation of manual configuration, and improve agile development efficiency. In specific implementation, the configuration process can be automatically executed according to preset rules, and attributes such as priority and module can be dynamically adjusted according to the project type. After creation, a configuration log is generated for subsequent viewing.

[0082] The automated switching and processing of interface-based operations replaces manual data entry, solves the problem of process interruption after API call failure, ensures that user stories can be successfully entered into the project management system, reduces manual intervention, and guarantees full automation of user story generation and management.

[0083] S54: After the project management system verifies the access credentials, complete the automatic login to the project management system and generate the end-user story.

[0084] In this embodiment, the project management system, through its own identity authentication system, verifies the access credentials and then transmits the complete structured user story data to and stores it within the project management system. This data is then sent to subsequent development projects, allowing developers to use the end-user story for project development. This embodiment utilizes an encrypted API interface for data writing, enabling rapid implementation of user stories and ensuring that requirements are quickly communicated to the development team. Preferably, the end-user story can be output through a UI interface for feedback to the user.

[0085] In practice, for example, in the financial sector, compliant cross-border payment stories are fully written into the project management system via API and synchronized to the development and testing teams in real time; in the medical sector, electronic prescription system stories are written in, triggering team collaboration processes.

[0086] S60: Based on the end-user story, update the knowledge base to form an executable automated task flow.

[0087] Understandably, the executable automated task workflow integrates the entire lifecycle management of user stories, from requirement input, structured processing, compliance verification, system integration to knowledge base updates, forming a standardized, closed-loop automated execution chain. This chain includes triggering conditions, execution logic, fault tolerance mechanisms, and output standards for each stage, allowing for seamless execution according to preset rules without manual intervention. In this embodiment, this workflow is both the core deliverable of the toolkit and a crucial link between user story generation and project development execution. It solidifies fragmented steps into reusable workflow templates, supporting rapid reuse for subsequent similar business needs. Furthermore, through automated interaction with the project management system, it transforms the final user story into task items that the development team can directly undertake, clearly defining task priorities, modules, acceptance criteria, and other core information to ensure consistency and efficiency in requirement implementation.

[0088] Step S60, by updating the knowledge base based on end-user stories, achieves continuous accumulation and iteration of domain knowledge. New industry terms, compliance requirements, and business logic relationships are added to the knowledge base, enabling subsequent AI-generated user stories to draw upon richer historical experience and the latest domain knowledge. This significantly improves the accuracy and compliance of requirement generation, solving the problems of lagging knowledge base updates and fragmented knowledge in traditional systems. The resulting executable automated task process completely breaks down the barriers between requirement generation and development execution, integrating previously scattered manual operations such as requirement organization, task creation, and knowledge archiving into a closed-loop automated process. This significantly shortens the cycle from requirement generation to implementation. Standardized process templates ensure consistency across types and projects, reducing communication costs and rework risks caused by process differences. Furthermore, the dynamic updating of the knowledge base and the reuse of process templates form a virtuous cycle. As usage frequency increases, the industry adaptability and automation efficiency of the tool suite continuously improve, providing core support for sustainable evolution in enterprise agile development.

[0089] In some embodiments of the present invention, such as Figure 6 As shown, a specific management scheme is provided. In S60, the knowledge base is updated based on the end-user story to form an executable automated task process, which specifically includes the following steps S61-S64.

[0090] S61: By using the object storage compatibility protocol configured in the project management system, historical user stories, industry terms, and compliance requirements data are classified and stored in the knowledge base to achieve mixed storage of structured and unstructured data.

[0091] Among them, object storage compatibility protocols refer to mainstream object storage protocols such as S3, Swift, and OSS; structured data refers to historical records in JSON format; and unstructured data refers to compliant PDF documents, meeting recordings, etc. This embodiment realizes unified storage of multiple types of data, solves the problem of scattered storage of different types of knowledge data, enables elastic expansion of the knowledge base, and supports efficient management and fast access to massive amounts of knowledge data.

[0092] For example, in the financial sector, historical cross-border payment stories in JSON format and anti-money laundering documents in PDF format are stored; in the medical sector, structured electronic medical record requirements and recordings of medical consultations are stored.

[0093] In practice, object storage can adopt an intelligent tiered storage strategy, storing hot data (common industry terms), warm data (recent user stories), and cold data (archived cases) on different types of storage media according to data access frequency, in order to improve access efficiency.

[0094] The classification and storage mechanism in step S61 directly supports the implementation of four core application scenarios of object storage: First, model pre-training data scenario, such as using collections of financial compliance documents and medical treatment manuals as model pre-training data, achieving efficient storage and retrieval through the batch read and write capabilities of object storage; Second, real-time knowledge retrieval scenario, such as the current project's specific financial business rule library and medical UI component library, after being classified and stored in this step, achieving low-latency access with the help of the SDK, providing real-time support for domain knowledge enhancement in step S23 and compliance verification in step S32; Third, version archiving scenario, such as historical versions of user story templates and AI model inference logs, after being classified and archived in this step, achieving full lifecycle traceability using the version control function of object storage; Fourth, multimodal knowledge management scenario, such as multimodal data like recordings of financial requirement review meetings and OCR results of medical interface screenshots, achieving unified management and rapid retrieval through the combination of structured metadata annotation and unstructured data storage in this step, further improving the multi-dimensional knowledge coverage of the knowledge base.

[0095] S62: Utilize the automated operation process of the interface to automatically collect historical data from the project management system, incrementally update the knowledge base, and construct a knowledge graph.

[0096] It is understandable that incremental updates refer to updating only newly added or changed data, while knowledge graphs refer to a network formed by structuring and linking knowledge data.

[0097] For step S62, data collection enables the dynamic evolution of the knowledge base, solving the problem of low efficiency in manually updating the knowledge base, and the knowledge graph improves the accuracy and relevance of knowledge retrieval.

[0098] In practice, for example, in the financial sector, newly added cross-border payment stories are automatically collected from the project management system to update the financial knowledge base and improve the knowledge link between "payment and compliance"; in the medical sector, newly added electronic medical record requirements are collected to build a knowledge graph of "diagnosis and treatment functions and privacy protection".

[0099] Furthermore, in this embodiment, Python Playwright is used to automatically crawl historical user stories, requirement change records, and other data from the project management system. After classifying the data according to preset rules, the knowledge base is updated, and the knowledge graph is automatically constructed and optimized based on the semantic associations of the data.

[0100] S63: An access control system based on the project management system configuration, which controls access permissions to the knowledge base and manages and traces multiple versions of knowledge base content through the version control function of object storage.

[0101] In this embodiment, an RBAC (Role-Based Access Control) system is used as the access control system. Users and permissions are connected through roles. Users refer to system users, such as product managers, developers, and testers; roles refer to sets of permissions, such as requirement administrators and compliance auditors; and permissions refer to authorized operations on system resources, such as creating user stories and viewing test reports. In this embodiment, the RBAC system is used to ensure the security of enterprise-level sensitive data and the isolation of permissions in multi-team collaboration. Only authorized personnel are allowed to edit user stories related to financial compliance; testers can only view acceptance criteria and cannot modify core functional descriptions.

[0102] In this embodiment, the version control function refers to recording the historical versions of the knowledge base content and supporting backtracking; this embodiment is used to ensure the security of the knowledge base and the traceability of data.

[0103] For step S63, the knowledge base access security issue is resolved by restricting access permissions and version control, preventing the leakage of sensitive knowledge, while supporting historical version rollback, which facilitates troubleshooting and reuse of historical knowledge.

[0104] S64: Based on the knowledge base, form an executable automated task process and output a traceable execution log.

[0105] In this embodiment, the automated task flow refers to the end-to-end automated process of user story generation, verification, interaction, and knowledge updating, while the execution log refers to a document recording the execution status of each step. The execution log includes information such as the execution time, results, involved data, and verification nodes for each step, and is stored in object storage, supporting querying and exporting by project, time, step type, and other dimensions. Step S64 achieves closed-loop automation of the entire method, solving the problem of process fragmentation. The execution log supports problem investigation and auditing, improving the manageability of the process.

[0106] As can be seen, the above solution achieves automatic collection and processing of user stories through multimodal input adaptation, structured processing, knowledge enhancement, and compliance integrity verification. Combined with automatic access programs and object storage-based knowledge base support, it enables automatic construction and management of user stories, effectively solving the problems of low efficiency and insufficient automation in traditional manual writing. It ensures the compliance and professionalism of user stories, realizes full automation of the user story generation and management task process, improves collaboration and development efficiency, and adapts to the convenient development needs of the financial and medical fields.

[0107] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0108] In one embodiment, the present invention provides a task flow automatic generation device 100, which corresponds one-to-one with the task flow automatic generation method in the above embodiments. For example... Figure 7 As shown, the automatic task flow generation device 100 includes an acquisition module 101, an extraction module 102, a structuring module 103, a verification module 104, a generation module 105, and a management module 106. Detailed descriptions of each functional module are as follows: The acquisition module 101 is used to acquire multimodal input information.

[0109] The extraction module 102 is used to extract the semantic elements of the multimodal input information, input the semantic elements into a preset story template, and inject domain knowledge into the story template to output a story framework; wherein, the domain knowledge is stored in a knowledge base, and the domain knowledge includes pre-collected industry terms and compliance requirements.

[0110] The structuring module 103 is used to perform structuring processing on the story framework to generate a structured user story.

[0111] The verification module 104 is used to perform compliance and integrity verification on the structured user story.

[0112] The generation module 105 is used to trigger a preset automatic access program based on the verified structured user story, and input the verified structured user story into the project management system to generate the final user story.

[0113] The management module 106 is used to update the knowledge base based on the end-user story to form an executable automated task process.

[0114] In one embodiment, the extraction module 102 is specifically used for: The input information is analyzed using a semantic understanding model to extract its natural language requirements and core semantic elements. The core semantic elements are input into a preset story template; wherein, the story template is a data structure for generating user stories, including fields, field types, and whether they are required. By calling the domain knowledge enhancement model through the model collaboration platform, the domain knowledge enhancement model injects industry terms and compliance requirements into the story framework by field based on the knowledge base, thus obtaining a complete story framework; Based on the completed story framework, a logical relationship network among the core semantic elements is established, and a standardized story framework is generated and output.

[0115] In one embodiment, the structured module 103 is specifically used for: The standardized story framework is restricted in terms of field types and formats using a structured output plugin. The standardized story framework after output constraints is the structured user story.

[0116] In one embodiment, the verification module 104 is specifically used for: By calling the classification model through the model collaboration platform, we can check whether the acceptance criteria for formatted user stories cover the core business logic. By calling the anomaly detection model through the model collaboration platform, compliance risks or logical vulnerabilities in formatted user stories can be identified. If both the classification model and the anomaly detection model fail the validation, a correction suggestion is output to regenerate the user story; If both the classification model and the anomaly detection model pass the verification, a qualified structured user story is output.

[0117] In one embodiment, the generation module 105 is specifically used for: Obtain the verification result of the structured user story. If the verification passes, obtain the access credentials for the project management system through the identity authentication system of the project management system. Call the API interface and send the validated structured user story to the project management system based on the access credentials; Once the access credentials are verified by the project management system, the automatic login to the project management system is completed, and an end-user story is generated. The step of calling the API interface and sending the validated structured user story to the project management system based on the access credentials also includes: If the API call fails, a fault-tolerant switching mechanism is initiated, through which the validated structured user story is sent to the project management system.

[0118] If the API call fails, a fault-tolerant switching mechanism is initiated, through which the validated structured user story is sent to the project management system, including: Trigger the automated operation scripts of the project management system interface; Based on the interface automation operation script, switch to the interface automation operation process and perform a visual operation for user story entry to simulate manual operation by the user. Call the visual recognition plugin to verify the accuracy of the visual operation results; If the visualization operation result is verified to be accurate, the verified structured user story is sent to the project management system. If the visualization operation result verification is inaccurate, the fault tolerance switching mechanism will be re-executed until the visualization operation result verification is accurate.

[0119] In one embodiment, the management module 106 is specifically used for: By using the object storage compatibility protocol configured in the project management system, historical user stories, industry terms, and compliance requirements data are categorized and stored in the knowledge base to achieve hybrid storage of structured and unstructured data. By utilizing the automated operation process of the interface, historical data from the project management system is automatically collected, incremental updates are performed on the knowledge base, and a knowledge graph is constructed. The access control system, configured based on the project management system, controls access permissions to the knowledge base and manages and traces multiple versions of the knowledge base content through the version control function of object storage. Based on the knowledge base, an executable automated task process is formed, and a traceable execution log is output.

[0120] Specific limitations regarding the automatic task flow generation device 100 can be found in the limitations of the automatic task flow generation method described above, and will not be repeated here. Each module in the aforementioned automatic task flow generation device 100 can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0121] In one embodiment, a computer device 200 is provided, which may be a server, and its internal structure diagram may be as follows: Figure 8 As shown. The computer device 200 includes a processor 220, memory, and a network interface 250 connected via a system bus 210. The processor 220 provides computing and control capabilities. The memory of the computer device 200 includes non-volatile and / or volatile storage media and internal memory 240. The non-volatile storage media 230 stores an operating system 231, computer programs 232, and a database 233. The internal memory 240 provides an environment for the operation of the operating system and computer programs in the non-volatile storage media 230. The network interface 250 of the computer device 200 is used to communicate with external clients via a network connection. When the computer program is executed by the processor 220, it implements the functions or steps of a task flow automatic generation method server. That is, when the processor 220 executes the computer program, it implements the following steps: Acquire multimodal input information; Semantic elements of the multimodal input information are extracted, and these semantic elements are input into a preset story template. Domain knowledge is then injected into the story template to output a story framework. The domain knowledge is stored in a knowledge base and includes pre-collected industry terms and compliance requirements. The story framework is then structured to generate structured user stories; Perform compliance and integrity checks on the structured user stories; Based on the verified structured user story, a preset automatic access program is triggered to input the verified structured user story into the project management system to generate the final user story. Based on the end-user story, the knowledge base is updated to form an executable automated task flow.

[0122] In one embodiment, a computer device 300 is provided, which may be a client, and its internal structure diagram may be as follows: Figure 9As shown. The computer device includes a processor 320, memory, network interface 350, display screen 370, and input device 360 ​​connected via a system bus 310. The processor 320 provides computing and control capabilities. The memory includes a non-volatile storage medium 330 and internal memory 340. The non-volatile storage medium 330 stores an operating system 331 and a computer program 332. The internal memory provides an environment for the operation of the operating system 331 and the computer program 332 in the non-volatile storage medium 330. The network interface 350 of the computer device 300 is used for communication with an external server via a network connection. When the computer program is executed by the processor 320, it implements the functions or steps of a task flow automatic generation method on the client side. That is, when the processor 320 executes the computer program 332, it implements the following steps: Acquire multimodal input information; Semantic elements of the multimodal input information are extracted, and these semantic elements are input into a preset story template. Domain knowledge is then injected into the story template to output a story framework. The domain knowledge is stored in a knowledge base and includes pre-collected industry terms and compliance requirements. The story framework is then structured to generate structured user stories; Perform compliance and integrity checks on the structured user stories; Based on the verified structured user story, a preset automatic access program is triggered to input the verified structured user story into the project management system to generate the final user story. Based on the end-user story, the knowledge base is updated to form an executable automated task flow.

[0123] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: Acquire multimodal input information; Semantic elements of the multimodal input information are extracted, and these semantic elements are input into a preset story template. Domain knowledge is then injected into the story template to output a story framework. The domain knowledge is stored in a knowledge base and includes pre-collected industry terms and compliance requirements. The story framework is then structured to generate structured user stories; Perform compliance and integrity checks on the structured user stories; Based on the verified structured user story, a preset automatic access program is triggered to input the verified structured user story into the project management system to generate the final user story. Based on the end-user story, the knowledge base is updated to form an executable automated task flow.

[0124] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0125] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0126] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0127] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for automatically generating task flows, characterized in that, include: Acquire multimodal input information; Semantic elements of the multimodal input information are extracted, and these semantic elements are input into a preset story template. Domain knowledge is then injected into the story template to output a story framework. The domain knowledge is stored in a knowledge base and includes pre-collected industry terms and compliance requirements. The story framework is then structured to generate structured user stories; Perform compliance and integrity checks on the structured user stories; Based on the verified structured user story, a preset automatic access program is triggered to input the verified structured user story into the project management system to generate the final user story. Based on the end-user story, the knowledge base is updated to form an executable automated task flow.

2. The method for automatically generating task flows according to claim 1, characterized in that, The process of extracting semantic elements from the multimodal input information, inputting these semantic elements into a preset story template, and injecting domain knowledge into the story template to output a story framework includes: The input information is analyzed using a semantic understanding model to extract its natural language requirements and core semantic elements. The core semantic elements are input into a preset story template; wherein, the story template is a data structure for generating user stories, including fields, field types, and whether they are required. By calling the domain knowledge enhancement model through the model collaboration platform, the domain knowledge enhancement model, based on the knowledge base, injects industry terms and compliance requirements into the story framework by field to obtain a complete story framework; Based on the completed story framework, a logical relationship network among the core semantic elements is established, and a standardized story framework is generated and output.

3. The method for automatically generating task flows according to claim 2, characterized in that, The step of structuring the story framework to generate structured user stories includes: The standardized story framework is restricted in terms of field types and formats using a structured output plugin. The standardized story framework after output constraints is the structured user story.

4. The method for automatically generating task flows according to claim 2, characterized in that, The compliance and integrity verification of the structured user stories includes: By calling the classification model through the model collaboration platform, we can check whether the acceptance criteria for formatted user stories cover the core business logic. By calling the anomaly detection model through the model collaboration platform, compliance risks or logical vulnerabilities in formatted user stories can be identified. If both the classification model and the anomaly detection model fail the validation, a correction suggestion is output to regenerate the user story; If both the classification model and the anomaly detection model pass the verification, a qualified structured user story is output.

5. The method for automatically generating task flows according to claim 4, characterized in that, The process of triggering a preset automatic access procedure based on the verified structured user story to input the verified structured user story into the project management system to generate the final user story includes: Obtain the verification result of the structured user story. If the verification passes, obtain the access credentials for the project management system through the identity authentication system of the project management system. Call the API interface and send the validated structured user story to the project management system based on the access credentials; Once the access credentials are verified by the project management system, the automatic login to the project management system is completed, and an end-user story is generated. The step of calling the API interface and sending the validated structured user story to the project management system based on the access credentials also includes: If the API call fails, a fault-tolerant switching mechanism is initiated, through which the validated structured user story is sent to the project management system.

6. The method for automatically generating task flows according to claim 5, characterized in that, If the API call fails, a fault-tolerant switching mechanism is initiated. This mechanism sends the validated structured user story to the project management system, including: Trigger the automated operation scripts of the project management system interface; Based on the interface automation operation script, switch to the interface automation operation process and perform a visual operation for user story entry to simulate manual operation by the user. Call the visual recognition plugin to verify the accuracy of the visual operation results; If the visualization operation result is verified to be accurate, the verified structured user story is sent to the project management system. If the visualization operation result verification is inaccurate, the fault tolerance switching mechanism will be re-executed until the visualization operation result verification is accurate.

7. The method for automatically generating task flows according to claim 6, characterized in that, The process of updating the knowledge base based on the end-user story to form an executable automated task flow includes: By using the object storage compatibility protocol configured in the project management system, historical user stories, industry terms, and compliance requirements data are categorized and stored in the knowledge base to achieve hybrid storage of structured and unstructured data. By utilizing the automated operation process of the interface, historical data from the project management system is automatically collected, incremental updates are performed on the knowledge base, and a knowledge graph is constructed. The access control system, configured based on the project management system, controls access permissions to the knowledge base and manages and traces multiple versions of the knowledge base content through the version control function of object storage. Based on the knowledge base, an executable automated task process is formed, and a traceable execution log is output.

8. A task flow automatic generation device, characterized in that, include: The acquisition module is used to acquire multimodal input information; An extraction module is used to extract semantic elements from the multimodal input information, input the semantic elements into a preset story template, and inject domain knowledge into the story template to output a story framework; wherein, the domain knowledge is stored in a knowledge base, and the domain knowledge includes pre-collected industry terms and compliance requirements; A structuring module is used to perform structuring processing on the story framework to generate structured user stories; The verification module is used to perform compliance and integrity verification on the structured user stories; The generation module is used to trigger a preset automatic access program based on the verified structured user story, and input the verified structured user story into the project management system to generate the final user story. The management module is used to update the knowledge base based on the end-user stories, forming an executable automated task flow.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the task flow automatic generation method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the task flow automatic generation method as described in any one of claims 1 to 7.

Images