Tool configuration method and apparatus, computer device, storage medium, and program product

Abstract

This application relates to a tool configuration method and apparatus, computer equipment, storage medium, and program product. The tool configuration method includes: determining node tools for a target node, where the target node is a node in a target workflow, the target workflow is a workflow used to execute a target task, the target node indicates a phased subtask of the target task, and the node tools are tools required during the execution of the phased subtasks. The node tools are determined from workflow tools based on the node configuration information of the target node, and the workflow tools are tools to be invoked during the execution of the target task; based on the node tools, determining a target toolset, where the tools in the target toolset are tools to be invoked during the execution of the phased subtasks. By configuring the callable tools during the execution of phased subtasks using this tool configuration method, it is beneficial to more accurately invoke tools, thereby improving the accuracy of target task execution.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a tool configuration method and apparatus, computer equipment, storage medium and program product. Background Technology

[0002] With the innovation and development of artificial intelligence (AI) technology, it is increasingly being applied to various fields such as code generation, intelligent customer service, data analysis, and automated operation and maintenance. Furthermore, the application of AI technology is evolving from simple chatbots to autonomous systems capable of performing complex tasks. As the application of AI technology becomes more widespread and users' demands for it increase, AI development frameworks are being optimized. How to improve the accuracy of using AI technology to perform tasks is a problem that requires continuous research in the application development of AI technology. Summary of the Invention

[0003] Therefore, it is necessary to provide a tool configuration method and apparatus, computer equipment, storage medium, and program product to address the aforementioned technical problems. Configuring the callable tools during the execution of phased sub-tasks facilitates more accurate tool invocation, thereby improving the accuracy of target task execution.

[0004] Firstly, this application provides a tool configuration method, the method comprising: The node tool for determining the target node is a node in the target workflow, which is a workflow used to execute the target task. The target node indicates a phased subtask of the target task. The node tool is a tool required in the process of executing the phased subtask. The node tool is determined from the workflow tool based on the node configuration information of the target node. The workflow tool is a tool that is called during the execution of the target task. Based on the node tools, a target toolset is determined, wherein the tools in the target toolset are those to be invoked during the execution of the phased sub-tasks.

[0005] Using the tool configuration method provided in the first aspect, the node tools of the target node are first determined. The target node is a node in the target workflow that executes the target task. The target node indicates a phased subtask of the target task. The workflow tool is the tool invoked during the execution of the target task. Based on the node configuration information of the target node, the node tools required for executing the phased subtask are determined from the workflow tools. Furthermore, a target toolset is determined based on the node tools to configure the tools that can be invoked during the execution of the phased subtask. This tool configuration method ensures that all tools in the target toolset are required for executing the phased subtask, that is, that all tools invoked during the execution of the phased subtask are the necessary tools. It effectively avoids the target toolset from containing workflow tools unrelated to the execution of the phased subtask. Furthermore, it effectively avoids invoking the wrong tools to execute the phased subtask and effectively avoids tool clutter interfering with the execution of the phased subtask and the target task. This tool configuration method facilitates more accurate invocation of tools from the target toolset, leading to more accurate execution of the phased subtask and the target task.

[0006] In one embodiment, prior to the node tool for determining the target node, the method further includes: Obtain the node metadata of the target node, wherein the node metadata is a collection of information configuring the execution process of the phased subtask, and the node metadata includes at least the information of the node tool; The node tool for determining the target node includes: Based on the information from the node tool, the node tool of the target node is determined.

[0007] In one embodiment, the method further includes: The tool constraint information of the target node is determined. The tool constraint information is used to indicate whether there is permission to call file system tools during the execution of the phased subtask. The file system tools are used to access or manage data files. The step of determining the target toolset based on the node tools includes: If the tool constraint information of the target node indicates that it has permission to invoke the file system tools, then the target toolset is determined based on the node tools and the file system tools; or If the tool constraint information of the target node indicates that it does not have permission to call the file system tools, then the target toolset is determined based on the node tools.

[0008] In one embodiment, the node metadata further includes information about dedicated tools, which are tools invoked during the execution of the phased subtasks. These dedicated tools are used to execute control instructions output by callback functions or to execute user-approved items included in the execution of the phased subtasks. The callback functions are used to control the execution process of the phased subtasks. The method further includes: Based on the information of the dedicated tool, determine the dedicated tool for the target node; The step of determining the target toolset based on the node tools includes: The target toolset is determined based on the node tools and the dedicated tools.

[0009] In one embodiment, the information of the dedicated tool includes approval information and visualization information. The approval information is used to indicate whether user approval is required when calling the dedicated tool to perform a task. The visualization information is used to indicate that if user approval is required, the process of calling the dedicated tool to perform the task will be displayed visually.

[0010] In one embodiment, the node metadata further includes at least one of interaction information and subtask target information; The interaction information is used to indicate whether the target node is an interactive node. If the execution of the phased subtask is interrupted when the target node is not an interactive node, the execution is resumed. If the execution of the phased subtask is interrupted when the target node is an interactive node, a user interaction instruction is received. The user interaction instruction is associated with the execution process of the phased subtask. The subtask target is used to guide the execution process of the phased subtask.

[0011] In one embodiment, the method further includes: Determine the tool configuration information for each tool in the target toolset, wherein the tool configuration information is used to indicate the function and usage of the corresponding tool; Determine the sub-task objectives of the target node, wherein the sub-task objectives are used to guide the execution process of the phased sub-tasks; Based on the tool configuration information of each tool and the sub-task objective, a target tool is determined from the target tool set, and the function and usage of the target tool are matched with the sub-task objective.

[0012] In one embodiment, before determining the target tool from the target tool set based on the tool configuration information of each tool and the sub-task objective, the method further includes: Determine context information and / or user profile; the context information includes at least one of the following: device type, task objective of the target workflow, the task objective being used to guide the execution process of the target task; The step of determining the target tool from the target tool set based on the tool configuration information of each tool and the sub-task objective includes: The target tool is determined from the target toolset based on the tool configuration information of each tool, the sub-task objective, the context information, and / or the user profile.

[0013] Secondly, this application also provides a tool configuration apparatus, the apparatus comprising: A processing unit is used to determine node tools for a target node, wherein the target node is a node in a target workflow, the target workflow is a workflow used to execute a target task, the target node indicates a phased subtask of the target task, the node tool is a tool required in the process of executing the phased subtask, the node tool is determined from a workflow tool based on the node configuration information of the target node, and the workflow tool is a tool to be called during the execution of the target task; based on the node tool, a target toolset is determined, wherein the tools in the target toolset are tools to be called during the execution of the phased subtask.

[0014] Thirdly, this application also provides a tool configuration method, the method comprising: A node tool for determining a target node, wherein the target node is a node in a target workflow, the target workflow is a workflow used to execute code generation tasks, the target node indicates a code release subtask in the code generation task, the node tool is a tool required in the process of executing the code release subtask, the node tool is determined from a workflow tool based on the node configuration information of the target node, and the workflow tool is a tool to be called during the execution of the code generation task; Based on the node tools, a target toolset is determined, wherein the tools in the target toolset are those to be invoked during the execution of the code release subtask.

[0015] Fourthly, this application also provides a tool configuration apparatus, the apparatus comprising: A processing unit is used to determine node tools for a target node, wherein the target node is a node in a target workflow, the target workflow is a workflow for executing a code generation task, the target node indicates a code release subtask in the code generation task, the node tool is a tool required in the process of executing the code release subtask, the node tool is determined from a workflow tool based on the node configuration information of the target node, and the workflow tool is a tool to be invoked during the execution of the code generation task; based on the node tool, a target toolset is determined, wherein the tools in the target toolset are tools to be invoked during the execution of the code release subtask.

[0016] Fifthly, this application also provides a computer device, including: a memory and a processor, wherein the memory stores computer program instructions; when the computer program instructions are executed by the processor, the processor causes the processor to perform the method as described in the first aspect or any embodiment of the first aspect, or to perform the method as described in the third aspect or any embodiment of the third aspect.

[0017] Sixthly, this application also provides a computer-readable storage medium storing computer program instructions; when the computer program instructions are executed on one or more processors, they perform the method as described in the first aspect or any embodiment of the first aspect, or perform the method as described in the third aspect or any embodiment of the third aspect.

[0018] In a seventh aspect, this application also provides a computer program product comprising computer program instructions; when the computer program instructions are executed on a computer, the computer causes the computer to perform the method as described in the first aspect or any embodiment of the first aspect, or to perform the method as described in the third aspect or any embodiment of the third aspect.

[0019] The beneficial effects that can be achieved by the tool configuration apparatus provided in the second aspect, the tool configuration method provided in the third aspect, the tool configuration apparatus provided in the fourth aspect, the computer equipment provided in the fifth aspect, the computer-readable storage medium provided in the sixth aspect, and the computer program product provided in the seventh aspect can refer to the beneficial effects of the tool configuration method provided in the first aspect. Attached Figure Description

[0020] To clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the embodiments will be described below.

[0021] Figure 1 A schematic diagram of the intelligent agent framework provided in the embodiments of this application; Figure 2A flowchart illustrating the tool configuration method provided in the embodiments of this application; Figure 3 A flowchart illustrating the tool configuration method provided in the embodiments of this application; Figure 4 This is a schematic diagram of the structure of the tool configuration device provided in the embodiments of this application; Figure 5 This is a schematic diagram of the structure of the tool configuration device provided in the embodiments of this application; Figure 6 A schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0022] To facilitate understanding of the embodiments of this application, a more comprehensive description of the embodiments of this application will be provided below with reference to the accompanying drawings. The drawings illustrate preferred embodiments of the embodiments of this application. However, the embodiments of this application can be implemented in many different forms and are not limited to the embodiments described herein. These embodiments are provided to make the disclosure of the embodiments of this application more thorough and comprehensive.

[0023] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which embodiments of this application pertain.

[0024] The terms “first,” “second,” etc., used in this document are used to distinguish different objects, not to describe a specific order. It should also be understood that the term “including / contains” means the presence of the stated features, wholes, steps, operations, parts, or combinations thereof, but does not exclude the possibility of the presence of one or more other features, wholes, steps, operations, parts, or combinations thereof.

[0025] With the innovation and development of artificial intelligence (AI) technology, AI is increasingly being applied to many fields such as code generation, intelligent customer service, data analysis, and automated operation and maintenance. Furthermore, the application of AI is evolving from simple chatbots to autonomous systems capable of completing complex tasks. As AI technology becomes more widespread and users demand higher functionality from it, AI development frameworks are being optimized. However, traditional AI development frameworks, such as LangGraph and AutoGen, adopt a "workflow engineering" design philosophy. They employ a linear workflow concatenation model, focusing on automating business processes and orchestrating and collaborating complete business processes to achieve end-to-end business goals. However, their control granularity is relatively coarse, making it difficult to perform fine-grained control or intervention during the execution of phased sub-tasks using AI technology. In this application, the AI ​​entity with autonomous decision-making capabilities, constructed using an AI development framework, is referred to as an agent. Therefore, the AI ​​development framework can also be called an agent development framework. Utilizing AI technology to perform tasks can also be called utilizing agents to perform tasks. This application proposes a finite state machine-based intelligent agent development framework (hereinafter referred to as "this intelligent agent framework") that adopts the "State Engineering" design concept. The tasks executed by the intelligent agent include several phased sub-tasks, each of which is defined as a state of the finite state machine. This intelligent agent framework focuses on the transitions between states of the finite state machine and the states themselves, enabling fine-grained control and intervention over each state. Specifically, while enabling the intelligent agent to make autonomous decisions within the state machine framework, it provides a full-cycle, multi-level callback function mechanism from the state machine level to the state level, supporting programmable intervention throughout the entire execution process. It also provides a complete user approval mechanism, supporting user approval intervention for states. Employing a state machine-driven loop architecture, while the intelligent agent makes its own decisions, it achieves fine-grained control over each state through the callback function mechanism and the user approval mechanism, and then the intelligent agent provides feedback on the task execution results, thus forming a closed-loop control. This intelligent agent framework based on finite state machines not only achieves finer-grained control but also offers stronger predictability and programmability.

[0026] For example, Figure 1 An intelligent agent framework is illustrated. For example... Figure 1As shown, in this intelligent agent framework, the first module receives task instructions from the user interaction layer and distributes tasks in response to these instructions. Specifically, the task is divided into phased subtasks, and different phased subtasks are distributed to the corresponding second modules for execution. There is a one-to-one correspondence between the second module, the phased subtasks, and the state. The second module is configured with callback functions, models (LargeLanguageModel, LLM), tools that can be invoked to achieve the subtask objectives of the corresponding phased subtasks, and information such as whether user interaction is required. The external service module can provide external service calls to the execution process of the phased subtasks. The external service module includes at least a model provisioning unit (LLMProvide) and a tool registration unit (ToolRegistry). For example, data flows in from the user interaction layer, is processed by the core engine of this intelligent agent framework, and then calls external services.

[0027] The entire execution process of a task can be viewed as a workflow, with each stage of the task acting as a node within that workflow. The finite state machine described above corresponds to the workflow, and the states correspond to the nodes. Each state during task execution is a node. For ease of description and understanding, the following text will use the concepts of workflow and nodes to describe the task's execution process and its stage-specific subtasks.

[0028] This application also provides a tool configuration method to configure tools that can be invoked by an agent built based on this agent framework in a target node. This tool configuration method helps the agent more accurately determine and invoke the required tools. Furthermore, it allows the agent to perform tasks more accurately using more precise tools. It should be understood that the agent built using this agent framework is a software system that runs and executes tasks on a computer device. This tool configuration method is executed by the computer device running the agent; the behavior of the agent described herein should be understood as the behavior performed by the computer device through running the agent.

[0029] Before providing a detailed explanation of the configuration method for this tool, the following explanation will clarify some of the terms used in the embodiments of this application.

[0030] An intelligent agent is an artificial intelligence (AI) entity built using an AI development framework, possessing autonomous decision-making capabilities. It can understand the goals of a task and execute it through model invocation, tool invocation, and other means.

[0031] State Engineering is an AI application development paradigm centered on state management, distinct from traditional workflow engineering.

[0032] State machine: A framework for controlling task execution. The execution of a phased subtask can be determined based on the state of the state machine.

[0033] Workflow. Describes the phased process by which an agent completes a task. It includes multiple nodes, each node corresponding to a phased subtask, representing a stage of task execution. A finite state machine with multiple states can be viewed as a workflow with multiple nodes.

[0034] Callback functions. Callback functions can be triggered at specific stages of a task's execution by the agent. With developer intervention and configuration, they are used to guide and control the agent's behavior, enabling fine-grained control over workflows and nodes.

[0035] Large Language Model (LLM). A core AI model used for agent reasoning and decision-making, referred to simply as "the model" in this paper.

[0036] Tools. Capability units with standardized interfaces and clear descriptions that can be invoked by intelligent agents to perform specific operations. Intelligent agents can determine how to invoke tools based on standardized interfaces, and when to invoke tools based on clear descriptions of functions, scenarios, parameters, and other information. As capability units, tools can be combined or decoupled by intelligent agents for use.

[0037] like Figure 2 As shown, the tool configuration method in this application embodiment includes the following steps S101 to S102.

[0038] S101, Determine the node tool for the target node. The target node is a node in the target workflow, which is a workflow used to execute the target task. The target node indicates the phased subtasks of the target task. The node tool is the tool required in the process of executing the phased subtasks. The node tool is determined from the workflow tool based on the node configuration information of the target node. The workflow tool is a tool that is called during the execution of the target task.

[0039] For example, using this agent framework, a feasible processing phase of the target node, i.e., a feasible execution process of a phased subtask, is as follows. The processing phase of the target phase includes an entry phase, an update phase, and an exit phase. Entering the target node can trigger the invocation of an entry callback function. Based on the preset configuration for the entry callback function, the model invocation instruction and tool invocation instruction are output through the entry callback function. Optionally, a preset configuration is performed according to the subtask objective of the phased subtask. Then, entering the update phase executes the model invocation instruction and the tool invocation instruction. Here, the update phase is the execution phase. Specifically, the model is invoked, and the target tool is determined from the target toolset using the model. The target tool is used to execute the items in the phased subtask to execute the phased subtask. Then, entering the exit phase, the target node can be exited by triggering the invocation of the exit callback function, and the next node can be entered. Referring to this example, this tool configuration method is used to determine the target toolset, i.e., to configure the tools that the agent can invoke in the target node.

[0040] The tools in this application embodiment can be understood as capability units with standardized interfaces and clear descriptions that can be invoked by intelligent agents to perform specific operations. Workflow tools are tools invoked during the execution of the target task. Specifically, workflow tools are determined through registration via ToolRegistry and are tools that can be invoked by the intelligent agent throughout the entire lifecycle of executing the target task and are shared across nodes. To determine the target toolset, the node tools of the target node are first determined from the workflow tools based on the node configuration information of the target node. Node tools are tools required during the execution of phased sub-tasks. For example, the node configuration information of the target node is information preset in the intelligent agent framework or information obtained at the current moment. The intelligent agent loads and obtains the node tools of the target node based on the node configuration information of the target node.

[0041] Intelligent agents are applied in many fields such as code generation, intelligent customer service, data analysis, and automated operation and maintenance. For ease of understanding, this application uses code generation as an example to illustrate the concept. Optionally, the code generation task can be broken down into multiple phased sub-tasks, such as generating the system architecture to be generated, generating code based on the system architecture, and code deployment. Taking the phased sub-task corresponding to the target node as the code deployment sub-task as an example, the node tools include, for example, testing tools and packaging tools.

[0042] In one embodiment, before determining the node tool for the target node in step S101 above, the tool configuration method further includes the following steps: Obtain the node metadata of the target node. Node metadata is a collection of information configuring the execution process of phased subtasks, and it includes at least information about the node tools.

[0043] In this embodiment, step S101, determining the node tool for the target node, includes the following steps: Based on the information from the node tool, determine the node tool for the target node.

[0044] Node metadata is a collection of information describing the node attributes, configuration, and execution process of the phased subtasks of the target node. Node metadata is determined based on the subtask objectives of the phased subtasks. In this embodiment, the key function of node metadata is to configure the execution process of the phased subtasks. The node metadata of the target node is an example of the node configuration information in the aforementioned embodiments. Node metadata at least includes information about node tools, which includes at least the name of the node tool and may also include at least one of the following: the function, usage method, and configuration parameters of the node tool. The node tool information indicates the workflow tools required during the execution of the phased subtasks, i.e., the workflow tools required to achieve the subtask objectives of the phased subtasks. Optionally, the node metadata of the target node is information pre-set in this agent framework, i.e., information obtained by the agent before executing the target task. Optionally, the node metadata of the target node is information obtained by the agent during the execution of the target task. The agent can load the node tools of the target node based on the node tool information.

[0045] For example, the node metadata of the target node is obtained. Node metadata is a collection of information configuring the execution process of the task publishing subtask, and it includes at least information about the node tools. The node tool information indicates the workflow tools required during the task publishing subtask process. Optionally, the node tool information is determined based on the subtask objective of the task publishing subtask. That is, the required workflow tools for this subtask are determined based on its subtask objective, and then the required workflow tools are configured using the node tool information. Based on the node tool information, the node tools required to execute the task publishing subtask can be determined.

[0046] S102, Based on the node tools, determine the target toolset. The tools in the target toolset are those that will be invoked during the execution of phased sub-tasks.

[0047] Based on the node configuration information of the target node, exemplarily, the node tools required for executing phased subtasks are determined from the workflow tools based on the node tool information. Then, a target toolset is determined based on the node tools to configure the tools that can be invoked during the execution of phased subtasks. This tool configuration method ensures that all tools in the target toolset are required for executing phased subtasks, meaning that all tools invoked during the execution of phased subtasks are necessary. It effectively avoids the target toolset containing workflow tools unrelated to the execution of phased subtasks. Furthermore, it effectively avoids invoking incorrect tools to execute phased subtasks and prevents tool clutter from interfering with the execution of phased subtasks and the target task. This tool configuration method facilitates more accurate invocation of tools from the target toolset, leading to more accurate execution of phased subtasks and the target task.

[0048] In one embodiment, before determining the target toolset based on the node tools in step S102, the tool configuration method further includes the following steps: Determine the tool constraint information for the target node. Tool constraint information indicates whether there is permission to invoke file system tools during the execution of phased subtasks. File system tools are used to access or manage data files.

[0049] Access includes at least one of reading and writing. Accessing or managing data files includes actions such as reading, writing, editing, deleting, and moving data files. File system tools are built into the agent. For example, with permission to invoke file system tools, the agent can integrate and use the functionality of a memory file system (MemFS) to access or manage data files in memory. Version control is also supported, allowing the agent to perform operations such as rollback, snapshots, and comparisons on data files. In this example, granting the agent permission to use the memory file system tools enables it to process temporary data files in memory, providing a file operation interface and mitigating the risk of file tampering. Tool constraint information indicates whether permission to invoke file system tools is available during the execution of phased subtasks. Optionally, tool constraint information is expressed as having or not having invocation permission. Optionally, the agent has invocation permission by default; configuring tool constraint information can prevent the agent from having invocation permission. For example, the node metadata of the target node also includes tool constraint information, and the agent determines the tool constraint information of the target node from the node metadata.

[0050] In this embodiment, step S102, determining the target toolset based on the node tools, includes the following steps: If the target node's tool constraint information indicates that it has permission to invoke file system tools, then determine the target toolset based on the node tools and file system tools; or If the tool constraint information of the target node indicates that it does not have permission to call file system tools, then the target toolset is determined based on the node tools.

[0051] Specifically, when the agent has permission to invoke file system tools, the target toolset is dynamically configured based on the node tools and file system tools. The target toolset includes both node tools and file system tools. When executing phased subtasks, the agent can invoke file system tools to access or manage data files, which is beneficial for efficiently executing tasks by reading information already recorded in the data files. When the agent lacks permission to invoke file system tools, the target toolset does not include file system tools. When executing phased subtasks, the agent cannot access or manage data files, which helps to effectively avoid data leakage, resource consumption, and other unforeseen risks.

[0052] In one embodiment, the node metadata also includes information about dedicated tools. These dedicated tools are invoked during the execution of phased subtasks. They are used to execute control commands output by callback functions or to handle user-approved matters included in the execution of phased subtasks. The callback functions are used to control the execution process of the phased subtasks. If the tool configuration method includes the step of obtaining the node metadata of the target node, before determining the target toolset based on the node tools in step S102, the tool configuration method further includes the following steps: Based on the information from the dedicated tools, determine the dedicated tools for the target node.

[0053] In this embodiment, step S102, determining the target toolset based on the node tools, includes the following steps: Determine the target toolset based on node tools and dedicated tools.

[0054] As described above, this agent framework enables agents to make autonomous decisions within a state machine framework while also providing a multi-layered callback function mechanism throughout the entire lifecycle, from the state machine level to the state level, supporting programmable intervention throughout the entire execution process. With developer intervention and configuration, callback functions can be triggered at specific stages of the agent's task execution to guide and control the agent's behavior, achieving fine-grained control over workflows and nodes. For example, an entry callback function is triggered when entering a target node, outputting a model invocation command to control the agent to invoke the model. Similarly, an exit callback function is triggered upon completion of the phased subtask corresponding to the target node, outputting a jump command to control the agent to enter the next node.

[0055] Unlike node tools determined from workflow nodes, dedicated tools, configured directly through node metadata, can be used to execute control commands output by callback functions. For example, the information of a dedicated tool includes at least its name and executable operations. For instance, dedicated tool A's executable operations include jump instructions. Therefore, in response to the jump instruction output by the callback function, dedicated tool A is invoked to execute node jump. Dedicated tools can be used to execute user-approved items included in the execution of phased subtasks. For example, the information of a dedicated tool includes at least its name, approval information (need approval), and visual information (hint). The approval information indicates whether user approval is required when calling the dedicated tool to execute an item; the visual information indicates that the process of calling the dedicated tool to execute an item will be visualized if user approval is required. For example, dedicated tool B's function is information publishing, and the approval information of dedicated tool B indicates that user approval is required when calling dedicated tool B to execute an item. In the case where the phased subtask is a code publishing subtask, and user approval of the code to be published is required before publication, dedicated tool B is invoked to execute the publishing item in the code publishing subtask. Furthermore, the process of executing release requests is visualized for user review and approval. During the approval process, users can check the rationality of the code to be released and the release flow before granting approval, thus authorizing the release. For example, the information of a dedicated tool includes its name, approval information (needapproval), visual information (hint), executable operations (execute), functional description (description), and tool parameters (parameters). The functional description indicates the tool's function and purpose, while the tool parameters indicate the parameters required to invoke the tool. By configuring and describing the dedicated tool in detail using the above information, its functions and attributes are clearly defined, facilitating accurate and efficient invocation of the required tools by the agent. For example, dedicated tools can be used to execute control instructions output by callback functions and to handle user-approved matters included in the execution of phased subtasks.

[0056] Having obtained the node metadata of the target node, the agent can load the target node's dedicated tools based on the dedicated tool information in the node metadata. In this embodiment, the target toolset is dynamically configured based on the dedicated tools and node tools. This ensures the target toolset includes both dedicated and node tools, allowing the agent to not only call the node tools registered through ToolRegistry but also the dedicated tools to execute control commands output by callback functions or perform tasks requiring user approval. This facilitates the effective application of the agent framework including the callback function mechanism, helps the agent meet task requirements requiring user approval, and enables the agent to execute more complex, phased subtasks.

[0057] For example, Figure 3 A flowchart illustrating the tool configuration method is provided. In this example, after obtaining the node metadata of the target node, which includes at least information about the node tools and the dedicated tools, the agent, upon entering the target node, loads the node tools and dedicated tools of the target node based on the node metadata. Furthermore, the agent determines the tool constraint information of the target node to ascertain whether it has permission to call file system tools. If so, file system tools are added; otherwise, they are skipped. Subsequently, a target toolset for the target node is constructed, which includes at least node tools and dedicated tools, and may also include file system tools. Figure 3 The sequence of loading node tools, loading dedicated tools, and determining permission to call file system tools shown is merely an example. There is no fixed order requirement between these three actions; optionally, they can be performed simultaneously. Exemplarily, the agent uses the model to determine target tools from the target toolset, and then uses these target tools to perform tasks within the phased subtask. The target toolset is the configuration information used by the model to determine the target tools. This tool configuration method in the example helps ensure that the target node is configured with only the necessary tools, avoids incorrect tool calls due to an excessive or complex set of tools, and prevents interference with the agent's execution of both the phased subtask and the target task.

[0058] In one embodiment, the node metadata also includes at least one of interaction information and subtask target information; Interaction information is used to indicate whether the target node is an interactive node. If the execution of a phased subtask is interrupted when the target node is not an interactive node, execution will resume. If the execution of a phased subtask is interrupted when the target node is an interactive node, user interaction instructions will be received. User interaction instructions are associated with the execution process of the phased subtask. The subtask target is used to guide the execution process of the phased subtask.

[0059] The interactive information indicates whether the target node is an interactive node, with "0" representing no and "1" representing yes. When the interactive information indicates the target node is an interactive node, it signifies that the execution of the phased subtask involves an interruption to allow interaction with the user. Therefore, if the execution of a phased subtask is interrupted, the agent waits for and receives user interaction commands. Configuring interactive information in the node metadata helps ensure the user's approval rights over the agent's execution of phased subtasks. User approval facilitates fine-grained control over the execution process of phased subtasks and the agent's task execution. User interaction commands are associated with the execution process of the phased subtask. For example, if the target node corresponding to the code release subtask is an interactive node, then in the event of an interruption in the code release subtask's execution, the agent waits for and receives user interaction commands. For example, user interaction commands are used to authorize the release of the generated code or to indicate that the generated code has failed approval.

[0060] When the interaction information indicates that the target node is not an interactive node, it signifies that the execution of the phased subtask does not involve any interruption to allow for interaction with the user. Therefore, if the execution of a phased subtask is interrupted, the agent automatically resumes its execution. Furthermore, by configuring interaction information in the node metadata, automatic recovery from execution interruptions can be achieved, improving the agent's task execution efficiency.

[0061] The subtask goal information includes at least the specific content of the subtask goal. The subtask goal of the target node can be dynamically established as a function or statically configured. The subtask goal serves as the behavioral guideline and decision-making basis for the agent during the execution of phased subtasks, precisely guiding the execution process of the phased subtasks. Configuring interaction information in the node metadata helps ensure the accurate execution of phased subtasks.

[0062] For example, node metadata includes information such as the target node's name, subtask goal, node tool, action, interaction, node model, target, end, trailing prompt, and callback function. The descriptions of the subtask goal, tool, action, and interaction information can be found in the preceding embodiments and will not be repeated here. The name in the node metadata indicates the name of the phased subtask corresponding to the target node, such as a task release subtask. The node model refers to the model required during the execution of the phased subtask, and is also the model that can be invoked during the execution of the phased subtask. The node model information includes at least the model's name. By configuring the node model information in the node metadata, the agent can invoke the required model to execute the phased subtask. For example, the node model information also includes a functional description of the model; the agent determines the model to invoke based on the subtask goal and the functional descriptions of each model. Then, the corresponding model is called from the external service module. Jumpable target information indicates the node from which the agent can jump, indicating the agent's next stage subtask. Last information indicates whether the target node is the last node in the target workflow. The agent can determine its next action based on the last information. Tail prompts are prompts located at the end of the message. The information for tail prompts includes at least the specific content of the tail prompt. Distinguishing it from prompts in other positions, placing the prompt at the end of the message ensures that the agent can see it at its focal point each time it reads or views the message. Configuring tail prompts helps improve the agent's attention to these prompts. Callback function information includes at least the names of the callback functions that the agent can trigger during the execution of stage subtasks. For example, if the subtask goal of the code release subtask is to verify and release code, the following callback functions can be configured in the node metadata: entry callback function (onEnter), exit callback function (onExit), update callback function (onUpdate), and repair callback function (onError). For example, entering a target node triggers the invocation of an entry callback function, and exiting a target node triggers the invocation of an exit callback function. During execution, code validation can trigger the invocation of an update callback function, and if model calls or code tests fail, a repair callback function can be invoked.

[0063] In one embodiment, after determining the target toolset according to the method in any of the foregoing embodiments, the tool configuration method further includes the following steps S201 to S203.

[0064] S201, Determine the tool configuration information for each tool in the target toolset. The tool configuration information is used to indicate the function and usage of the corresponding tool.

[0065] Optionally, the tool configuration information of a node tool can be configured through node metadata, meaning that when configuring node tool information, the node tool information includes the tool configuration information. Alternatively, the tool configuration information of a node tool can be configured when registering a workflow tool through the ToolRegistry. The tool configuration information of a dedicated tool is configured through node metadata, and the dedicated tool information includes the dedicated tool's own tool configuration information. For example, the dedicated tool's tool configuration information includes its needapproval information, hint information, executeable operations, description, and parameters. The agent determines the tool configuration information of each tool from the node metadata or relevant data from tool registration to obtain the function and usage of each tool.

[0066] S202, Determine the sub-task objectives of the target node. Sub-task objectives are used to guide the execution process of phased sub-tasks.

[0067] Specifically, the agent determines the sub-task objectives of the target node from the node's metadata. A description of the sub-task objectives can be found in the preceding embodiments.

[0068] S203. Based on the tool configuration information of each tool and the sub-task objectives, determine the target tools from the target tool set. The functions and usage methods of the target tools are matched with the sub-task objectives.

[0069] Specifically, the agent identifies the intent of each stage of the sub-task based on its sub-task objective. For example, if the sub-task objective of the code release sub-task is to verify and release the code, the agent can identify that the intent of this stage of the sub-task is to test the generated code and release it. Based on this, the agent determines the testing tool and the release tool from the target toolset according to the functions and usage of each tool. The testing tool and the release tool can be used to achieve the sub-task objective of verifying and releasing the code.

[0070] For example, when the agent enters the update phase of the target node, it executes the model invocation instruction and tool invocation instruction output by the callback function. Specifically, it invokes the model and uses the model to determine the target tool from the target tool set, and uses the target tool to perform the tasks in the phased sub-task. The model determines the target tool from the target tool set based on the tool configuration information of each tool in the target tool set and the sub-task objectives.

[0071] In this embodiment, matching the tool configuration information of each tool in the target toolset with the sub-task objectives helps the agent to more accurately determine the target tools from the required target toolset, which is beneficial for the agent to make accurate decisions.

[0072] In one embodiment, before determining the target tool from the target tool set based on the tool configuration information of each tool and the sub-task objective in step S203, the tool configuration method further includes the following steps: Determine the context information and / or user profile. The context information includes at least one of the following: the task objective of the target workflow and the device type. The task objective of the target workflow is the objective of the target task, used to guide the execution process of the target task.

[0073] In this embodiment, step S203, determining the target tool from the target tool set based on the tool configuration information and sub-task objectives of each tool, includes the following steps: Target tools are determined from the target toolset based on the tool configuration information, sub-task objectives, context information, and / or user profiles of each tool.

[0074] Specifically, the task objective serves as the behavioral guideline and decision-making basis for the intelligent agent during the execution of the target task, precisely guiding the execution process. Taking code generation as an example, the task objective refers to the overall goal of the code generation task. For instance, if the user interaction layer inputs "Please generate easily understandable code," the task objective could be to generate and publish easily understandable code. By gaining a deeper understanding of the intent of the phased sub-tasks based on the sub-task objectives "validate and publish code" and the task objective "generate and publish easily understandable code," the intelligent agent can identify not only the intent to test and publish the generated code but also the intent to simplify or parse the code. Based on this, the intelligent agent determines testing tools, publishing tools, and annotation tools from the target toolset according to the functions and usage of each tool. Then, it executes the phased sub-tasks according to the determined target tools to publish correct, annotated, and parsed code. This improves the alignment between the task execution results and user needs, enhancing the user experience.

[0075] Device type refers to the type of computer device running the intelligent agent, such as mobile phones, watches, and desktop computers. Different device types have different screen sizes. For example, based on the sub-task objective "verify and release code," the intelligent agent identifies the intent of the phased sub-task and determines to call the testing and release tools. Simultaneously, based on the mobile phone device type, it determines to call the line-breaking tool. The line-breaking tool edits the code, displaying it as multi-line short sentences to fit the mobile phone screen size, making it easier for users to view and understand. For example, based on the user profile, the agent marks the user's preference for viewing task results presented in short sentence format based on historically executed tasks. Based on this, the intelligent agent also determines to call the line-breaking tool from the target toolset to display the code as multi-line short sentences.

[0076] In this embodiment, in addition to the tool configuration information and sub-task objectives of each tool, the target tool is also determined from the target tools based on context information and / or user profiles to execute the phased sub-tasks and target tasks. This not only helps the agent accurately determine the target tool and execute tasks, but also improves the fit between task execution results and user needs, thus enhancing the user experience.

[0077] This application also provides a tool configuration method, which includes the following steps: The node tool determines the target node, which is a node in the target workflow. The target workflow is a workflow used to execute code generation tasks. The target node indicates the code release subtask in the code generation task. The node tool is the tool required in the process of executing the code release subtask. The node tool is determined from the workflow tool based on the node configuration information of the target node. The workflow tool is a tool that is called during the execution of the code generation task. Based on the node tools, determine the target toolset. The tools in the target toolset are those that are called during the execution of the code release subtask.

[0078] The tool configuration method in this embodiment indicates that the tool configuration method in the foregoing embodiments is applied to a scenario where the target task is a code generation task and the phased subtask is a code release subtask. For a discussion of this tool configuration method, please refer to the examples in the foregoing embodiments where the target task is a code generation task and the phased subtask is a code release subtask; further details will not be repeated here.

[0079] like Figure 4As shown in the illustration, this application embodiment also provides a tool configuration device 10. The tool configuration device 10 includes a processing unit 101, configured to determine node tools for a target node. The target node is a node in a target workflow, which is a workflow used to execute a target task. The target node indicates a phased subtask of the target task. The node tools are tools required during the execution of the phased subtasks. The node tools are determined from workflow tools based on the node configuration information of the target node. The workflow tools are tools invoked during the execution of the target task. The device also includes a processing unit 101 for determining a target toolset based on the node tools. The tools in the target toolset are tools invoked during the execution of the phased subtasks.

[0080] In one embodiment, the tool configuration device 10 further includes an acquisition unit for acquiring node metadata of the target node. The node metadata is a collection of information about the execution process of the configuration phased subtasks, and the node metadata includes at least information about the node tools. The processing unit 101 is further configured to determine the node tools of the target node based on the node tool information.

[0081] In one embodiment, the processing unit 101 is further configured to determine tool constraint information of the target node. The tool constraint information indicates whether the node has permission to invoke file system tools during the execution of a phased subtask. File system tools are used to access or manage data files. The processing unit 101 is further configured to, if the tool constraint information of the target node indicates permission to invoke file system tools, determine a target toolset based on the node tools and the file system tools; or if the tool constraint information of the target node indicates no permission to invoke file system tools, determine a target toolset based on the node tools.

[0082] In one embodiment, the node metadata also includes information about dedicated tools. Dedicated tools are tools invoked during the execution of phased subtasks. These tools are used to execute control commands output by callback functions or to execute user-approved items included in the execution of phased subtasks. The callback functions are used to control the execution process of the phased subtasks. The processing unit 101 is further configured to determine the dedicated tools of the target node based on the information about the dedicated tools. The processing unit 101 is also configured to determine a target toolset based on the node tools and the dedicated tools.

[0083] In one embodiment, the information of the dedicated tool includes approval information and visualization information. The approval information is used to indicate whether user approval is required when calling the dedicated tool to perform a task. The visualization information is used to indicate that if user approval is required, the process of calling the dedicated tool to perform the task will be visualized.

[0084] In one embodiment, the node metadata further includes at least one of interaction information and subtask target information. The interaction information indicates whether the target node is an interactive node. If the target node is not an interactive node, execution resumes if the execution of the phased subtask is interrupted. If the target node is an interactive node, user interaction instructions are received if the execution of the phased subtask is interrupted, and these user interaction instructions are associated with the execution process of the phased subtask. The subtask target guides the execution process of the phased subtask.

[0085] In one embodiment, the processing unit 101 is further configured to determine the tool configuration information of each tool in the target toolset, the tool configuration information being used to indicate the function and usage of the corresponding tool; determine the sub-task target of the target node, the sub-task target being used to guide the execution process of the phased sub-task; and determine the target tool from the target toolset based on the tool configuration information and the sub-task target, the function and usage of the target tool being matched with the sub-task target.

[0086] In one embodiment, the processing unit 101 is further configured to determine context information and / or user profile. The context information includes at least one of the following: the task objective of the target workflow and the device type, wherein the task objective is used to guide the execution process of the target task. The processing unit 101 is further configured to determine the target tool from the target toolset based on the tool configuration information and sub-task objectives of each tool, as well as the context information and / or user profile.

[0087] For explanations of the terms, please refer to the relevant descriptions in the aforementioned tool configuration method embodiments, which will not be elaborated here.

[0088] It should be noted that the specific execution process of the aforementioned tool configuration device 10 can be found in [reference needed]. Figures 1 to 3 The specific details of the corresponding embodiments will not be elaborated here.

[0089] like Figure 5 As shown in the illustration, this application embodiment also provides a tool configuration device 20. The tool configuration device 20 includes a processing unit 201, configured to determine node tools for a target node. The target node is a node in a target workflow, which is a workflow used to execute a target task. The target node indicates a phased subtask of the target task. The node tools are tools required during the execution of the phased subtasks. The node tools are determined from workflow tools based on the node configuration information of the target node. The workflow tools are tools invoked during the execution of the target task. The device also includes a processing unit 201, configured to determine a target toolset based on the node tools. The tools in the target toolset are tools invoked during the execution of the phased subtasks.

[0090] For explanations of the terms, please refer to the relevant descriptions in the aforementioned tool configuration method embodiments, which will not be elaborated here.

[0091] It should be noted that the specific execution process of the aforementioned tool configuration device 20 can also be found in [reference needed]. Figures 1 to 3 The specific details of the corresponding embodiments will not be elaborated here.

[0092] like Figure 6 As shown in the illustration, this application also provides a computer device 30. Exemplarily, the computer device 30 may include a processor 301, a communication interface 302, a communication bus 303, and a memory 304. Specifically, the computer device 30 may include: The system includes at least one processor 301, such as a CPU, at least one communication interface 302, a memory 304, and at least one communication bus 303. The communication bus 303 is used to enable communication between these components. The communication interface 302 may optionally include a standard wired interface, a wireless interface (such as a Wi-Fi interface or a Bluetooth interface), etc. The memory 304 may be high-speed RAM or non-volatile memory, such as at least one disk drive. Optionally, the memory 304 may also be at least one storage device located remotely from the aforementioned processor 301. Figure 6 As shown, the memory 304, which serves as a computer storage medium, may include an operating system and program instructions.

[0093] For example, processor 301 can be used to implement the above. Figure 4 The steps or methods executed by the processing unit 101, or those used to implement the above. Figure 5 The steps or methods executed by the processing unit 201 in the process.

[0094] Understandably, the above method is merely an example, and the above can also be executed by the processor 301 and other modules in the aforementioned computer device 30. Figure 4 The steps or methods executed by the processing unit 101, or in conjunction with the above-mentioned steps or methods. Figure 5 The steps or methods executed by the processing unit 201 in the document. This document does not limit this.

[0095] exist Figure 6 In the computer device 30 shown, the processor 301 can be used to load computer program instructions stored in the memory 304 and specifically perform the following operations: The node tool determines the target node. The target node is a node in the target workflow. The target workflow is a workflow used to execute the target task. The target node indicates the phased subtasks of the target task. The node tool is the tool required in the process of executing the phased subtasks. The node tool is determined from the workflow tool based on the node configuration information of the target node. The workflow tool is a tool that is called during the execution of the target task. Based on the node tools, determine the target toolset. The tools in the target toolset are those that can be called during the execution of phased sub-tasks.

[0096] For explanations of the terms, please refer to the relevant descriptions in the aforementioned tool configuration method embodiments, which will not be elaborated here.

[0097] It should be noted that the specific execution process can be found in [link to relevant documentation]. Figures 1 to 3 The specific details of the corresponding embodiments will not be elaborated here.

[0098] This application also provides a computer-readable storage medium that can store multiple computer program instructions. These computer program instructions are adapted to be loaded and executed by a processor as described above. Figures 1 to 3 The corresponding tool configuration method embodiment provides the method steps. For a detailed execution process, please refer to [link / reference]. Figures 1 to 3 The specific details of the corresponding tool configuration method embodiments are not elaborated here.

[0099] As used in the above embodiments, depending on the context, the term "when..." can be interpreted as meaning "if...", or "after...", or "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if (the stated condition or event) is detected" can be interpreted as meaning "if determining...", or "in response to determining...", or "when (the stated condition or event) is detected", or "in response to detecting (the stated condition or event)".

[0100] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially as a computer program product. This computer program product includes one or more computer program instructions. When these computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer program instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer program instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0101] 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. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

[0102] The technical features of the above embodiments can be arbitrarily combined. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, they should be considered to be within the scope of this specification.

[0103] The embodiments described above are merely examples of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application.

Claims

1. A tool configuration method, characterized in that, The method includes: The node tool for determining the target node is a node in the target workflow, which is a workflow used to execute the target task. The target node indicates a phased subtask of the target task. The node tool is a tool required in the process of executing the phased subtask. The node tool is determined from the workflow tool based on the node configuration information of the target node. The workflow tool is a tool that is called during the execution of the target task. Based on the node tools, a target toolset is determined, wherein the tools in the target toolset are those to be invoked during the execution of the phased sub-tasks.

2. The method according to claim 1, characterized in that, Prior to the node tool for determining the target node, the method further includes: Obtain the node metadata of the target node, wherein the node metadata is a collection of information configuring the execution process of the phased subtask, and the node metadata includes at least the information of the node tool; The node tool for determining the target node includes: Based on the information from the node tool, the node tool of the target node is determined.

3. The method according to claim 1 or 2, characterized in that, The method further includes: The tool constraint information of the target node is determined. The tool constraint information is used to indicate whether there is permission to call file system tools during the execution of the phased subtask. The file system tools are used to access or manage data files. The step of determining the target toolset based on the node tools includes: If the tool constraint information of the target node indicates that it has permission to invoke the file system tools, then the target toolset is determined based on the node tools and the file system tools; or If the tool constraint information of the target node indicates that it does not have permission to call the file system tools, then the target toolset is determined based on the node tools.

4. The method according to claim 2, characterized in that, The node metadata also includes information about dedicated tools, which are tools invoked during the execution of the phased subtasks. These dedicated tools are used to execute control commands output by callback functions or to execute user-approved items included in the execution of the phased subtasks. The callback functions are used to control the execution process of the phased subtasks. The method further includes: Based on the information of the dedicated tool, determine the dedicated tool for the target node; The step of determining the target toolset based on the node tools includes: The target toolset is determined based on the node tools and the dedicated tools.

5. The method according to claim 4, characterized in that, The information of the dedicated tool includes approval information and visualization information. The approval information is used to indicate whether user approval is required when calling the dedicated tool to perform a task. The visualization information is used to indicate that the process of calling the dedicated tool to perform a task will be displayed visually when user approval is required.

6. The method according to claim 2, characterized in that, The node metadata also includes at least one of the following: interaction information and information about the subtask objective; The interaction information is used to indicate whether the target node is an interactive node. If the execution of the phased subtask is interrupted when the target node is not an interactive node, the execution is resumed. If the execution of the phased subtask is interrupted when the target node is an interactive node, a user interaction instruction is received. The user interaction instruction is associated with the execution process of the phased subtask. The subtask target is used to guide the execution process of the phased subtask.

7. The method according to any one of claims 1 to 6, characterized in that, The method further includes: Determine the tool configuration information for each tool in the target toolset, wherein the tool configuration information is used to indicate the function and usage of the corresponding tool; Determine the sub-task objectives of the target node, wherein the sub-task objectives are used to guide the execution process of the phased sub-tasks; Based on the tool configuration information of each tool and the sub-task objective, a target tool is determined from the target tool set, and the function and usage of the target tool are matched with the sub-task objective.

8. The method according to claim 7, characterized in that, Before determining the target tool from the target tool set based on the tool configuration information of each tool and the sub-task objective, the method further includes: Determine context information and / or user profile; the context information includes at least one of the following: device type, task objective of the target workflow, the task objective being used to guide the execution process of the target task; The step of determining the target tool from the target tool set based on the tool configuration information of each tool and the sub-task objective includes: The target tool is determined from the target toolset based on the tool configuration information of each tool, the sub-task objective, the context information, and / or the user profile.

9. A tool configuration method, characterized in that, The method includes: A node tool for determining a target node, wherein the target node is a node in a target workflow, the target workflow is a workflow used to execute code generation tasks, the target node indicates a code release subtask in the code generation task, the node tool is a tool required in the process of executing the code release subtask, the node tool is determined from a workflow tool based on the node configuration information of the target node, and the workflow tool is a tool to be called during the execution of the code generation task; Based on the node tools, a target toolset is determined, wherein the tools in the target toolset are those to be invoked during the execution of the code release subtask.

10. A tool configuration device, characterized in that, The device includes: A processing unit is used to determine node tools for a target node, wherein the target node is a node in a target workflow, the target workflow is a workflow used to execute a target task, the target node indicates a phased subtask of the target task, the node tool is a tool required in the process of executing the phased subtask, the node tool is determined from a workflow tool based on the node configuration information of the target node, and the workflow tool is a tool to be called during the execution of the target task; based on the node tool, a target toolset is determined, wherein the tools in the target toolset are tools to be called during the execution of the phased subtask.

11. A tool configuration device, characterized in that, The device includes: A processing unit is used to determine node tools for a target node, wherein the target node is a node in a target workflow, the target workflow is a workflow for executing a code generation task, the target node indicates a code release subtask in the code generation task, the node tool is a tool required in the process of executing the code release subtask, the node tool is determined from a workflow tool based on the node configuration information of the target node, and the workflow tool is a tool to be invoked during the execution of the code generation task; based on the node tool, a target toolset is determined, wherein the tools in the target toolset are tools to be invoked during the execution of the code release subtask.

12. A computer device, characterized in that, include: A memory and a processor, wherein the memory stores computer program instructions; when executed by the processor, the computer program instructions cause the processor to perform the method as described in any one of claims 1 to 8, or to perform the method as described in claim 9.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions; when the computer program instructions are run on one or more processors, they perform the method as described in any one of claims 1 to 8, or perform the method as described in claim 9.

14. A computer program product, characterized in that, The computer program product includes computer program instructions; when the computer program instructions are executed on a computer, the computer causes the computer to perform the method as claimed in any one of claims 1 to 8, or to perform the method as claimed in claim 9.

Images