Interface MCP method based on APIG

Abstract

The invention discloses an interface MCP method based on APIG, and relates to the technical field of cloud computing and large models. Comprising the steps that 1, an APIG-based interface MCP system is created, 2, interface metadata information is collected from a Swagger document, an Excel interface list or an API gateway through an interface information collection module, and the interface metadata information comprises URL, request types, parameter structures and attributes, parameter default values and limiting condition information of an interface; the method comprises the following steps of: 1, acquiring interface metadata information, 2, automatically analyzing the interface metadata information through an MCP service packet generation module, and generating a Server code packet conforming to an MCP protocol format and a corresponding JSON configuration file which are used for two communication modes of STDIO and SSE, and 4, simulating an MCP Client to call an MCP Server interface through an MCP service packet test module, verifying response correctness and context coherence after the service packet is applied to the interface, and sending the response correctness and the context coherence to the MCP server interface through the MCP service packet test module. And step 5, deploying the service package passing the test to a local or cloud operation environment through an MCP deployment and release module, and carrying out service registration, authentication information issuing and operation state monitoring.

Description

Technical Field

[0001] This invention discloses an interface MCP method based on APIG, which relates to the fields of cloud computing and large model technology. Background Technology

[0002] In the current era of rapid development in artificial intelligence, the integration of existing applications with large-scale model tools has become an important direction for promoting the intelligent upgrading of various industries. However, existing application interfaces typically use protocols such as RESTful and SOAP, which differ significantly from the MCP protocol required by large-scale model tools. This results in a large amount of complex modification and adaptation work required when interfaces are integrated with large-scale model tools.

[0003] In existing technologies, interface MCP (Multi-Channel Programming) transformation mainly relies on manual modification. Developers need to manually parse the request-response structure of the original interface, rewrite the parameter mapping logic according to the MCP specification, write adaptation layer code and verify compatibility, and configure large model calling rules individually for each interface. This not only consumes a lot of manpower and time, but is also prone to human error, resulting in low efficiency in interface integration with large model tools and seriously hindering the process of application integration with large models. In addition, due to the lack of a unified automated processing mechanism, it is difficult to standardize and normalize the interface transformation process of different applications, further increasing the difficulty of intelligent application transformation.

[0004] APIG (API Gateway), as an important component for managing and distributing API requests in applications, has achieved unified management and traffic control of interfaces to a certain extent. However, when connecting with large model tools, due to the significant differences between its native protocol and the MCP (Model Context Protocol) used by large models, it is difficult to directly meet the needs of large models for standardized context information exchange. Summary of the Invention

[0005] This invention addresses the problems of existing technologies by providing an APIG-based interface MCP method, which solves the problems of time-consuming, labor-intensive, and error-prone manual modification processes in existing technologies, and enables rapid and seamless integration of applications with large model tools.

[0006] This invention provides an APIG-based interface MCP method, comprising:

[0007] Step 1: Create an APIG-based interface MCP system. This system includes an interface information acquisition module, an MCP service package generation module, an MCP service package testing module, and an MCP deployment and release module.

[0008] Step 2: Collect API metadata information from the Swagger documentation, Excel API list, or API gateway using the API information collection module. This metadata includes the API URL, request type, parameter structure and attributes, default parameter values, and limiting conditions.

[0009] Step 3: The MCP service package generation module automatically parses the interface metadata information and generates a Server code package conforming to the MCP protocol format and a corresponding JSON configuration file for both STDIO and SSE communication modes.

[0010] Step 4: Simulate an MCP Client calling the MCP Server interface using the MCP service package testing module. Verify the correctness of the response and the consistency of the context after the service package is applied to the interface, and output a test report.

[0011] Step 5: Deploy the tested service package to the local or cloud runtime environment through the MCP deployment and release module, and perform service registration, authentication information distribution and runtime status monitoring.

[0012] Furthermore, in step 2 of the APIG-based interface MCP method, the interface library established by the interface information collection module stores interface and metadata information in a structured and searchable form, constructs an interface library index and retrieval structure, ensures that the subsequent MCP service package generation module can call the interface definition, and provides a complete and standardized data source for subsequent stages.

[0013] Furthermore, in step 3 of the APIG-based interface MCP method, the parameters, types, default values, and required attributes of each interface are determined based on the interface library metadata, and the natural language descriptions of the interface and parameters are extracted. Then, according to the built-in code template, all the above service descriptions, interface descriptions, parameter descriptions, and basic information of the interface and parameters are filled into the template to generate a code file. The compiler is then called to compile the code file into an executable program package.

[0014] The MCP service package generation module generates two types of MCP service packages: STDIO and SSE. The STDIO package is directly available for users to download and use after generation, while the SSE package is deployed and used in conjunction with the MCP deployment and publishing module.

[0015] The MCP service package generation module automatically generates a JSON configuration file that conforms to the Anthropic MCP protocol definition.

[0016] Furthermore, in step 4 of the APIG-based interface MCP-ization method, the MCP service package testing module is embedded with an interaction engine that simulates an MCP client. Following the MCP protocol standard, it simulates the context request process of a real large-scale tool for the service package, including parameter input, service invocation, and response result parsing operations.

[0017] In the testing process, the MCP service package testing module first loads the generated STDIO or SSE MCPServer package and configuration file to automatically complete the service startup and connection tests; then, through the automated test script, each interface is called in turn to simulate various request scenarios, verify whether the response data meets the expected structure, and output a test report.

[0018] For interfaces that do not conform to MCP semantics or have incorrect responses, highlight them and indicate possible causes to help developers quickly locate and fix problems;

[0019] The MCP service package testing module also integrates a dialogue context simulation function, which simulates the context memory call scenarios during multi-round interaction of the model, and is used to verify the performance of the interface in complex applications.

[0020] Furthermore, in step 5 of the APIG-based interface MCP-ization method, an interface is provided through the MCP deployment and release module for call log collection and performance monitoring, real-time monitoring of service operation status and alarm configuration; after deployment, the deployment status, call frequency and failure information of all MCP-ized interfaces can be viewed through the APIG console, realizing unified operation and maintenance and management of the entire MCP-ized interface system.

[0021] This invention also provides an APIG-based interface MCP system, including an interface information acquisition module, an MCP service package generation module, an MCP service package testing module, and an MCP deployment and release module.

[0022] The API information collection module gathers API metadata information from Swagger documentation, Excel API lists, or the API gateway. This metadata includes the API URL, request type, parameter structure and attributes, default parameter values, and limiting conditions.

[0023] The MCP service package generation module automatically parses the interface metadata information and generates a Server code package conforming to the MCP protocol format, along with a corresponding JSON configuration file, for use in both STDIO and SSE communication modes.

[0024] The MCP service package testing module simulates an MCP Client calling an MCP Server interface to verify the correctness and contextual consistency of the response after the service package is applied to the interface, and outputs a test report.

[0025] The MCP deployment and release module will deploy the tested service packages to the local or cloud runtime environment, and perform service registration, authentication information distribution, and runtime status monitoring.

[0026] Furthermore, the interface information collection module of the APIG-based interface MCP system establishes an interface library that stores interface and metadata information in a structured and searchable form, constructs an interface library index and retrieval structure, ensures that the subsequent MCP service package generation module can call the interface definition, and provides a complete and standardized data source for subsequent stages.

[0027] Furthermore, the MCP service package generation module of the APIG-based interface MCP system determines the parameters, types, default values, and required attributes of each interface based on the interface library metadata, and extracts the natural language descriptions of the interfaces and parameters. Then, according to the built-in code template, it fills in all the above service descriptions, interface descriptions, parameter descriptions, and basic information about the interfaces and parameters into the template to generate a code file. The compiler is then called to compile the code file into an executable program package.

[0028] The MCP service package generation module generates two types of MCP service packages: STDIO and SSE. The STDIO package is directly available for users to download and use, while the SSE package is used in conjunction with the MCP deployment and release module for deployment.

[0029] The MCP service package generation module automatically generates a JSON configuration file that conforms to the Anthropic MCP protocol definition.

[0030] Furthermore, the MCP service package testing module of the APIG-based interface MCP system has a built-in interactive engine that simulates an MCP client. Following the MCP protocol standard, it simulates the context request process of a real large-scale tool for the service package, including parameter input, service invocation, and response result parsing operations.

[0031] In the testing process, the MCP service package testing module first loads the generated STDIO or SSE MCP Server package and configuration file, and automatically completes the service startup and connection tests; then, through automated test scripts, it calls each interface in sequence to simulate various request scenarios, verify whether the response data meets the expected structure, and outputs a test report.

[0032] For interfaces that do not conform to MCP semantics or have incorrect responses, highlight them and indicate possible causes to help developers quickly locate and fix problems;

[0033] The MCP service package testing module also integrates a dialogue context simulation function, which simulates the context memory call scenarios during multi-round interaction of the model, and is used to verify the performance of the interface in complex applications.

[0034] Furthermore, the MCP deployment and release module of the APIG-based interface MCP system provides an interface for call log collection and performance monitoring, real-time monitoring of service operation status and alarm configuration; after deployment, the deployment status, call frequency and failure information of all MCP-based interfaces can be viewed through the APIG console, realizing unified operation and management of the entire MCP-based interface system.

[0035] The advantages of this invention are:

[0036] 1. Significantly reduce transformation costs: By automatically parsing traditional interface documents and generating MCP service packages, the process of manually rewriting a large amount of adaptation logic is avoided, significantly reducing the manpower and time costs of interface transformation.

[0037] 2. Improve development efficiency: Supports one-click generation of complete MCP Server services and configuration files from interface definitions. Developers only need a few operations to complete MCP deployment, greatly shortening the interface launch cycle.

[0038] 3. Reduce error risk: By using a templated structure and automatic verification mechanism, the system significantly reduces the risk of errors such as semantic inconsistencies and parameter omissions that are common in manual modification, thereby enhancing system stability.

[0039] 4. Compatible with multiple deployment environments: Supports both STDIO and SSE protocols to meet different deployment needs and adapt to various operating environments such as local development, private cloud, hybrid cloud and public cloud. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0041] APIG (API Gateway) is a middleware that provides API hosting services. It sits between the client and the backend service, serving as the sole entry point for clients to access the backend service. All requests from clients first pass through the API Gateway, which then routes the requests to the backend service. Like a gatekeeper, it performs authentication, permission checks, and traffic control to ensure the security and stability of API requests.

[0042] Large models refer to deep learning models with a large number of parameters, typically reaching billions or even more. These models can handle complex tasks such as natural language processing, image recognition, and recommendation systems, and can provide high-quality results.

[0043] MCP (Model Context Protocol) is an open-source protocol introduced by Anthropic, designed to integrate large language models with external data sources and tools, defining a standardized way for applications to provide context to LLMs. You can think of MCP as a USB-C port for AI applications. Just as USB-C provides a standardized way to connect devices to various peripherals and accessories, MCP provides a standardized way to connect AI models to different data sources and tools. The MCP protocol is divided into client and server sides. The client refers to the tools and clients that call various services through the MCP protocol, while the server side refers to the servers that can provide various services and tools through the MCP protocol.

[0044] MCP communication modes include two types: STDIO and SSE. STDIO refers to the MCP server running on the same machine as the MCP client, transmitting data via standard input, output, and error streams. SSE refers to the MCP server running remotely; it's a unidirectional data streaming method based on the HTTP protocol, allowing the server to proactively push real-time data to the client. SSE maintains a persistent HTTP connection to stream data to the client.

[0045] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0046] Example 1

[0047] This invention provides an APIG-based interface MCP method, comprising:

[0048] Step 1: Create an APIG-based interface MCP system, which includes an interface information acquisition module, an MCP service package generation module, an MCP service package testing module, and an MCP deployment and release module.

[0049] Step 2: Collect interface metadata information from Swagger documentation, Excel interface list, or API gateway using the interface information collection module. Interface metadata information includes the interface URL, request type, parameter structure and attributes, parameter default values, and restriction information.

[0050] The API information acquisition module standardizes and parses API definitions from various sources and with varying levels of structure, including Swagger documentation, Postman API documentation, and Excel API lists. It extracts information such as the API URL, request type, parameter structure and attributes, default parameter values, and limitations. It also supports API creation via a graphical interface. The module largely reuses existing APIG functionality; existing APIs configured in APIG can be used directly without requiring re-entry of configurations.

[0051] An interface library was established to store interface and metadata information in a structured and searchable form. An interface library index and retrieval structure were built to ensure that the subsequent MCP generator can quickly and accurately call the interface definition and provide a complete and standardized data source for subsequent stages.

[0052] Step 3: The MCP service package generation module automatically parses the interface metadata information and generates a Server code package conforming to the MCP protocol format and a corresponding JSON configuration file for both STDIO and SSE communication modes.

[0053] Users need to select their already entered API interfaces, treat them as a single MCP service, and describe them. Next, the interface structure mapper of the MCP service package generation module determines the MCP structured capability definition for each interface based on the interface library metadata, identifying parameters, types, default values, and required attributes, and extracting natural language descriptions of the interface and parameters. Then, the MCP service package generation module uses a built-in code template to fill in all the above service descriptions, interface descriptions, parameter descriptions, and basic information about the interface and parameters, generating a code file. Finally, a compiler is invoked to compile the code file into an executable package.

[0054] The MCP service package generation module can generate two types of MCP service packages: STDIO and SSE. The STDIO package can be downloaded and used directly by users after generation, while the SSE package needs to be deployed and used in conjunction with the subsequent "MCP Deployment and Release Module".

[0055] The MCP service package generation module can automatically generate JSON configuration files that fully conform to the Anthropic MCP protocol definition, determining how to use the MCP service package, related environment information, and other configurations. This function is also template-based, and can generate both STDIO and SSE JSON configuration files.

[0056] Step 4: Simulate the MCP Client calling the MCP Server interface using the MCP service package test module to verify the correctness of the response and the consistency of the context after the service package is applied to the interface, and output a test report.

[0057] The MCP service package testing module incorporates an interactive engine that simulates an MCP client. It can simulate the context request process of a real large-scale model tool for the service package according to the MCP protocol standard, including parameter input, service call, and response result parsing.

[0058] In the testing process, the module first loads the generated STDIO or SSE MCP Server package and its configuration file, automatically completing service startup and connection tests. Then, through automated test scripts, each interface is called sequentially to simulate various request scenarios, verifying whether the response data conforms to the expected structure, and outputting a test report. For interfaces that do not conform to MCP semantics or have incorrect responses, the system will highlight them and indicate possible causes, helping developers quickly locate and fix problems.

[0059] In addition, the MCP service package testing module also integrates the "dialogue context simulation" function, which can simulate the context memory call scenario in the multi-turn interaction process of the model, and is used to verify the performance of the interface in complex applications.

[0060] Step 5: Deploy the tested service package to the local or cloud runtime environment using the MCP deployment and release module, and perform service registration, authentication information distribution, and runtime status monitoring. The MCP deployment and release module supports various deployment methods such as local deployment, private cloud deployment, and public cloud deployment, and provides compatibility with various cloud platforms such as Kubernetes, OpenShift, and Inspur Cloud Native Platform, enabling a standardized, low-code service deployment process.

[0061] The MCP deployment and release module supports MCP Server in SSE mode. The SSE service package is deployed in a cloud-native environment via containerization and automatically generates JSON configurations that can be used by large model tools.

[0062] Furthermore, the APIG deployment and release module provides an interface for call log collection and performance monitoring, enabling real-time monitoring and alarm configuration of service operation status. After deployment, the APIG console displays the deployment status, call frequency, and failure information of all MCP-enabled interfaces, achieving unified operation and management of the entire MCP-enabled interface system.

[0063] Example 2

[0064] This invention also provides an APIG-based interface MCP system, including an interface information acquisition module, an MCP service package generation module, an MCP service package testing module, and an MCP deployment and release module.

[0065] The API information collection module gathers API metadata information from Swagger documentation, Excel API lists, or the API gateway. This metadata includes the API URL, request type, parameter structure and attributes, default parameter values, and limiting conditions.

[0066] The MCP service package generation module automatically parses the interface metadata information and generates a Server code package conforming to the MCP protocol format, along with a corresponding JSON configuration file, for use in both STDIO and SSE communication modes.

[0067] The MCP service package testing module simulates an MCP Client calling an MCP Server interface to verify the correctness and contextual consistency of the response after the service package is applied to the interface, and outputs a test report.

[0068] The MCP deployment and release module will deploy the tested service packages to the local or cloud runtime environment, and perform service registration, authentication information distribution, and runtime status monitoring.

[0069] The information interaction and execution process between the modules in the above system are based on the same concept as the method embodiment of the present invention, and the specific details can be found in the description in the method embodiment of the present invention, and will not be repeated here.

[0070] Similarly, the advantages of the system of the present invention are:

[0071] 1. Significantly reduce transformation costs: By automatically parsing traditional interface documents and generating MCP service packages, the process of manually rewriting a large amount of adaptation logic is avoided, significantly reducing the manpower and time costs of interface transformation.

[0072] 2. Improve development efficiency: Supports one-click generation of complete MCP Server services and configuration files from interface definitions. Developers only need a few operations to complete MCP deployment, greatly shortening the interface launch cycle.

[0073] 3. Reduce error risk: By using a templated structure and automatic verification mechanism, the system significantly reduces the risk of errors such as semantic inconsistencies and parameter omissions that are common in manual modification, thereby enhancing system stability.

[0074] 4. Compatible with multiple deployment environments: Supports both STDIO and SSE protocols to meet different deployment needs and adapt to various operating environments such as local development, private cloud, hybrid cloud and public cloud.

[0075] It should be noted that not all steps and modules in the above processes and system structures are mandatory; some steps or modules can be omitted as needed. The execution order of each step is not fixed and can be adjusted as required. The system structures described in the above embodiments can be physical or logical structures. That is, some modules may be implemented by the same physical entity, or some modules may be implemented by multiple physical entities, or they may be jointly implemented by certain components in multiple independent devices.

[0076] The above-described embodiments are merely preferred embodiments provided to fully illustrate the present invention, and the scope of protection of the present invention is not limited thereto. Equivalent substitutions or modifications made by those skilled in the art based on the present invention are all within the scope of protection of the present invention. The scope of protection of the present invention is defined by the claims.

Claims

1. An APIG-based interface MCP method, characterized by: include: Step 1: Create an APIG-based interface MCP system. This system includes an interface information acquisition module, an MCP service package generation module, an MCP service package testing module, and an MCP deployment and release module. Step 2: Collect API metadata information from the Swagger documentation, Excel API list, or API gateway using the API information collection module. This metadata includes the API URL, request type, parameter structure and attributes, default parameter values, and limiting conditions. Step 3: The MCP service package generation module automatically parses the interface metadata information and generates a Server code package conforming to the MCP protocol format and a corresponding JSON configuration file for both STDIO and SSE communication modes. Step 4: Simulate an MCP Client calling the MCP Server interface using the MCP service package testing module. Verify the correctness of the response and the consistency of the context after the service package is applied to the interface, and output a test report. Step 5: Deploy the tested service package to the local or cloud runtime environment through the MCP deployment and release module, and perform service registration, authentication information distribution and runtime status monitoring.

2. The APIG-based interface MCP method according to claim 1, characterized in that: In step 2, the interface library established by the interface information collection module stores interface and metadata information in a structured and searchable form, constructs an interface library index and retrieval structure, ensures that the subsequent MCP service package generation module can call the interface definition, and provides a complete and standardized data source for subsequent stages.

3. The APIG-based interface MCP method according to claim 1, characterized in that in step 3, the parameters, types, default values, and required attributes of each interface are determined according to the interface library metadata, and the natural language descriptions of the interface and parameters are extracted; then, according to the built-in code template, all the above service descriptions, interface descriptions, parameter descriptions, and basic information of the interface and parameters are filled into the template to generate a code file, and the compiler is called to compile the code file into an executable program package. The MCP service package generation module generates two types of MCP service packages: STDIO and SSE. The STDIO package is directly available for users to download and use after generation, while the SSE package is deployed and used in conjunction with the MCP deployment and publishing module. The MCP service package generation module automatically generates a JSON configuration file that conforms to the Anthropic MCP protocol definition.

4. The APIG-based interface MCP method according to claim 1, Its key feature is that in step 4, the MCP service package testing module incorporates an interaction engine that simulates an MCP client. Following the MCP protocol standard, it simulates the context request process of a real large-scale tool for the service package, including parameter input, service invocation, and response result parsing. In the testing process, the MCP service package testing module first loads the generated STDIO or SSE MCP Server package and configuration file to automatically complete the service startup and connection tests; then, through the automated test script, each interface is called in turn to simulate various request scenarios, verify whether the response data meets the expected structure, and output a test report. For interfaces that do not conform to MCP semantics or have incorrect responses, highlight them and indicate possible causes to help developers quickly locate and fix problems; The MCP service package testing module also integrates a dialogue context simulation function, which simulates the context memory call scenarios during multi-round interaction of the model, and is used to verify the performance of the interface in complex applications.

5. The APIG-based interface MCP method according to claim 1, characterized in that: In step 5, the MCP deployment and release module provides an interface for calling log collection and performance monitoring, and performs real-time monitoring and alarm configuration of service operation status. After deployment, the deployment status, call frequency and failure information of all MCP-enabled interfaces can be viewed through the APIG console, realizing unified operation and maintenance and management of the entire MCP-enabled interface system.

6. An APIG-based interface MCP system, characterized by: It includes an interface information collection module, an MCP service package generation module, an MCP service package testing module, and an MCP deployment and release module. The API information collection module gathers API metadata information from Swagger documentation, Excel API lists, or the API gateway. This metadata includes the API URL, request type, parameter structure and attributes, default parameter values, and limiting conditions. The MCP service package generation module automatically parses the interface metadata information and generates a Server code package conforming to the MCP protocol format, along with a corresponding JSON configuration file, for use in both STDIO and SSE communication modes. The MCP service package testing module simulates an MCP Client calling an MCP Server interface to verify the correctness and contextual consistency of the response after the service package is applied to the interface, and outputs a test report. The MCP deployment and release module will deploy the tested service packages to the local or cloud runtime environment, and perform service registration, authentication information distribution, and runtime status monitoring.

7. The APIG-based interface MCP system according to claim 6, characterized in that the interface... The interface library established by the information collection module stores interface and metadata information in a structured and searchable form, constructs an interface library index and retrieval structure, ensures that the subsequent MCP service package generation module can call the interface definition, and provides a complete and standardized data source for subsequent stages.

8. The APIG-based interface MCP system according to claim 6, characterized in that MCP... The service package generation module determines the parameters, types, default values, and required attributes of each interface based on the interface library metadata, and extracts the natural language descriptions of the interface and parameters. Then, based on the built-in code template, it fills in all the above service descriptions, interface descriptions, parameter descriptions, and basic information about the interface and parameters into the template, generates a code file, and calls the compiler to compile the code file into an executable program package. The MCP service package generation module generates two types of MCP service packages: STDIO and SSE. The STDIO package is directly available for users to download and use, while the SSE package is used in conjunction with the MCP deployment and release module for deployment. The MCP service package generation module automatically generates a JSON configuration file that conforms to the Anthropic MCP protocol definition.

9. The APIG-based interface MCP system according to claim 6, characterized in that: The MCP service package testing module has a built-in interaction engine that simulates an MCP client. Following the MCP protocol standard, it simulates the context request process of a real large-scale tool for a service package, including parameter input, service invocation, and response result parsing. In the testing process, the MCP service package testing module first loads the generated STDIO or SSE MCP Server package and configuration file, and automatically completes the service startup and connection tests; then, through automated test scripts, it calls each interface in sequence to simulate various request scenarios, verify whether the response data meets the expected structure, and outputs a test report. For interfaces that do not conform to MCP semantics or have incorrect responses, highlight them and indicate possible causes to help developers quickly locate and fix problems; The MCP service package testing module also integrates a dialogue context simulation function, which simulates the context memory call scenarios during multi-round interaction of the model, and is used to verify the performance of the interface in complex applications.

10. The APIG-based interface MCP system according to claim 6, characterized in that: The MCP deployment and release module provides interfaces for call log collection and performance monitoring, enabling real-time monitoring and alarm configuration of service operation status. After deployment, the APIG console allows users to view the deployment status, call frequency, and failure information of all MCP-enabled interfaces, achieving unified operation and management of the entire MCP-enabled interface system.

Images