CN-122027612-A - AI application system and operation method thereof

Abstract

The invention provides an AI application system and an operation method thereof, wherein the system comprises an application front end and an application server; the application front end comprises an interaction interface and a first MCP server, the application server comprises a first MCP client and a proxy MCP server, the interaction interface sends an interaction request, the first MCP client generates a server calling request and sends the server calling request to the proxy MCP server, the proxy MCP server establishes communication connection with the first MCP server and sends the server calling request to the first MCP server, the first MCP server responds to the server calling request to generate response data and returns the response data to the proxy MCP server, the proxy MCP server returns the response data to the first MCP client, and the first MCP client provides a request result corresponding to the interaction request to the interaction interface based on the response data. The method can realize intelligent user interaction while guaranteeing data security.

Inventors

• ZHAO YUSEN
• SONG RUI

Assignees

• 杭州网易智企科技有限公司

Dates

Publication Date

20260512

Application Date

20251230

Claims (10)

1. 1. The AI application system is characterized by comprising an application front end and an application service end, wherein the application front end comprises an interactive interface and a first MCP service end, the application service end comprises a first MCP client and a proxy MCP service end, the first MCP client and the proxy MCP service end are connected with each other, The interactive interface is used for receiving an interactive request and sending the interactive request to the first MCP client; the first MCP client is used for generating a server call request corresponding to the interaction request and sending the server call request to the proxy MCP server; the proxy MCP server is used for establishing communication connection with the first MCP server and sending the server calling request to the first MCP server; the first MCP server is used for responding to the server call request, generating response data and returning the response data to the proxy MCP server; The proxy MCP server is also used for returning the response data to the first MCP client; the first MCP client is further used for providing a request result corresponding to the interaction request to the interaction interface based on the response data.
2. 2. The system of claim 1, wherein the application front-end further comprises a reverse transmission module for: Receiving the server call request sent by the proxy MCP server; and transmitting the server call request to the first MCP server based on a preset network communication protocol.
3. 3. The system of claim 1, wherein the application front-end further comprises a front-end SDK in which the first MCP server and the reverse transmission module are disposed, a server SDK in which the proxy MCP server is disposed; The front-end SDK is used for responding to the interaction request received by the interaction interface, generating a client identifier corresponding to the front-end SDK, carrying the client identifier to the interaction request and sending the client identifier to the server-end SDK; the server SDK is used for establishing a communication connection relation between the client identifier and the first MCP server and storing the communication connection relation.
4. 4. The system of claim 3, wherein the proxy MCP server is further configured to: After receiving a server call request corresponding to the interaction request, acquiring a client identifier corresponding to the interaction request; and acquiring the front-end SDK corresponding to the client identifier from the pre-established communication connection relationship, and sending the server call request to a first MCP server in the front-end SDK.
5. 5. The system of claim 3, wherein the front-end SDKs comprise a plurality of front-end SDKs having the same source data, wherein different front-end SDKs having the same source data have corresponding child identifications, wherein the source data comprises at least one of protocol data, domain name data, and port data; The front-end SDK is further used for responding to the interaction request received by the interaction interface, generating a client identifier and a sub identifier corresponding to the front-end SDK, carrying the client identifier and the sub identifier to the interaction request, and sending the client identifier and the sub identifier to the server-end SDK; the server SDK is used for establishing a communication connection relation between the client identifier and the front-end SDK and storing the communication connection relation.
6. 6. The operation method of the AI application system is characterized in that the method is applied to terminal equipment, the system comprises an application front end and an application service end, the application front end is operated on the terminal equipment and comprises an interaction interface and a first MCP service end, the application service end comprises a first MCP client and a proxy MCP service end, and the method comprises the following steps: Receiving an interaction request through the interaction interface, sending the interaction request to the first MCP client, generating a server call request corresponding to the interaction request through the first MCP client, and sending the server call request to the proxy MCP server; establishing communication connection with the first MCP server through the proxy MCP server, and sending the server calling request to the first MCP server; And responding to the service-end calling request through a first MCP service end, generating response data, returning the response data to the proxy MCP service end, wherein the proxy MCP service end is further used for returning the response data to the first MCP client end, and providing a request result corresponding to the interaction request to the interaction interface through the first MCP client end based on the response data.
7. 7. The operation method of the AI application system is characterized in that the method is applied to a server, the system comprises an application front end and an application service end, the application service end is operated on the server, the application front end comprises an interaction interface and a first MCP service end, the application service end comprises a first MCP client and a proxy MCP service end, and the method comprises the following steps: Generating a server call request corresponding to an interaction request through the first MCP client, and sending the server call request to the proxy MCP server, wherein the interaction request is received and sent by the application front end through the interaction interface; establishing communication connection with the first MCP server through the proxy MCP server, and sending the server calling request to the first MCP server; Responding to the server call request through the first MCP server, generating response data, and returning the response data to the proxy MCP server; Returning the response data to the first MCP client through the proxy MCP server; and providing a request result corresponding to the interaction request to the interaction interface by the first MCP client based on the response data.
8. 8. A terminal device comprising a processor and a memory, the memory storing computer-executable instructions executable by the processor, the processor executing the computer-executable instructions to implement the method of operating the AI application system of claim 6.
9. 9. A server comprising a processor and a memory, the memory storing computer-executable instructions executable by the processor, the processor executing the computer-executable instructions to implement the method of operating the AI application system of claim 7.
10. 10. A computer-readable storage medium storing computer-executable instructions that, when invoked and executed by a processor, cause the processor to implement the method of operating an AI application system of claim 6 or 7.

Description

AI application system and operation method thereof Technical Field The invention relates to the technical field of application programs, in particular to an AI application system and an operation method thereof. Background In an MCP (Model Context Protocol ) based AI application, the MCP is used to provide a communication interface to the AI model so that the AI model performs tasks, and the AI application provides relevant application services to the user based on the results of the task execution by the AI model. The MCP comprises an MCP client and an MCP server, and the AI application comprises an application front end and an application server, wherein in the related technology, when the MCP client and the MCP server are both arranged at the application front end, the high data security risk is realized, and when the MCP client and the MCP server are both arranged at the application server, the AI application is difficult to realize intelligent user interaction due to the fact that the user interaction is generated at the application front end, and therefore, the current AI application is difficult to realize intelligent user interaction while ensuring the data security. Disclosure of Invention In view of the above, the present invention aims to provide an AI application system and an operation method thereof, so as to realize intelligent user interaction while ensuring data security. In a first aspect, the embodiment of the invention provides an AI application system, which comprises an application front end and an application service end, wherein the application front end comprises an interaction interface and a first MCP service end, the application service end comprises a first MCP client and a proxy MCP service end, the interaction interface is used for receiving an interaction request and sending the interaction request to the first MCP client, the first MCP client is used for generating a service end calling request corresponding to the interaction request and sending the service end calling request to the proxy MCP service end, the proxy MCP service end is used for establishing communication connection with the first MCP service end and sending the service end calling request to the first MCP service end, the first MCP service end is used for responding to the service end calling request and generating response data and returning the response data to the proxy MCP service end, and the proxy MCP service end is further used for returning the response data to the first MCP client and providing a request result corresponding to the interaction request to the interaction interface based on the response data. In a second aspect, the embodiment of the invention provides an operation method of an AI application system, which is applied to terminal equipment, wherein the system comprises an application front end and an application service end, the application front end is operated at the terminal equipment, the application front end comprises an interaction interface and a first MCP service end, the application service end comprises a first MCP client and a proxy MCP service end, the method comprises the steps of receiving the interaction request through the interaction interface, sending the interaction request to the first MCP client, generating a service end calling request corresponding to the interaction request through the first MCP client, sending the service end calling request to the proxy MCP service end, establishing communication connection with the first MCP service end through the proxy MCP service end, sending the service end calling request to the first MCP service end, responding to the service end calling request through the first MCP service end, generating response data through the first MCP service end, returning the response data to the proxy MCP service end, and further used for returning the response data to the first MCP client through the proxy MCP service end, and providing a request result corresponding to the interaction request to the interaction interface based on the response data through the first MCP client. The embodiment of the invention provides an operation method of an AI application system, which is applied to a server, wherein the system comprises an application front end and an application service end, the application service end is operated on the server, the application front end comprises an interaction interface and a first MCP service end, the application service end comprises a first MCP client and a proxy MCP service end, the method comprises the steps of generating a service end calling request corresponding to the interaction request through the first MCP client, sending the service end calling request to the proxy MCP service end, wherein the interaction request is received and sent by the application front end through the interaction interface, communication connection is established between the proxy MCP service end and the first MCP service end, the