Agent container operation system

Abstract

The embodiment of the invention discloses an Agent container operation system. The system comprises an Agent monitoring module, an agent module, a docker container, an Agent bin and a docker mirror image bin, the Agent monitoring module monitors a user request in real time, when the user request is received, a mirror image required by the Agent is obtained from the docker mirror image bin, and the docker container is created according to the mirror image; the Agent monitoring module obtains an Agent code needing to be operated from an Agent bin according to a user request, the Agent code is loaded to a docker container, and an Agent process is created in the docker container; the Agent module sends the user request to an Agent process, and the Agent process comprises an Agent framework module; and the Agent framework module operates according to the operation logic of the Agent process. According to the system, dependency conflicts among different Agents are avoided, and the resource utilization efficiency is improved. Meanwhile, the Agent module is used for forwarding the user request, so that the accuracy and timeliness of request transmission are ensured, the operation of the whole system is more stable and flexible, and the operation and maintenance cost is reduced.

Description

Technical Field

[0001] This invention relates to the field of agent operation technology, and in particular to an agent container operation system. Background Technology

[0002] An agent, or intelligent agent, is a computational entity capable of autonomously perceiving its environment, making decisions, and taking actions. The core functions of an agent include perception, decision-making, and action. These functions enable agents to autonomously perform tasks in dynamic environments, thereby reducing reliance on human intervention.

[0003] Since agents typically possess the ability to execute code and OS commands (hereinafter referred to as code execution for simplicity), this can potentially disrupt the runtime environment and cause security issues. Therefore, a solution for secure code execution is needed. A common approach is to execute code within a container or sandbox when the agent needs to do so. For example, CREWAI (a production-grade Python framework for orchestrating role-playing autonomous AI agents) supports running specified Python code in a container; Manus runs code in a sandbox; and Wuying AgentBay provides an environment for agent code execution. These solutions all rely on an external sandbox or container (hereinafter referred to as a container unless otherwise specified). Containers run an independent OS image. The agent creates a container for each user session. Since the code executed by the agent is typically small, for example, typically less than 200 lines, running an independent OS image for such a small amount of code results in a significant waste of resources; while a large portion of the agent's own code runs on the host machine, leading to extremely uneven system load. In addition, each sandbox or runtime environment provider offers different SDKs, which allow the Agent to send code to the sandbox or runtime environment for execution. This forces Agent developers to bind to a particular sandbox or runtime environment provider or to develop complex adaptation layers, resulting in severe coupling between systems. Summary of the Invention

[0004] This invention provides an Agent container runtime system, which aims to solve the technical problems of complex Agent runtime environment configuration and easy dependency conflicts between different Agents in traditional methods.

[0005] In a first aspect, embodiments of the present invention provide an Agent container runtime system, including an Agent listening module, an agent module, a Docker container, an Agent repository, and a Docker image repository, wherein:

[0006] The Agent listening module listens for user requests in real time. When a user request is received, it retrieves the image required by the Agent from the Docker image repository and creates the Docker container based on the image. The Agent listening module also retrieves the Agent code to be run from the Agent repository according to the user request, loads the Agent code into the Docker container, and creates an Agent process in the Docker container. The Docker image repository stores multiple image images with different functions and versions, and the Agent repository stores Agent code with different functions and versions.

[0007] The Agent proxy module sends the user request to the Agent process, and the Agent process includes the Agent framework module;

[0008] The Agent framework module runs according to the running logic of the Agent process.

[0009] Furthermore, the Agent process also includes code tools and other tools, and the Agent framework module runs according to the running logic of the Agent process, including:

[0010] The Agent framework module interacts with a pre-built large language model, parses the interaction information returned by the pre-built large language model, obtains the parsing result, and calls the code tool or other tools based on the parsing result.

[0011] Furthermore, the Agent container runtime system also includes:

[0012] When the Agent generates an interaction message with the user, it sends the interaction message to the Agent proxy module;

[0013] The Agent module receives the interaction message and sends it to the user.

[0014] Furthermore, the Agent container runtime system also includes:

[0015] When the Agent monitoring module detects that the current interaction has ended, it deletes the Docker container corresponding to the current interaction.

[0016] Furthermore, the Agent container runtime system also includes:

[0017] When the Agent monitoring module detects that the current interaction has ended, it stores the Agent's log file in the database.

[0018] Furthermore, the Agent container runtime system also includes:

[0019] The Agent monitoring module includes an Agent registry and a Docker image registry. The Agent registry registers Agent function and version description information, and the Docker image registry registers Docker image function and version description information.

[0020] This invention provides an Agent container runtime system. The system includes an Agent listening module, a proxy module, a Docker container, an Agent repository, and a Docker image repository. The Agent listening module listens for user requests in real time. Upon receiving a user request, it retrieves the required image from the Docker image repository and creates a Docker container based on the image. The Agent listening module then retrieves the required Agent code from the Agent repository based on the user request, loads the Agent code into the Docker container, and creates an Agent process within the Docker container. The Agent proxy module sends the user request to the Agent process, which includes an Agent framework module. The Agent framework module runs according to the Agent process's execution logic. This system avoids dependency conflicts between different Agents, improving resource utilization efficiency. Simultaneously, by forwarding user requests through the Agent proxy module, the accuracy and timeliness of request delivery are ensured, resulting in a more stable and flexible overall system operation and reduced maintenance costs. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic block diagram of an embodiment of the Agent container running system provided in this invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0025] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0026] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0027] Please see Figure 1 This is a schematic block diagram of an Agent container runtime system provided in an embodiment of the present invention. The system includes an Agent listening module, an Agent proxy module, a Docker container, an Agent repository, and a Docker image repository, wherein:

[0028] The Agent listening module listens for user requests in real time. When a user request is received, it retrieves the image required by the Agent from the Docker image repository and creates the Docker container based on the image. The Agent listening module also retrieves the Agent code to be run from the Agent repository according to the user request, loads the Agent code into the Docker container, and creates an Agent process in the Docker container. The Docker image repository stores multiple image images with different functions and versions, and the Agent repository stores Agent code with different functions and versions.

[0029] The Agent proxy module sends the user request to the Agent process, and the Agent process includes the Agent framework module;

[0030] The Agent framework module runs according to the running logic of the Agent process.

[0031] In this embodiment, the Docker image repository stores multiple images with different functions and versions. These images contain various dependencies and system configurations required for the Agent to run. Each image corresponds to a specific runtime environment, which can meet the running requirements of different Agents. The Agent repository stores Agent code with different functions and versions, which is the core of the Agent to implement specific functions.

[0032] The Agent listening module plays a crucial role in receiving and processing user requests. It is constantly listening, waiting for user requests to arrive. Upon receiving a request, the Agent listening module first analyzes it to determine the image corresponding to the Agent that needs to be run. Then, it retrieves the image from the Docker image repository and creates a Docker container based on that image.

[0033] After the Docker container is created, the Agent listening module retrieves the required Agent code from the Agent repository based on the user request and loads the Agent code into the created Docker container. Then, an Agent process is created inside the Docker container, allowing the Agent code to run in an independent container environment.

[0034] The primary function of the Agent module is to handle the transmission of user requests. After the Agent listening module creates the Agent process, the Agent module accurately sends user requests to that Agent process. The Agent process contains the Agent framework module, which is the core support component for Agent operation. This framework module processes and responds to received user requests according to the Agent process's own operational logic, thereby enabling the Agent's various functionalities.

[0035] The Agent monitoring module includes an Agent registry and a Docker image registry. The Agent registry registers Agent functionality and version description information, which allows for a quick understanding of the characteristics and version differences of various Agents. The Docker image registry registers Docker image functionality and version description information, facilitating accurate matching of the image environment required for Agent operation.

[0036] Furthermore, the Agent process also includes code tools and other tools. The Agent framework module runs according to the execution logic of the Agent process, including:

[0037] The Agent framework module interacts with a pre-built Large Language Model (LLM), parses the interaction information returned by the LLM, obtains the parsing results, and calls code tools or other tools based on the parsing results. For code tools, code can be run directly in a Docker container. All tools can interact with databases (DBs), the web, and other environments. The Agent framework interacts directly with the database, saving session history.

[0038] Furthermore, when the Agent generates an interaction message with the user, it sends the interaction message to the Agent proxy module; the Agent proxy module receives the interaction message and sends it to the user, thus realizing a smooth interaction between the user and the Agent.

[0039] Furthermore, when the Agent monitoring module detects that the current interaction has ended, it will perform two operations: first, delete the Docker container corresponding to the current interaction to release system resources and avoid resource waste; second, store the Agent's running log files in the database. These log files contain various key information during the Agent's operation, providing an important basis for subsequent system maintenance, troubleshooting, and performance analysis.

[0040] Through the above system, each Agent runs in an independent Docker container, and the Agents are isolated from each other, effectively avoiding dependency conflicts. At the same time, the centralized management of Agent code and images through Agent repositories and Docker image repositories, as well as the collaborative work of Agent listening modules and Agent proxy modules, makes Agent deployment, startup, operation, and management more efficient and flexible, greatly improving the stability and maintainability of the Agent running system.

[0041] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An Agent container runtime system, characterized in that, This includes an Agent listening module, an Agent proxy module, a Docker container, an Agent repository, and a Docker image repository, among which: The Agent listening module listens for user requests in real time. When a user request is received, it retrieves the image required by the Agent from the Docker image repository and creates the Docker container based on the image. The Agent listening module also retrieves the Agent code to be run from the Agent repository according to the user request, loads the Agent code into the Docker container, and creates an Agent process in the Docker container. The Docker image repository stores multiple image images with different functions and versions, and the Agent repository stores Agent code with different functions and versions. The Agent proxy module sends the user request to the Agent process, and the Agent process includes the Agent framework module; The Agent framework module runs according to the running logic of the Agent process.

2. The Agent container runtime system as described in claim 1, characterized in that, The Agent process also includes code tools and other tools. The Agent framework module runs according to the execution logic of the Agent process, including: The Agent framework module interacts with a pre-built large language model, parses the interaction information returned by the pre-built large language model, obtains the parsing result, and calls the code tool or other tools based on the parsing result.

3. The Agent container runtime system as described in claim 1, characterized in that, Also includes: When the Agent generates an interaction message with the user, it sends the interaction message to the Agent proxy module; The Agent module receives the interaction message and sends it to the user.

4. The Agent container runtime system as described in claim 1, characterized in that, Also includes: When the Agent monitoring module detects that the current interaction has ended, it deletes the Docker container corresponding to the current interaction.

5. The Agent container runtime system as described in claim 1, characterized in that, Also includes: When the Agent monitoring module detects that the current interaction has ended, it stores the Agent's log file in the database.

6. The Agent container runtime system as described in claim 1, characterized in that, Also includes: The Agent monitoring module includes an Agent registry and a Docker image registry. The Agent registry registers Agent function and version description information, and the Docker image registry registers Docker image function and version description information.

Images