Introduction to Plug-ins

On the agent development platform, plug-ins are important extensions of large model capabilities. In modular mode, plug-ins can provide more professional skills and complex task processing capabilities for large models, meeting user requirements more efficiently in diverse scenarios.

Applications interconnected with plug-ins can provide capabilities that are not supported by the large model. Plug-ins provide various external service APIs. When a task is executed, the model detects the applicable plug-ins based on the prompts, automatically invokes the plug-ins, obtains the result from the external service, and returns the result. This design enables the Agent to intelligently process complex tasks and even solve problems across domains, implementing automatic processing of complex problems.

The agent development platform supports two types of plug-ins:

• Preset plug-ins: These plug-ins are provided by the platform for developers and users. They are ready to use and do not require custom development. For example, the Python interpreter plug-in provided by the platform can automatically generate Python code based on the questions entered by users, and execute the code to obtain the result. This plug-in provides powerful computing, data processing, and analysis functions for agents. You only need to add the plug-in to an application to extend its functions.
• Custom plug-ins: To meet more personalized requirements, the platform allows developers to create custom plug-ins. APIs can be quickly created as plug-ins through configuration and called by agents. In this way, developers can add dedicated functions to applications based on specific requirements.