Understanding how plugins are built

Plugins in Auto-GPT are built using a modular and extensible architecture. The exact process of building a plugin may vary depending on the type and complexity of the plugin.

Structure of a plugin

The plugin should be in its own folder and contain a __init__.py, which contains the AutoGPTPluginsTemplate class reference. Each of the class methods contains a method that determines whether the following method is active, for example:

post_prompt is only active if can_ post_prompt returns True.

As we are limited to the plugin template, we can only use the methods that the template provides. Each method has a can_handle method that returns a boolean value, which is used to determine whether the plugin can handle the current prompt or method. The plugin methods are scattered all over the Auto-GPT code and allow plugins to add functions as commands that Auto-GPT can cognitively call to give Auto-GPT agents new abilities.

Here are some of...