Introduction

The Cheshire Cat is a ready-to-use AI micro-framework. Once installed and connected to a Language Model (LLM), it can be queried through APIs. These APIs return the responses provided by the LLM.

But this is just the beginning.

Previous Conversation History

All previous conversations are stored in a local database called episodic memory. When you ask a question, the Cat answers taking into account the past conversations.

Loading Documents

You can load text documents as well. These documents are also saved in a local database called declarative memory. When answering, the Cat will consider the information within these documents. Documents can be uploaded through the APIs or the Admin Portal.

The Rabbit Hole is the component responsible for the document ingestion.

Performing Actions

The Cheshire Cat isn't limited to just answering questions; it can also perform actions. You can write Python functions called Tools and have the LLM execute this code. The only limit to the Python code's capabilities is your imagination.

Extending the Core

Additionally, it's possible to customize the Cheshire Cat's core. In the main process flow, there are predefined adaptation points called Hooks. You can write Python functions that can be attached onto these Hooks. The attached code will be invoked during the flow's execution and can modify the Cheshire Cat's internal behavior, without directly modifying the core of the Cheshire Cat.

Tools and Hooks are packaged into Plugins that can be installed by placing files in a specific folder or using the Admin Portal. The Mad Hatter is the component that manages plugins.

If desired, you can publish your Plugins on the public registry. Other users will be able to install them with just a single click from the Admin Portal.

Admin Portal

A web portal for Admin users completes the framework. Using this portal, the admin can configure the settings, install plugins, upload documents and use it as a playground tool. You can chat with the Cheshire Cat, inspect its responses and directly query its memories.

Next step

In the next step, you will learn how to install the Cat, set the LLM and the basics of this all.

We will be transforming the Cat into a sock seller. More in detail, we will upload some knowledge (documents) about socks knitting. Also, the Cat will be able to tell the price of socks according to the requested color (using a Tool). In the end, we will transform the sock seller into a poetic socks seller, changing its personality (using a Hook).

The example is light and fun, it should give you an idea of what is possible.