Background

I have used generative AI tools like Gemini and ChatGPT for a while. Over time, I’ve created several prompts for various tasks.

I started developing an application to collect, version control, and manage these prompts a few months ago. Initially, the application was designed as a prompt library, but I later integrated it with generative AI APIs.

There are several reasons:

• Prompts can be modified and executed with a single click.
• The results are influenced by my custom system prompts and not the default system prompt.
• Unlike most tools that retain history for only a few weeks, I can keep my results as long as possible.
• I can switch between different vendors (Like Claude, Open AI, or Gemini) using a drop-down box.
• I can add any features I want to (For example, custom AI agents and tooling).

Theme

The application applies a Game theme. 
There is a village (application) where scholars (prompts) live. I can recruit or hire scholars to perform quests (requests) using scrolls (generative AI APIs). And the results (conversations, images, voices) are stored in the Great Library.

Scrolls (Vendors)

The application is designed to support multiple vendors through a common interface. This provides a multi-vendors, multi-modals, and cross-vendors execution environment.

Scholars (Prompts, Agents, Audits, Instructions)

Scholars can perform a 1-time execution or conversions. 

Scholar consists of system prompts, audit logs, instructions, and optional agents.

Running Quests

1) Text to text

2) Text to Image

3) Text/Audio to Text

4) Text/Image to Text