I created (: Smile after over two years of iterating on how to structure prompts. As I often iterate with Chat models (like ChatGPT, Claude, Perplexity or Kimi), use a variety of inference API’s, and open source models, I needed something that would work across all of them without compiling into a different language.
When I looked around, I found some great prompt engineering tools, especially for leveraging programming (Like POML), but most instruction languages seemed to be based on existing markups that people were already familiar with. Like a language in react components.
It wasn’t satisfying to me. So I made (: Smile! As the backbone of my own LLM apps, agents and prompts I’m ready to open it up to feedback and ideas for extending it into something that you would use.
I created (: Smile after over two years of iterating on how to structure prompts. As I often iterate with Chat models (like ChatGPT, Claude, Perplexity or Kimi), use a variety of inference API’s, and open source models, I needed something that would work across all of them without compiling into a different language.
When I looked around, I found some great prompt engineering tools, especially for leveraging programming (Like POML), but most instruction languages seemed to be based on existing markups that people were already familiar with. Like a language in react components.
It wasn’t satisfying to me. So I made (: Smile! As the backbone of my own LLM apps, agents and prompts I’m ready to open it up to feedback and ideas for extending it into something that you would use.
Let me know what you think!