ImPromptr

# Prompt Engineering Over And Over **Story Time** I am very particular regarding what and how I use AI. I am not saying I am a skeptic; quite the opposite actually. I know that AI/LLM tools are capable of great things **AS LONG AS THEY ARE USED PROPERLY**. For the longest time, whenever I needed the optimal results with an AI tool or chat-bot, this is the process I would go through: 1. Go to the Github repo of [friuns2/BlackFriday-GPTs-Prompts](https://github.com/friuns2/Black...) 2. Go to the file [Prompt-Engineering.md](https://github.com/friuns2/Black...) 3. Select the [ChatGPT 4 Prompt Improvement](https://github.com/friuns2/Black...) 4. Copy and paste that prompt over to my chatbot of choice 5. Begin my prompting my hyper-specific, multi-paragraph prompt 6. Read and respond to the 3/6 questions that the chat-bot came up with so the next iteration of the prompt will be even more specified. 7. After many cycles of prompting, re-prompting, and answering, use the final prompt that was refined to get the ultimate optimal result While this process was always exhilarating to repeat multiple times a day, for some reason I kept yearning for a faster, more efficient, and better organized method of going about this. Coincidentally, winter break began for me around November, I had over a month of free time, and a menial task that I was craving to overengineer. The result, [ImPromptr](https://impromptr.com), the iterative prompt engineering tool to help you get your best results. It doesn't just stop at prompts, though, as each chat instance where you are improving your prompts has the ability to generate markdown context files for your esoteric use cases. In many cases online, you can almost always find a prompt that you are looking for with 98.67% accuracy. With ImPromptr, you don't have to sacrifice your precious percentage points. Each saved prompt allows you to modify the prompt in its entirety to your hearts desire **WHILE** maintaining a strict version control system that allows you to go through the lifecycle of the prompt. Once again, I truly do believe that AI assisted *everything* is the future, whether it be engineering, research, education, or more. The optimal scenario with AI is that given **exactly** what you are looking for, the tools will be able to understand exactly what it needs to do and execute on it's task with clarity and context. I hope this project that I made can help everyone out with the first part.