HyperPrompter

HyperPrompter is a desktop idea-exploration studio where the interface itself acts like an AI thought partner, not just a chat box. Instead of linear prompt-response threads, users orchestrate multiple personas, regenerate prompts from blended perspectives, evaluate output against custom criteria, and extract reusable instructions that recursively improve future prompts.
The UI creates an innovative way to start exploration and help get to better prompts through self-learning and persona advisors. Instead of constant back and forth text were the conversation can get polluted with off-topic statements, HyperPrompter iterates on improving through advisors and criteria for evaluation.

We used Electrobun to build a desktop reactjs app and using anthropic models for generating structured data validated with zod.

just a empty github repo