gpt-prompt-engineer is an experimental prompt optimization toolkit built around notebooks and LLM-assisted evaluation. It lets users describe a task, provide test cases, and generate many candidate prompts automatically. The system then tests those prompts against the examples and ranks their performance through an ELO-style scoring process. The repository includes versions for general prompt generation, classification tasks, Claude-based workflows, and model-to-model prompt conversion. It is meant to make prompt engineering more systematic than manual trial and error. The project is useful for AI builders who need repeatable prompt selection before deploying a task-specific assistant, classifier, or automation workflow.
Features
- Automatic prompt generation
- Prompt testing against cases
- ELO-style prompt ranking
- Classification task notebook
- Claude and GPT workflow variants
- Model-to-model prompt conversion notebooks