GPT Prompt Lab

A local prompt engineering sandbox powered by Ollama, Streamlit, and optional GPT comparisons built for testing, comparing, and improving prompt responses across multiple LLMs.

Features

• Multi-Prompt Comparison
  Enter multiple prompts and compare LLM responses side by side.

• Ollama Model Integration
  Works offline with models like Mistral, LLaMA, Gemma, etc.

• Prompt Presets & Templates
  Load preset prompts for specific tasks like summarization, Q&A, code generation.

• Light/Dark Mode
  Easily switch contrast and text styling from the sidebar.

• Response History & Feedback
  View past results and prompt experiments.

• “Make Prompt Better” Button
  Suggest improved prompt variants with an LLM.

• AI Summary of Responses
  GPT-4 generates insights and summaries from responses.

• Interactive Diff Viewer (Advanced Mode)
  Toggle a side-by-side diff to visually compare response differences.

Screenshots

Installation

git clone https://github.com/yourusername/gpt-prompt-lab.git
cd gpt-prompt-lab
python -m venv venv
.\venv\Scripts\activate
pip install -r requirements.txt
streamlit run app.py

Requires Python 3.9 or later. Ollama should be running while using local models.