Thank you for your interest in contributing to AI Reply Index! This project is a hybrid of a desktop app and an open repository for collecting well-formatted AI prompts and responses from a wide range of models. All contributions help grow a reliable benchmark archive for the AI community.
We require that contributors use the AI Reply Index app to ensure a consistent format for all entries. The app:
- Generates folder structures automatically
- Formats Markdown using
mdformat - Tags models and metadata properly
- Supports drag-and-drop for DOCX input
- Supports multiple response versions per prompt
Install the app using these instructions before proceeding.
Here’s how to add your prompt + responses to the index:
-
Fork the repository
-
Clone it locally
git clone https://github.com/YOUR_USERNAME/ai-reply-index.git cd ai-reply-index -
Install the app
- With Poetry:
poetry install
- With Poetry:
-
Run the app
python app/main.py
-
Use the "New Entry" tab to input:
- The prompt
- The model name
- The AI response
- Optional tags
- Choose to save as Markdown + JSON
-
Verify folder output under
prompts/Each entry folder should include:prompts/YYYY-MM-DD_slug/ ├── prompt.md ├── MODEL_NAME.md └── metadata.json
-
Commit your prompt folder
git checkout -b add-my-prompt git add prompts/2025-04-19_my_test_prompt/ git commit -m "Add: AI response to my_test_prompt using Claude" git push origin add-my-prompt -
Submit a Pull Request
- Title your PR clearly
- Mention the models used and what the prompt tests
✅ Use meaningful prompt slugs
✅ Add clear tags (e.g., reasoning, ethics, summarization)
✅ Test multiple models on the same prompt
✅ Prefer Markdown-friendly input
✅ Avoid long preambles in model replies (trim where needed)
If you'd like to preview your contribution as an HTML archive:
python scripts/export_static_site.pyHTML will be generated to dist/index.html for easy sharing.
All contributions are licensed under the MIT License. By submitting a pull request, you agree that your contributions will be included under this license.
Let’s build a transparent, open-source AI prompt archive — one entry at a time. Thank you! 🙏