Prompt2Data is a conversational AI tool that turns natural language prompts into structured data specifications. It helps automate dataset generation by capturing goals, schema, and metadata — starting with an intelligent prompt interpreter powered by Claude 4.

Devpost Link - https://devpost.com/software/prompt2data-autonomous-data-pipeline?ref_content=my-projects-tab&ref_feature=my_projects

Think of it as ChatGPT for dataset design and pipeline automation.

• 🧠 Prompt-to-JSON task parsing using Claude 4 (Anthropic)
• 💬 Continuous chat context using Streamlit
• 📦 Outputs a fully structured spec: goal, data fields, location, time, format
• 🔁 Remembers past instructions, updates specs incrementally
• ✅ Plug-in ready for multi-agent expansion (dataset search, cleaning, etc.)

• Python 3.9+
• Streamlit (for UI)
• Anthropic Claude API (for intent parsing)
• Dotenv (for API key handling)

• Python 3.9+
• A GitHub account (for cloning)
• A free Anthropic Claude API Key

git clone https://github.com/your-username/Prompt2Data.git
cd Prompt2Data

python -m venv venv
venv\Scripts\activate

pip install streamlit anthropic python-dotenv

In your root folder, add a new file called .env:

CLAUDE_API_KEY=sk-ant-xxxxxxxxxxxxxxxxxxxx

Prompt2Data/
├── venv/
├── .env
├── intent_ui.py
├── intent_agent.py
├── requirements.txt

streamlit run intent_ui.py

It will open a browser window at:
http://localhost:8501

I want data on housing prices and crime rates in California
Make it from 2010 to 2020
Output should be a CSV
Add air quality if available

Each message updates your task spec!

PRs and ideas welcome — DM us or submit issues if you’d like to extend agents or plug in other APIs!