milanmarkovic-source/creative-fatigue-monitor ? reverse-engineered prompt

Build me a simple Python tool that helps me spot creative fatigue in paid social ads much faster.

I want to upload or generate a dataset, compare the most recent 7 day window against the previous 7 day window, and only flag ads when all three things happen at once, frequency goes up, CTR goes down, and CPA goes up. Please include a minimum impressions filter so low volume ads under 1,000 impressions do not get flagged by accident. I also want the results sorted by the worst CPA increase so the most urgent ones are at the top.

Make it easy to use in a small Streamlit app with a button to generate mock test data and a clear table of flagged ads, plus a way to run the same logic in a headless script for scheduled checks. If it helps, look up current docs online for any Streamlit or pandas details you need.