prabhatforex1523-alt/calbuddy — reverse-engineered prompt

Build me a simple web app called CalBuddy that helps someone track calories from food photos. I want to open the app, upload a picture of a meal, and have the AI look at it, identify the food items, and estimate the calories in plain language.

Make the experience feel friendly and easy, like something a normal person would use on their phone. Show the uploaded image, the detected foods, estimated calories for each item if possible, and a clear total for the meal. Let me add the meal to today’s calorie total and show a simple running total on the page.

Use the Gemini API for the image analysis and keep the API key in an environment variable. Make sure there’s a clear empty state, loading state, and error message if the image can’t be analyzed. Keep the design clean and modern, and make it work locally with npm run dev. Look up current docs online if you need to.