leigest519/ScreenCoder ? reverse-engineered prompt

Build me a simple tool that takes a UI screenshot or design mockup and turns it into clean, editable HTML and CSS.

I want to be able to drop in one image, run it locally, and get a result that keeps the layout, spacing, text, buttons, and image areas as close to the original as possible. If something is an image in the screenshot, make it a real image placeholder at first, then swap in the cropped image in the final output. The code should be easy to customize after generation, not just a flat screenshot clone.

Please wire up the full flow so it can detect the UI pieces, map them into page regions, generate the HTML, and produce a final output file I can open in a browser. If there are model settings or API keys needed, make it easy to configure. Use the project’s existing Python setup, and if needed look up current docs online so the generation part works well.