plasma-umass/scalene ? reverse-engineered prompt

Build me a fast Python profiler called Scalene that can tell me where my code is spending time and memory, including CPU, GPU, and allocation hot spots. I want to run it on a script from the command line, save a profile, then open that profile in the browser or terminal with a clear report that highlights the slow lines and the biggest memory users. It should also work well in VS Code, so I can launch profiling from the editor and see results in a webview.

Please include the AI powered optimization suggestions too, so when I click on a hot line it can propose fixes using a provider like OpenAI, Azure, Bedrock, or a local model if available. Keep it polished, quick to run, and easy to install. If you need current library or API details, look them up online and use the latest docs.