a prompt for ChatGPT that enables it respond on "Overlay green boxes over humans you see on the screen": "write a js code that will be run in CombineWebView with dimensions 768x1388. underneath it you user can see a human figure, code that you will write can receive the coordinates of the recognized hands as an array with objects with format: {x:, y:} on js event "received labeled data"

"create a text field and an preview box, each takes half of the screen. when text in text field is edited, markdown or mermaid or latex is rendered in the preview box. speech of user and editing instructions that you will receive are transformed into the code that manipulates text in the box accordingly to instructions"

for all available browser tabs: attach to each text view a code that monitors if user periodically enters text into it. find an element on the page that is being updated few seconds after user presses enter. then, display that you found such tab. then, marshall users speech input into that text field and updated element from the page - onto the overlay view

"respeat entering a random description of beautiful Midjorney image prompt, hit enter button and wait for update of the biggest image on the page until on screen human recognition will show exactly 7 humans. don't use '7' in the prompt. when picture with 7 people is detected copy third from the left, create a mail in Gmail tab to address standardtemplateconstruct_test_email@gmail.com, include both full and cropped pictures.

use built in console to checkout master of NodeJS and use creative approaches trying to port it to web asm