analyze user's prompt and output relevant to it json in following format.
{“enabledSources”:  array of enum Camera, Screen
       “enabledPipelines:  array of enum HandRecognizer, HumanBodyRecognizer, TextRecognizer, CatRecognizer
       “webviewTag” enum CameraOverlay, BrowserOverlay, PiPView, SideBarTop, SideBarBottom}
user prompt: "place a button to picture in picture, on its bottom, when user clicks it - save all text"
do not output any explanations!

Write a code in JavaScript, that’s being executed in environment where following events are being emitted:
  “onScreenData” on document with object { screenData: pixelBuffer, textRecognizer: {boxes: [ {x, y, width, height, “recognized text in box1”}, {x, y, width, height, “recognized text in box2” }…]}}. Note that the only way to receive a screen data is to subscribe to this event
Following functions are already attached to document object:
  submitLabel(label: String, callback: (error)->void)
User prompt: “place a half transparent button over each recognized box of text. When user click on it, show a dropdown with options: good, bad, average. When label is submitted show alert”
Output separately minified html of div#content, css and js like that
😳HTML
```
(Only div#content and its children)
```
🧐JS
```
(JS without any comments, prettified)
```
😎CSS
```
(Minified css)
```

Do not output any explanations!





modify previous response according to user prompt. output only modified lines in DIFF format with line numbers, inside of 
😳HTML
```
html diff goes here
```
🧐JS
```
js diff goes here
```
😎CSS
```
css diff goes here
```
as needed.
user prompt is: "make buttons blue, barely visible, with corner radius 0.4"
Do not output any explanations!