# Adom Desktop Control

A skill pack of **tested patterns for driving an Adom user's real machine** from the
cloud AI through Adom Desktop: their logged-in Chrome/Edge, headless pup browsers,
native app windows, the Hydrogen editor, sibling Claude tabs, plus a full
screen-recording and video post-production pipeline.

Every skill here was distilled from a real session and labels **what works, what fails,
and why**, so the next agent doesn't relearn it the hard way.

```bash
adompkg install adom/adom-desktop-control
```

## Pick the right surface

| You want to drive... | Use | Skill |
|---|---|---|
| The user's REAL logged-in Chrome/Edge | `nbrowser_*` (extension, CDP) | **native-browser-control** |
| A throwaway / clean-to-record browser | `browser_*` (pup) | **pup-browser-control** |
| A native app window (Fusion, dialogs) | `desktop_*` (UIA + SendInput) | **desktop-ui-control** |
| The Hydrogen editor / its Claude panel | (mostly can't, read why) | **hydrogen-web-control** |
| A sibling Claude tab | depends where it runs | **claude-control** |
| Capture the screen to video | `desktop_record_*` + `desktop_caption` | **screen-recording** |
| Cut raw clips into a finished demo | ffmpeg + `video-post` | **video-post-production** |

## The rules that apply everywhere

- **Background by default, never steal focus.** CDP eval, UIA Invoke, and window
  screenshots all run on a background window. Only OS SendInput needs foreground, so
  avoid it when a CDP/UIA path exists.
- **Drive it yourself, including login.** You can take the user's real browser through an
  OAuth sign-in end to end. Never ask them to click or log in.
- **Upload via DataTransfer, not the OS file picker.** Inject the bytes, set the
  `<input type=file>`. Works in pup and nbrowser.
- **Recording captures whatever is topmost.** Re-foreground the target before every
  visible step or you record the wrong window.
- **Verify by screenshot.** OS input silently no-ops on many webview editors; confirm the
  text/state actually landed before building on it.

## What's inside

- **native-browser-control** - the user's real Chrome/Edge over CDP (incl. `browser:"edge"`,
  DataTransfer uploads, driving OAuth login, the cross-origin / no-frame-targeting limits).
- **pup-browser-control** - headless Puppeteer browsers for anonymous or clean-to-record work.
- **desktop-ui-control** - native windows via UIA (`find_control`/`ui_click`) and SendInput
  (`click`/`type`/`press_key`), with screenshot-to-coordinate mapping.
- **hydrogen-web-control** - what's reachable in the Hydrogen editor and why its embedded
  Claude input is not.
- **claude-control** - when to delegate to a sibling Claude vs do it yourself.
- **screen-recording** - WGC single-window capture (`desktop_record_window_start`):
  background, no picker, no banner, occlusion-safe, the right way to record a window
  (whole-desktop capture and its foreground theft is the wrong default).
- **video-post-production** - crop, speed up, brand-font captions, Ken Burns on stills,
  TTS voiceover, concat, and wiki-ready keyframes.

## Pairs with

[adom-desktop](https://wiki.adom.inc/adom/adom-desktop) (the app + relay),
[adom-browser-extension](https://wiki.adom.inc/adom/adom-browser-extension) (native CDP),
[adom-tts](https://wiki.adom.inc/adom/adom-tts) (voiceover),
[demo-recording](https://wiki.adom.inc/adom/demo-recording) (the broader demo workflow).