@steipete
Captures windows with OCR so your agent can read what is happening on-screen.
when context lives inside a desktop app, not a browser tab
exec