It's both amazing and painful to watch codex use browser + computer use to open…
- Source
- x.com
- Date
It's both amazing and painful to watch codex use browser + computer use to open Chrome, go to my PR, tap on comment and wrangle with the macOS picker - all TO UPLOAD AN IMAGE. GitHub has no API doesn't stop anyone. I let my codex run in VMs so they don't steal app focus.
Context
A coding 's work can extend into ordinary desktop controls. Peter Steinberger described watching OpenAI's Codex navigate to a pull-request comment and operate a file picker to upload an image. The attached screenshot shows the picker open over the comment editor.
Steinberger says their Codex sessions run in virtual machines so they do not take app focus. The example exposes a practical constraint of computer use: an agent interacting with windows can compete with the person using the same desktop. Running the agent in a virtual machine is the arrangement Steinberger describes for avoiding that interference.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Checking sign-in…
Loading comments…