Loading ToolBench
An AI agent uses a real GNOME/Wayland desktop - pressing actual widgets, not guessing at pixels - optionally on an invisible second GNOME session, so your own screen stays yours.
Scoring was not performed
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 23 | 2026-07-28+ | v2 |