Mahal Systems UI-CUBE
Fits RPA-style enterprise agents; borrow the split into simple interactions versus multi-step workflows and the resolution and widget-variant sweep.
What it asks the agent to do
- perform a single UI interaction (click, select, type) across widget variants and resolutions
- copy data between applications
- complete an enterprise application scenario end to end
What to borrow from it
taxonomygrading design
Caveats
Vendor-built; apps are synthetic stand-ins; repo has little documentation and no stated license; agents score 67 to 85 on simple tasks but 9 to 19 on workflows.
Details
| Workloads | computer use, workflow automation, crm |
|---|---|
| Interaction | single turn |
| Environment | desktop |
| End state checked | Yes |
| Tools | VNC desktop, X11 actions |
| Grading | state diff, programmatic check |
| Size | 226 |
| Data | synthetic |
| Personal data | none |
| License | unknown |
| Harness | own repo |
| Status | active |