Mahal Systems WorkArena
Fits browser agents working inside ServiceNow or similar ITSM/HR portals. Borrow the task templates (33 atomic tasks instantiated into thousands of variants) and the per-task validate() reward on a live instance.
What it asks the agent to do
- fill in a form in an enterprise web app
- filter and sort a list view
- order an item from a service catalog
- navigate menus and read a dashboard
- look up a knowledge base article
What to borrow from it
sandbox designcase templatebaseline
Caveats
Needs a ServiceNow developer instance and a gated Hugging Face dataset; L1 tasks are now mostly solved by strong agents, so use WorkArena++ for a current signal.
Details
| Workloads | it ops, workflow automation, web browsing, computer use |
|---|---|
| Interaction | single turn |
| Environment | browser |
| End state checked | Yes |
| Tools | browser actions via BrowserGym |
| Grading | programmatic check |
| Size | 33 |
| Data | synthetic |
| Personal data | none |
| License | Apache-2.0 |
| Harness | own repo |
| Status | saturated |