Mahal Systems WebMall
Fits comparison-shopping agents; borrow the multi-shop offline sandbox built from Common Crawl product pages and the basic / advanced task split.
What it asks the agent to do
- find a specific product across four shops and compare prices
- find substitutes or compatible products from a vague description
- add to cart and check out
What to borrow from it
sandbox designcase template
Caveats
Small; shops are simplified; license not stated; best agents 62 to 75 percent on basic tasks and 19 to 41 on advanced.
Details
| Workloads | shopping, retail, web browsing |
|---|---|
| Interaction | single turn |
| Environment | browser |
| End state checked | Yes |
| Tools | four self hosted simulated shops |
| Grading | programmatic check, state diff |
| Size | 91 |
| Data | mixed |
| Personal data | none |
| License | unknown |
| Harness | own repo |
| Status | active |