A
licenseA
qualityB
maintenanceEnables AI agents to test web screens by reading a text map of elements (roles, names, values, states, positions) instead of screenshots, and to act by intent such as clicking "Save" or filling "Email" rather than guessing coordinates. Agents can compare maps before and after an action to see what appeared, disappeared, changed or moved, inspect network requests, page errors and native dialogs, and detect visual issues like overflow, cut text, low contrast or small touch targets.
6
MIT