Introduction
Cursor's cloud agents (originally called background agents) run in remote machines, so a developer can hand off a task, close the laptop and come back to a finished branch. With Cursor 3's Agents Window, many developers now run several agents in parallel across local worktrees and the cloud. That creates a new job: supervising agents when you are not at your desk. An agent may finish, get stuck waiting for input, fail its tests or open a pull request while you are on a train or in a meeting.
This question asks us to design a mobile experience for that job. I'll clarify scope, explain why it matters, pick a user segment, map pain points, design a solution and define metrics.
For agent products, the hard design problem is trust: how much can someone approve from a 6-inch screen without reading every line? Make that tension the centre of your answer.
Step 1
Clarify Questions (3 minutes)
Propose the Goal
Let developers keep cloud agents moving when they are away from their desk, so less agent time is wasted waiting on a human, without lowering code quality or security.
Define the Scope
In scope: notifications, agent status, reviewing summaries and diffs, replying to agents, approving pull requests within policy, starting simple tasks. Out of scope: full code editing on a phone.
Practice similar questions
Subscribe to access the full answer