Cursor's Projects: One AI to Rule Thousands of Agents

On September 10, Cursor—the AI code editor that's been turning heads—launched a beta feature called Projects. It's designed to tackle the heavy lifting in software development: feature builds, large-scale migrations, and even entire applications. The secret sauce? A coordinator agent that orchestrates thousands of sub-agents running in parallel. No more babysitting individual agents; you just give the orders.

How It Works Under the Hood

Projects brings three key capabilities to the table.

First, default cloud execution with on-demand local testing. Each Project lives on its own cloud computer, so even if you shut your laptop, the work keeps chugging along. That means you can scale sub-agents far beyond your local machine's limits. When you need to test locally, the coordinator spins up a local agent automatically.

Second, shared system context. Every Project maintains a set of continuously synced files where agents stash their research, codebase insights, and preferences. Over time, the coordinator gets smarter and faster.

Third, an intelligent subscription mechanism. The coordinator can watch Slack channels, run on a schedule, or track every pull request. When it spots a signal, it takes action—no constant manual prompts required.

What You Can Actually Do With It

Projects shines in three practical scenarios.

Feature development: Agents start by researching and building shared context. Then the coordinator drafts a plan and assigns sub-agents to implement and test in parallel. After launch, it keeps an eye on logs and squashes bugs.

Code migration: Think large-scale framework upgrades or style replacements—easy to start, hard to finish. The system can grind through hundreds of PRs. Early on, humans review strictly; later, as the solution matures, the coordinator takes the wheel.

Ongoing "gardening" maintenance: Perfect for code quality monitoring or regression testing. For example, it can automatically extract design system components and generate rules online for efficient automation governance.

The Numbers Don't Lie

Cursor's team has been testing Projects for months, using it on migrations and feature releases involving hundreds of PRs. The results? New users merged 30% more PRs, while those who leaned heavily on Projects saw a staggering six-fold increase in merged PRs.

Projects is in beta and rolling out to all users starting today. If you're tired of juggling agents, this might be your new best friend.

Key Points

Coordinator agent manages thousands of sub-agents in parallel.
Cloud execution keeps work going even when your laptop is closed.
Shared context makes the coordinator smarter over time.
Intelligent subscriptions monitor Slack, schedules, and PRs proactively.
Early results: 30% more PRs merged for new users; 6x for heavy users.
Beta rollout begins today.

Cursor's Projects: One AI to Rule Thousands of Agents