A single AI coding chat can help with a small fix. But large features, code migrations, and long-term maintenance require more than one prompt—and often more than one agent.
Cursor Projects is designed for that larger class of software work. Cursor describes Projects as a way to maintain context for months, delegate tasks to thousands of subagents, and run recurring work without waiting for a new prompt.
What Is Cursor Projects?
Cursor Projects is a long-running workspace managed through a coordinating agent. Developers describe the work they want to complete, while the coordinating agent plans tasks and delegates implementation and testing to other agents.
The coordinating agent does not write code itself. Its role is to organize the work and direct coding agents, allowing it to continue responding while other agents execute assigned tasks.
Cursor announced Projects as a beta feature on September 10, 2026, with a gradual rollout to all users.
How Cursor Projects Works
Cloud-first execution with local support
Projects run in the cloud by default, according to Cursor. Each Project operates on its own computer, so work can continue even when a developer closes their laptop.
When a task needs access to the developer’s local machine, the coordinating agent can start a local agent for testing or other local work.
Shared context across agents
Each Project maintains a shared collection of files synchronized across the cloud and local machines used by its agents.
These files can contain research, generated artifacts, codebase knowledge, testing instructions, and information about a developer’s preferred workflow. Cursor says this allows later agents to reuse information discovered by earlier agents instead of repeating the same investigation.
Event-based automation
Projects can also respond to external signals. Cursor says a coordinating agent can monitor Slack channels, run on a schedule, or track pull requests.
For example, it can take action when a pull request is opened or merged, including responding to continuous integration issues.
Cursor Agent vs Projects
The main difference is the level of abstraction. With a regular coding agent, the developer manages the task. With Projects, the developer defines a broader outcome while the coordinator breaks that work into tasks and manages the agents responsible for completing them.
Three Practical Use Cases for Cursor Projects
1. Large feature development
Cursor describes a workflow in which a Project first researches the existing system and records its findings as shared context. The coordinating agent then creates a plan and assigns different implementation and testing tasks to multiple agents.
After the feature is released, the same Project can continue monitoring logs and handling bug reports while retaining the context behind earlier technical decisions.
2. Code migrations
Migrations are another central use case. Cursor says it has used Projects across hundreds of pull requests for framework adoption and style-system replacement.
The process begins with a migration strategy. The coordinating agent then applies that strategy across the codebase in stages. Cursor describes a review model in which engineers inspect early pull requests closely and reduce manual review as the migration proves reliable.
3. Ongoing codebase maintenance
Projects can be used for maintenance work that continues over time, such as monitoring code quality and identifying regressions.
Cursor describes an internal design-system Project that scans new pull requests, identifies relevant components, and adds a lint rule when the same issue appears for a second time. Cursor says this Project is expected to process between 20 and 100 pull requests per day; this is an internal company example, not a guaranteed user result.
What Cursor Reports About Productivity
Cursor says it used Projects for several months to complete migrations involving hundreds of pull requests, maintain design-system consistency, and build Projects itself.
The company reports that new users increased their merged pull requests by 30%, while users who primarily used Projects reached six times their previous merge volume. These figures are Cursor’s own reported results; the announcement does not provide an independent benchmark or detailed methodology.
How to Start Using Cursor Projects
Cursor says users can create a Project from the left navigation bar and describe what they want to build.
The feature is intended for work that cannot be completed in a single chat, including multi-pull-request feature development, code migrations, and tasks that should continue while the developer is away.
Why Cursor Projects Matters
Projects move Cursor’s AI coding workflow from one-off assistance toward coordinated, long-running work.
The key idea is simple: the developer defines the outcome, while a coordinating agent organizes research, implementation, testing, monitoring, and follow-up tasks. Whether this approach delivers consistent results will depend on factors such as code review, testing quality, and the reliability of the agents involved. Cursor’s announcement introduces the product and its use cases but does not provide pricing, usage limits, or independent performance evaluations.
FAQ
What is Cursor Projects?
Cursor Projects is a long-running workspace that uses a coordinating agent to delegate software development tasks to multiple coding agents.
What does the coordinating agent do?
It plans and delegates work rather than writing code directly. Other agents handle implementation and testing.
Can Cursor Projects run when my laptop is closed?
Yes. Cursor says Projects run in the cloud by default. A local agent can be started when work requires access to the developer’s computer.
What are the main use cases?
Cursor highlights feature development, code migrations, and ongoing maintenance such as code-quality monitoring and regression detection.
Is Cursor Projects available now?
Projects entered beta on September 10, 2026, and Cursor says it is gradually rolling out the feature to all users.
