Cursor Projects: what it is and how the coordinator delegates tasks
MarkdownCursor Projects, as announced by Cursor: how the coordinator delegates tasks, where shared context is stored, how Subscriptions work and what this beta is for.
Read full article
On September 10, 2026, Cursor announced the beta version of Projects—a feature for large tasks with shared context that persists for months.
A coordinator leads the work in Projects. It does not write code itself: it plans the work and delegates parts of it to other agents. A project has a shared set of files containing the material accumulated along the way. You can also ask the coordinator to watch for events and respond without a new prompt. Below is a look at the capabilities and limitations described in Cursor's announcement.
Ailoot reference documentation: ailoot.ai/docs/.
What is Cursor Projects, in plain language?
Key takeaway
Cursor Projects is a feature for working on a large task: a new feature, a migration or an entire application. A coordinator plans the work and distributes it among implementation agents. The coordinator does not write code itself. According to the September 10, 2026 announcement, Projects is available in beta from Cursor's left navigation panel; access is gradually being rolled out to all users.
Cursor's official announcement describes the purpose of the feature as follows:
Projects lets you take on larger bodies of work, such as a feature, a migration, or a full app. It maintains context over months of work, delegates tasks to thousands of subagents, and performs recurring work without being prompted.
In other words: Projects lets you take on large tasks, such as a new feature, a migration or an entire application. It maintains context over months of work, delegates tasks to thousands of subagents and performs recurring work without a user prompt.
How does the coordinator differ from a regular Cursor agent?
Key takeaway
A project's coordinator plans the work, distributes it among implementation agents and brings the finished work to you for review. It does not write code itself. Coordinators create and manage agents on your behalf, running as many in parallel as the work requires.
Here is how Cursor describes the division of roles:
The coordinator agent in a project doesn't write code itself; it plans the work, delegates it to agents that implement it, and brings the finished work back to you to check. Coordinators create and manage agents on your behalf, running as many in parallel as the work needs.
In other words: a project's coordinator does not write code itself. It plans the work, delegates it to implementation agents and brings the finished work to you for review. Coordinators create and manage agents on your behalf, running as many in parallel as the work requires.
You can compare the division of roles to a foreman and a crew: the foreman plans the work and assigns people to carry it out. In Projects, those carrying it out are AI agents managed by the coordinator. The coordinator still brings the finished work to you for review. This model resembles the “orchestrator and workers” arrangement in multi-agent systems—see the guide to distributing tasks among agents.
The project runs in the cloud, but the coordinator can launch a local agent to test something on your machine:
A Project runs on its own computer in the cloud, so closing your laptop doesn't stop it. When something needs testing on your machine, the coordinator spins up a local agent to run it there.
In other words: the project runs on its own cloud computer, so closing your laptop does not stop it. When something needs testing on your computer, the coordinator launches a local agent there to perform that check.
How does a project's shared memory work?
Key takeaway
Each project has a set of files that sync across all the cloud and local machines its agents use. Agents put their findings, artifacts and what they learn about the code and your preferred way of working into those files. If one agent figures out how to test a service, future agents can use those instructions.
Cursor calls this mechanism Shared context:
Each Project maintains a set of files that sync across every cloud and local machine its agents use. Agents add research and artifacts, along with what they learn about the codebase and how you prefer work to be done. If one agent figures out how to test a service, for example, every future agent can use those instructions. The shared context grows with the Project, making the coordinator more effective over time.
In other words: each project keeps a set of files that sync across all of its agents' cloud and local machines. Agents add research results, artifacts and information about the code and your work preferences. If one agent finds out how to test a service, every future agent can use its instructions. Shared context grows with the project, which Cursor says makes the coordinator more effective over time.
The coordinator can draw on accumulated material. This is close to the idea of a Second Brain, but here it means a project's shared memory in Cursor rather than your personal memory. Agents add to it themselves as they work.
If you are interested in building similar personal memory in Claude Code, there is a separate guide to the Second Brain in Claude Code. It covers the related idea of accumulating material for later work.
Accumulating context across tasks is part of context engineering: managing what goes into AI memory and where it goes.
What are Subscriptions in Cursor Projects?
Key takeaway
Subscriptions let you ask the coordinator to watch a Slack channel, run on a schedule or follow all your pull requests. The coordinator then responds to signals on its own, without a new prompt. An example from the announcement: connect Slack and specify a bug-report channel, and the coordinator starts delegating a task whenever a new bug report arrives.
Shared memory holds the project's material, while Subscriptions specify when the coordinator should act. Once configured this way, it does not need to wait for you to type a new task into the chat:
Tell the coordinator agent to watch a Slack channel, run on a schedule, or follow all your PRs. This way it can take action based on signals it detects, without waiting for your prompt.
In other words: ask the coordinator to watch a Slack channel, run on a schedule or follow all your pull requests, and it can act on the signals it detects without waiting for your prompt.
An example from the same announcement:
Connect Slack and point it at a bug-report channel, and it starts delegating each time a bug comes in.
In other words: connect Slack and specify a channel for bug reports, and the coordinator starts delegating a task whenever a new report arrives.
The official announcement does not explain which other events can trigger work beyond Slack, schedules and PRs, and it does not give a price for the feature. To configure it, open Projects from Cursor's left navigation panel. Detailed setup steps are not provided here: follow the interface in the version available to you.
When to consider Cursor Projects
Key takeaway
Cursor offers Projects for large tasks: a new feature, a migration or an entire application. My editorial recommendation: for a one-off change or a small task, a regular chat agent is probably enough. Consider Projects for large tasks that extend over time.
The announcement describes the feature's purpose as “larger bodies of work, such as a feature, a migration, or a full app”—large tasks such as a new feature, a migration or an entire application. That use case is the basis for my recommendation. If a task fits into a single conversation with an agent, a regular chat agent is probably enough. If you regularly return to one large task and the same code, Projects' shared context may help: future agents can use the instructions and findings already accumulated.
If you are choosing between Cursor and other tools, there is a separate comparison—Cursor vs Claude Code. It looks at how they differ in everyday work, without Projects.
What next
Cursor Projects was in beta at the time of its September 10, 2026 announcement. The official announcement does not give a price, so check pricing and limits within the product when you enable it rather than relying on this article.
Continue exploring Ailoot at ailoot.ai.
More details
Questions about Cursor Projects: what it is and how the coordinator delegates tasks
When is a regular agent chat enough instead of Projects?
For a one-off change that fits into one conversation, a regular agent chat is a reasonable starting point. Projects is aimed at larger work that continues over time and benefits from shared findings and instructions.
What should I check before enabling a Subscription?
Decide which signal the coordinator should watch, such as a Slack bug-report channel, a schedule, or pull requests. Check the interface, pricing, and limits available to your account; the announcement does not supply a price or a complete setup guide.