Cursor has announced the general availability of self-hosted cloud agents. This solution allows enterprises to run cloud agents in their own networks and infrastructure, and the codebase, tool execution and construction products do not have to leave the local environment, mainly for teams with high financial and compliance requirements or rely on intranet resources.
Functionally, the self-hosted version retains isolation of remote environments, multi-model access, plug-in extensions, and team permission management, and supports the execution of tasks through workers. There is no need to open inbound ports or establish a VPN during deployment, and it can be launched on a single machine or scaled up and down at scale with the help of Helm charts, Kubernetes operators, and fleet management APIs.
This capability is currently available in the Cursor Dashboard. At the same time, the official reminds that enterprises still need to evaluate risks such as model calls, code indexing, extended signatures, and whitelist configurations in combination with their own security policies.
FAQs
Q: What is Cursor self-hosted cloud agent?
A: This is a solution that runs Cursor cloud agents on the company's own infrastructure.
Q: Which teams are better suited for Cursor self-hosted plans?
A: Financial, highly regulated industries, and teams that rely on private caches, libraries, and intranet endpoints are more suitable.
Q: How do Cursor self-hosted workers connect to the platform?
A: Workers can actively connect to the Cursor cloud via HTTPS without opening the inbound port.
Q: Does Cursor self-hosted plan support large-scale deployment?
A: Yes, Kubernetes-related capabilities and fleet management APIs are officially provided.
